China's Grassland Policies and the Inner Mongolian Grassland System

China’s grassland policies and the Inner Mongolian grassland system

Min Liu

China’s grassland policies and the Inner Mongolian

grassland system

Thesis committee

Promotor Prof. Dr W.J.M. Heijman

Professor of Regional Economics Wageningen University

Co-promotors Min Liu Prof. Dr J. Huang

Professor of Agricultural Economics China Centre for Agricultural Policy (CCAP), Peking University Dr L.K.E. Dries Associate professor, Agricultural Economics and Rural Policy Group Wageningen University

Dr X. Zhu Associate professor, Environmental Economics and Natural Resources Group Thesis Wageningen University submitted in fulfilment of the requirements for the degree of doctor at Wageningen University Other members by the authority of the Rector Magnificus, Prof. Dr J.D. van der Ploeg, Wageningen University Prof. Dr A.P.J. Mol, Dr N.B.M. Heerink, Wageningen University in the presence of the Prof. Dr Robert Finger, Swiss Federal Institute of Technology in Zurich, Switzerland Thesis Committee appointed by the Academic Board Prof. Dr P.P.S. Ho, Delft University of Technology to be defended in public on Tuesday 24 January 2017 at 4 p.m. in the Aula. This research was conducted under the auspices of the Wageningen School of Social Sciences (WASS)

ii iii

China’s grassland policies and the Inner Mongolian

grassland system

Thesis committee

Promotor Prof. Dr W.J.M. Heijman

Professor of Regional Economics Wageningen University

Co-promotors Min Liu Prof. Dr J. Huang

ii iii

TABLE OF CONTENTS

1. INTRODUCTION ...... 2

1.1 Problem statement ...... 2 1.2 Background information ...... 4

1.2.1 Grasslands of China ...... 4

1.2.2 Grassland degradation in China ...... 6 1.2.3 Livestock production in the pastoral areas of China ...... 7

1.2.4 The overview of China’s grassland policies ...... 9

1.2.5 Research region: the pastoral areas of Inner Mongolia ...... 10 1.3 Objectives and research questions ...... 12

1.4 Materials and methods ...... 13

1.4.1 Conceptual framework ...... 13 1.4.2 Sampling and data collection ...... 15

1.4.3 Model selection ...... 16

1.5 Contribution of the dissertation ...... 17 1.6 Organization of the dissertation ...... 18 2. RURAL LAND ALLOCATION IN ECO-FRAGILE REGIONS OF NORTHERN CHINA ...... 21

2.1 Introduction ...... 22 2.2 Conceptual framework ...... 23 2.3 Model specification ...... 25

2.3.1 The utility function of each stakeholder group ...... 25

2.3.2 The social welfare function ...... 26 Min Liu 2.4 Research region and data collection ...... 27 China’s grassland policies and the Inner Mongolian grassland system 2.5 Estimation of the model parameters ...... 29 163 pages. 2.5.1 The parameters ...... 29

2.5.2 The welfare weightܽ݅ǡ ܾ݅ǡ ܿ݅ǡ ...... ݀݅ 30 PhD thesis, Wageningen University, Wageningen, NL (2017) 2.6 Results ...... 31 With references, with summary in English ሺߚ݅ሻ ISBN: 978-94-6343-024-1 2.6.1 Land use conflicts among stakeholder groups ...... 31 DOI: 10.18174/399371 2.6.2 The optimal social allocation ...... 32 2.6.3 Comparison of individual optimum, social optimum and actual land allocation...... 32 2.7 Conclusion ...... 33 3. LAND TENURE REFORM AND GRASSLAND DEGRADATION IN INNER MONGOLIA, CHINA ...... 37 3.1 Introduction ...... 38

iv v

TABLE OF CONTENTS

1. INTRODUCTION ...... 2

1.1 Problem statement ...... 2 1.2 Background information ...... 4

1.2.1 Grasslands of China ...... 4

1.2.2 Grassland degradation in China ...... 6 1.2.3 Livestock production in the pastoral areas of China ...... 7

1.2.4 The overview of China’s grassland policies ...... 9

1.2.5 Research region: the pastoral areas of Inner Mongolia ...... 10 1.3 Objectives and research questions ...... 12

1.4 Materials and methods ...... 13

1.4.1 Conceptual framework ...... 13 1.4.2 Sampling and data collection ...... 15

1.4.3 Model selection ...... 16

1.5 Contribution of the dissertation ...... 17 1.6 Organization of the dissertation ...... 18 2. RURAL LAND ALLOCATION IN ECO-FRAGILE REGIONS OF NORTHERN CHINA ...... 21

2.1 Introduction ...... 22 2.2 Conceptual framework ...... 23 2.3 Model specification ...... 25

2.3.1 The utility function of each stakeholder group ...... 25

3.2 Materials ...... 40 5.3.2 Comparison between the estimated results and actual grassland condition ...... 100 3.2.1 Study area ...... 40 5.4 Discussion ...... 101 3.2.2 Data sources ...... 41 5.5 Conclusions ...... 102 3.2.3 Land tenure reform for the grasslands ...... 42 6. DISCUSSION AND CONCLUSIONS ...... 106 3.3 Empirical model ...... 45 6.1 Introduction ...... 106 3.3.1 Fixed effects model ...... 45 6.2 Synthesis of research findings ...... 107 3.3.2 Variable definitions and expected results...... 45 6.2.1 Different preference on grassland use by different stakeholders ...... 107 3.3.3 Descriptive statistics on the variables ...... 49 6.2.2 Grassland privatisation is a significant reason to cause grassland degradation in a long term 3.4 Results ...... 51 ...... 108 3.5 Discussion ...... 55 6.2.3 Grassland privatisation puts a ceiling on the livestock production of pastoral areas, but it promotes the development of livestock productivity ...... 108 3.6 Conclusions ...... 56 6.2.4 Ecological Construction Programs have significant impacts on preventing grassland 4. HOW DOES LAND TENURE REFORM IMPACT UPON PASTORAL LIVESTOCK degradation ...... 108 PRODUCTION? AN EMPIRICAL STUDY FOR INNER MONGOLIA, CHINA ...... 59 6.3 Contribution to scientific debates ...... 109 4.1 Introduction ...... 60 6.3.1 The tragedy of the commons versus the tragedy of privatisation ...... 109 4.2 Research region ...... 62 6.3.2 Deterministic stocking rates versus stochastic stocking rates ...... 111 4.2.1 Land tenure reform in the pastoral areas of Inner Mongolia...... 63 6.3.3 Top-down intervention versus bottom-up intervention ...... 112 4.2.2 Livestock production in Inner Mongolia ...... 65 6.3.4 The one-size-fits-all versus tailor-made policies for resource governance ...... 114 4.3 Data description ...... 68 6.4 Policy implications ...... 115 4.3.1 Data collection...... 68 6.5 Limitations and recommendations for future research ...... 118 4.3.2 Descriptive analysis...... 69 References ...... 120 4.4 Empirical model ...... 73 Appendixes ...... 134 4.5 Model results ...... 76 Appendix 2-A: Derivation of the individually optimal allocation of land ...... 134 4.6 Discussion ...... 80 Appendix 2-B: Derivation of the socially optimal allocation of land ...... 136 4.7 Conclusions ...... 82 Appendix 2-C: Questionnaires and results of Analytic Hierarchy Process ...... 139 5. THE IMPACT OF ECOLOGICAL CONSTRUCTION PROGRAMS ON GRASSLAND CONSVERVATION IN THE PASTORAL AREAS OF CHINA ...... 86 Appendix 2-D: Estimation of parameter ...... 144 5.1 Introduction ...... 87 Appendix 4-A: Descriptive statistics of theߚ݅ model variables ...... 145 5.2 Material and methods ...... 89 Appendix 4-B: The model specification with quadratic term ...... 146 5.2.1 Research region ...... 89 Summary ...... 149 5.2.2 Data collection...... 91 Acknowledgements ...... 153 5.2.3 Statistical analysis on grassland condition ...... 92 Biography ...... 155 5.2.4 Fixed effects model ...... 93 5.2.5 Descriptive statistics for the variables ...... 96 5.3 Results ...... 98 5.3.1 Model results ...... 98

vi vii

1. INTRODUCTION

1.1 Problem statement

Chapter 1 According to the FAO (2008), grasslands occupy 26% of the total land area of the world, and 70% of the global agricultural area. They are among the largest ecosystems in the world, contribute to the livelihoods of more than 800 million people, and are important as feed sources for livestock, as habitats for wildlife, for environmental protection efforts and for the in situ conservation of plant genetic resources (FAO, 2008). The sustainable use and management of grasslands have been of great concern to academic circles, policy-makers and NGOs, especially in the context of the growing worldwide Introduction problem of grassland degradation, rapid increases in demand for livestock products and the high level of poverty among pastoralists (Suttie et al., 2005; FAO, 2015). As defined by Fernandez-Gimenez and Swift (2003), sustainable grassland use and management should maintain the productive and adaptive capacity of grassland ecosystems while providing for the wellbeing of human communities. In practice, various policy interventions have been made, devoted to balancing the conservation of grasslands, production of livestock and the livelihoods of local households in agricultural societies.

However, the validity of existing policy interventions must be examined in terms of sustainable grassland use and management, given that grassland degradation is still ongoing and rural poverty is deepening in some pastoral areas (Hua and Squires, 2015). In this regard, research in multiple disciplines has found that some government policies accelerate resource destruction (Ostrom, 2009). To date, the policy interventions for grassland use and management widely employed by governments around the world aim to transform traditional pastoralism, by developing sedentarisation instead of nomadism, privatisation1 of grasslands instead of communal tenure, intensification instead of extensive grazing, commercialisation of livestock systems instead of subsistence, deterministic stocking rates instead of opportunistic strategies with stochastic stocking rates, and so forth (Homewood, 1995; Li and Zhang, 2009). Traditional pastoralism is subsistence-oriented and has been perceived as primarily mobile, manifesting in such diverse forms as nomadism, transhumance, and migration as well as pasture rotation on a large scale (Fernandez-Gimenez and Le Febre, 2006). Widespread sentiments identify traditional pastoralism as an irrational, ecologically destructive and economically inefficient system (Homewood, 1995; Ho and Azadi, 2010). So, what are the actual outcomes of the government-driven

1 Note that privatisation generally refers to the assignment of property rights over an asset or resource from the public (state) to the private sector. In the case of China, formal property rights over grasslands have remained in the public (or collective) sector. Hence, when we use the term privatisation for grasslands in the Chinese context we include the transfer of use rights over grasslands from the public to the private sector rather than the assignment of formal property rights. 1 2

Introduction Chapter 1 1

(Homewood, Azadi,2010).So,what 1995;Hoand ar traditional pastoralismasanirra and (Fernandez-Gimenez rotation onalargescale nomadism, as forms diverse mobile, manifesting in such Zhang, pastoralismissubsistenc Traditional 2009). st withstochastic strategies instead ofopportunistic grazing,inst oflivestock systems commercialisation 1 grasslands occupy (2008), FAO the According to 1.1 Problemstatement 1. INTRODUCTION nomadism, privatisation around the world aim to transform traditional past governments employed by managementwidely fordate, thepolicy and grassland interventions use government policies some disciplines hasfoundthat and Squire insomepastoralareas(Huadeepening given grasslandgrassland and management, use that policy existing thevalidity of However, intervent production oflivestockandthelivelihoodslo de various policy made, interventionshave been capacity of grassland ecosystems while providing for th and managem use grassland Swift (2003),sustainable (Suttieetal.,2005; of poverty pastoralists among grasslandin problem of rapid increases degradation, espe NGOs, and policy-makers circles, to academic manag and The sustainableuse resources 2008). (FAO, a efforts for for protection environmental wildlife, livelihoods ofmorethan800millionpe the global agricultural area. among They lar are the over grasslandsfrom public the tothe private sector termwe privat usethe when sector. Hence, to the private sector. In the case of China, formal property Note thatNote generally privatisation refersprope of tothe assignment 1 of grasslands instead of communal tenur communal of grasslands insteadof tional, ecologically destructive a tional, ecologically destructive isation for grasslands in the Chinese contex Chinese the in for grasslands isation ople, and are important as feed s rather than the assignment of formal property rights. formal of property the rights. assignment rather than cal households in agricultural societies. householdsinagriculturalcal societies. rights over grasslands have remained in the public (or collective) FAO, 2015). As defined by Fernandez-Gimenez and byFAO, 2015).Asdefined Fernandez-Gimenez 26% of the total land area of the world, and 70% of 70% and world, of the area total land 26% of the voted to balancing the conservation of grasslands, theconservation ofgrasslands, voted tobalancing 2 ocking rates, and so forth (Homewood, 1995; Li andLi 1995; (Homewood, forth andso ockingrates,

oralism, by developing accelerate resource destruction (Ostrom, 2009). To (Ostrom, 2009). destruction resource accelerate Le sentimentsidentify 2006).Widespread Febre, in terms of sustainable in terms ofsustainable beexamined must ions nd for the in situ conservation of plant genetic plant of conservation insitu for the nd e-oriented and has been perceived as primarily as primarily perceived been andhas e-oriented demand forproducts and livestock the level high cially in the context of the growing worldwide growing worldwide of the context ciallyin the gest ecosystems in the world, contribute to the ead of subsistence, stocking deterministic rates ead degradation isstillongoing andrural poverty is ent should maintain the productive and adaptive adaptive and the productive ent shouldmaintain e the actual outcomes of the government-driven actuale of the outcomes government-driven the s, 2015). Ins, 2015). thisregard, research inmultiple ement of grasslands have been of great concern great concern of have been grasslands ement of e wellbeing of human communities. Ine wellbeing communities. ofhuman practice, rty rights over orre anasset transhumance, and migration as well as pasture aspasture as well migration and transhumance, e, intensification insteadofextensivee, intensification nd economically inefficient system inefficientsystem nd economically ources for livestock, as habitats for livestock, ources t we include the transfer of use rights sedentarisation instead of source from public the (state )

1 transformation of traditional pastoralism? In other words, what are the impacts of the prevailing policy flexibility and reciprocity of grassland use (Fernandez-Gimenez and Le Febre, 2006). It has been interventions on the grassland system2? advocated that the government-driven transformation of traditional pastoralism broke the balance between the grassland ecosystem and livestock production, increased vulnerability to natural disasters The global tendency of grassland policy in the twentieth century has been toward privatisation (or and threatened pastoral livelihoods. In particular, privatisation and enclosure in arid and semi-arid individualisation, subdivision, etc.), which transforms the property rights or use rights over grassland pastoral areas seriously threatens the integrity of ecological systems because these areas are resources from communities to individuals (Mwangi, 2007; Li and Zhang, 2009; Li and Huntsinger, characterised by temporal variability and spatial heterogeneity (Mwangi, 2007). It is necessary to 2011). Since then, grassland privatisation has become the most significant transformation of traditional examine whether a tragedy of privatisation may have occurred, rather than the well-known tragedy of pastoralism, and the result of establishing physical boundaries between grassland parcels has led to a the commons. In addition, some scholars have implied that grassland degradation as a result of series of considerable changes in traditional modes of livestock production and household livelihoods overgrazing is not possible in traditional pastoralism, because animals would be moved to new places (Sheehy et al., 2006; Li and Zhang, 2009). In general, land-ownership regimes other than private for available forage before the vegetation of one area is significantly impacted by grazing (Li and property are considered by many economists to be inefficient and prone toward the overuse of resources Huntsinger, 2011). Furthermore, the restriction of livestock grazing inevitably causes economic loss to (Mwangi, 2007). Particularly, the conventional wisdom based on the concept of “the tragedy of the herders, which may impede the implementation of the environmental policies and programs. As such, commons” (Hardin, 1968) and the theorem of property rights (Coase, 2013), contends that the open the environmental policies and programs do not seem to be effective for preventing grassland access to resources and unclear property rights of traditional pastoralism would inevitably cause overuse degradation, especially those focussed on controlling livestock grazing. and increasing grassland degradation (Li et al., 2014). As such, grassland privatisation has been long embraced as the solution for achieving the sustainable use and management of grasslands (Mwangi, What are the outcomes of these prevailing policy interventions, i.e. the outcomes of the government- 2007; Ybarra, 2009). The development of grassland privatisation has also been influenced by the driven transformation of traditional pastoralism? Are they valid in light of the sustainable use and changes in the economy, such as technological innovation, changes in relative factor scarcities management of grasslands? In this dissertation, these questions will be examined by researching the especially with increasing populations and the creation of new markets (Demsetz, 1967; Mwangi, 2007). following issues: what is the impact of grassland privatisation on grassland condition and livestock However, an increasing number of scholars have explored some of the unintended effects of grassland production? What is the impact of the environmental policies and programs on grassland conservation? privatisation, challenging the conventional wisdom of the “the tragedy of the commons”. Although some academic studies and governmental reports have presented the results of policy interventions on grassland systems, the existing literature lacks quantitative analysis, based on large- After the period of implementation of grassland privatisation, various environmental policies and scale areas and long-term observations, in order to dissect the policy effectiveness, and has thus far programs have been introduced, influenced by growing attention to grassland ecosystems and fallen short of thorough and systematic discussion on the outcomes of government-driven worldwide grassland degradation in recent decades. The policy interventions for grassland conservation transformation of traditional pastoralism (Lu et al., 2015). Therefore, this research employs empirical mainly target at controlling livestock grazing, because overgrazing is believed to be one of the main analysis based on panel data involving extensive grasslands of China and spanning several decades, causes of grassland degradation by most of policy makers, such as in South Africa, South American with a focus on the effects of the transformation of pastoralism on both grassland condition and livestock Campos, China etc. (Suttie et al., 2005; Miao et al., 2015; Liu et. al., 2016). These environmental production. policies and programs further prompted pastoralists to transform their traditional lifestyles, particularly developing the intensification of animal husbandry, in place of extensive, grazing practices (Cao et al., 1.2 Background information 2013b). 1.2.1 Grasslands of China Nevertheless, traditional pastoralism had prevailed for centuries. Increasing literature has criticised China has around 392 million hectares of grasslands, the second largest area in the world after Australia, existing policy interventions as impairing the strategies of traditional pastoralism, reducing mobility, accounting for 12% of the world’s grasslands and 41.7% of the national land area (Fan et al., 2008). Nearly 80% of these grasslands are in arid and semi-arid regions, which are characterised as most 2 The grassland system in this research includes both the grassland condition (grassland area and quality) and livestock production. This is an adaptation of the resource system as defined by Ostrom (2009) in her framework of social- vulnerable to degradation, desertification and salinisation (Feng et al., 2008; National Bureau of ecological systems. 3 4

2 systeminterventions onthegrassland policy In oftheprevailing are theimpacts words, other what transformation pastoralism? oftraditional privatisation, challenging the expl have scholars number of increasing an However, crea populationsandthe withincreasing especially changes intheeconomy the by grassland2007; beeninfluenced Ybarra, development privatisation hasalso 2009). of The embraced as the solution for sustai achieving the degradational., 2014) grassland and increasing (Li et oftr rights toresources propertyaccess andunclear theoremofprope andthe commons” (Hardin,1968) 2007).Particularly,(Mwangi, the property are considered by many economists to be in (Sheehy et al., In andZhang, 2006; Li 2009). gene intraditionalmodes changes oflivestockproductionandhouseholdlivelihoodsseries ofconsiderable ofestablishingpastoralism, andtheresult physical grassland2011). Since privati then, resources from communitiesto individuals (Mwa individualisation, subdivision, etc.), whichtransfor The global tendencyThe global grassland policy of inthetw ecological systems. ecological systems. production. This is an adaptation of the system framework resource asdefined bysocial- of Ostrom (2009)in her existing policyexisting interventions asimpairingthestra pastoralismhadprevailed traditional Nevertheless, 2013b). in animal husbandry, of developing theintensification to promptedpastoralists policies andprograms further Campos, Chinaetc.(Suttieetal.,2005;Miao causes ofgrassland by degradation mostofpolicy ma grazing, beca livestock at controlling mainly target The decades. recent in degradation grassland worldwide by influenced introduced, been programs have e privatisation,various grassland implementationof After periodof the The grassland the includes both The research system this in , such as technological innovation,such changes as technological relative factor in scarcities conventional wisdom ofconventional the commons”. “thetragedy wisdom of the conventional wisdom basedonthe conventional sation has become the most significant transformation the most significant sation has become of traditional 2 ? grassland (grassland condition area and quality) livestock and et al., 2015; Liu et al.,2015; et.al.,2016).These environmental tion of new markets (Demsetz, 1967;Mwangi, (Demsetz, tion ofnewmarkets 2007). nable use and management of grasslands (Mwangi, ofgrasslands (Mwangi, nable useandmanagement aditional pastoralismwouldinevitably overuse cause ngi, 2007; Lingi,and Zhang, 2007; 2009;Li andHuntsinger, tegies of traditional pastoralism, reducing mobility, mobility, reducing pastoralism, traditional tegies of 3

ms theproperty rights or use rights over grassland ed to be one of the main ed tobeoneofthe isbeliev use overgrazing growing attention to grassland ecosystems and grassland and ecosystems growing attention to boundaries between grassland parcels hasledtoa parcels grassland boundaries between efficient and proneefficient th and toward entieth toward privatisation century (or has been for centuries. Increasing literature has criticised Increasing literature has for centuries. rty rights (Coase, 2013), contends that the that contends 2013), (Coase, rightsrty ored some of the unintended effects of grassland of effects unintended the ored someof place of extensive, grazing practices (Cao et al., et (Cao grazing practices extensive, placeof . As such, grassland privatisation has been long grassland privatisationhasbeenlong . Assuch, ral, land-ownership regimes other than private regimes thanprivate other land-ownership ral, transform their traditional lifestyles,transform theirtraditional particularly kers, such as in South Africa, South American kers, suchasinSouthAfrica,American policy interventions for grassland conservation grassland conservation interventions for policy concept the of“thetragedy of nvironmental policies and nvironmental policies and e overuse of resources resources of e overuse

open open and threatened pastoral li pastoral and threatened producti ecosystem andlivestock grassland between the transformati government-driven advocated thatthe flexibilityand reciprocitygrassland use(Fer of vulnerable to degradation, desertification and sa and desertification todegradation, vulnerable mi-arid regions, are which as most characterised se grasslands are in aridNearly and 80% of these 41.7% world’sgrasslands and foraccounting 12%ofthe th grasslands, million hectaresof around 392 China has 1.2.1 Grasslands China of 1.2 Background information production. of ofthetransformation effects with afocus onthe decades, severalspanning and China grasslands of data involvingextensive onpanel based analysis transformation (Lu oftraditionalpastoralism et al., fallen shortofthorough andsystematic disc observations, todi inorder andlong-term scale areas interventions systems, ongrassland li theexisting results policy of the presented governmental have reports and studies academic Although some What is theimpactproduction? oftheenvironmental following issues:what istheimpactof grasslan management In ofgrasslands? thisdissertation,these traditionalpastoralism?driven transformation of Are What are theoutcomesoftheseprevailing policyin focussedcontrollingdegradation, those on grazing. especially livestock the environmentalpoliciesandprograms seem donot of theimplementation may impede which herders, live of restriction Furthermore, the Huntsinger, 2011). one before of thevegetation forage for available in traditional pastoralis overgrazing is not possible impl Inaddition, somescholarshave the commons. of tragedy thewell-known than rather occurred, may have privatisation a tragedy whether of examine variabilityand spatial temporal characterised by threatens theintegrity seriously pastoral areas velihoods. In velihoods. particular,privatisation d privatisationongrassla nandez-Gimenez and Le Febre,It and 2006). been has nandez-Gimenez pastoralism on both grassland condition and livestock andlivestock onbothgrassland condition pastoralism of ecological systems because these areas are are areas these because systems ofecological terature lacks quantitative analysis, based on large- terature quantitative lacks the environmental policies and programs. Assuch, programs. and policies theenvironmental linisation (Feng etal.,2008; National Bureau of ussion on the outcomes of government-driven ussion ontheoutcomes ofgovernment-driven 4 m, because animals would be moved to new places places new be moved to would animals m, because

heterogeneity (Mwangi, 2007) (Mwangi, heterogeneity area is significantly impacted by grazingand is by significantly (Li impacted area 2015). Therefore, this research employs empirical terventions, i.e.theoutcomesofgovernment- on oftraditionalpastoralism broke thebalance policies andprograms ongrassland conservation? stock grazing inevitably causes economic loss toeconomic stock loss grazing causes inevitably e second largest area in the world after Australia, Australia, after in theworld area largest e second ssect thepolicyfar effectiveness, andhasthus questions willbeexamined bythe researching ied that grassland degradation as a result of thatgrasslanda result degradationied as they in light valid of the sustainable use and on, increased vulnerability on, increased disasters tonatural of the national land area (Fan et al., 2008). to be effective forgrassland preventing tobeeffective and enclosure in arid and semi-arid and enclosureinaridsemi-arid nd condition and livestock nd conditionandlivestock . It. necessaryis to

1 Statistics of China, 2009). Approximately 17 million people maintain their livelihoods on the grasslands called the traditional pastoral areas (Ho and Azadi, 2010), and which were home to the world’s largest of China (Li et al., 2014). The extensive grasslands of China are concentrated in six provinces and population of sheep and goats and the fourth largest concentration of cattle in the 1990s (Zhang and autonomous regions: Xinjiang, Tibet, Qinghai, Sichuan, Gansu and Inner Mongolia. They account for Yang, 1990; National Research Council, 1992; Ho and Azadi, 2010). They are considered to be the 75% of grasslands and accommodate 70% of the grazing livestock of China (Suttie et al., 2005). These major areas for livestock production in China, with a long history of producing meat, milk, wool and areas maintained traditional pastoralism over hundreds of years, before undergoing thorough land tenure fur. Moreover, local animal husbandry provides livelihoods to millions of people, with a majority reform since the 1980s and a series of Ecological Construction Programs since 2000. These policy existing as traditional self-sufficient enterprises and a few commercial grazing businesses (Li et al., interventions have been implemented widely in favour of transforming traditional pastoral practices. 2014). The ecological function of China’s grasslands is being emphasized increasingly, such as biodiversity conservation, carbon sequestration, soil retention, soil fertility maintenance, and catchment protection (Suttie et al., 2005). It was shown that the grasslands of northern China significantly impact both regional climates and the global carbon cycle (Ni, 2002). In addition, the majority of pastoral areas are populated by ethnic minorities, such as Mongolians and Ewenkis, thus protecting the grasslands of China also has a social function of maintaining cultural diversity and social stability (Kang et al., 2007). These socio-economic and environmental functions of grasslands have invoked the increasing attention of central and local governments to adopt policy interventions for sustaining the productivity of—in addition to protecting—the grassland ecosystem.

1.2.2 Grassland degradation in China

Grassland degradation, a worldwide problem, is particularly severe in China (Ho and Azadi, 2010). Grassland degradation can manifest itself in the lowering of grassland productivity, reduction in soil fertility, soil compaction, increased presence of unpalatable species of grass, or a combination thereof (Li et al., 2013). Although lacking clear documentation and differing in specifics regarding the extent

of China’s grassland degradation during the last several decades, it is unanimous among herders, scholars and governments that the grasslands of China have experienced degradation, with significant Figure 1.1 The distribution of grasslands in China regional variation (Harris, 2010; Waldron et al., 2010). An often-cited figure is that about 90% of

As shown in Figure 1.1, the grasslands of China are mainly distributed on the Inner Mongolia Plateau, China’s grasslands were degraded to various extents by the 2000s (Unkovich and Nan, 2008; Waldron the Loess Plateau and the Qinghai-Tibetan Plateau. The extensive grasslands are classified into four et al., 2010), and that degradation was increasing at a rate of two million hectares per year (Ren et al., major types based on the analysis of plant community in phytology (Wu et al., 1980; Kang et al., 2007), 2007; Harris, 2010; Cao et al., 2013b). The grassland areas shrunk significantly by the beginning of the including meadow steppes, typical steppes, desert steppes and alpine steppes. These four types of twenty-first century. In particular, the northern grassland boundary had moved about 200 km southward grasslands have a combined distribution ranging from the northeast plain adjacent to Mongolia to south and the western boundary about 100 km eastward compared to the previous hundred years (Liu 2008; of the Tibetan Plateau (Kang et al., 2007). The meadow steppes and typical steppes are the most Cao et al., 2013b). commonly used grasslands for grazing and other economic activities related to livestock production, A series of natural disasters in China in the late 1990s are to some extent responsible for degradation and most of them are located on the Inner Mongolian Plateau (Kang et al., 2007). and desertification of the grasslands in northern China (Kang et al., 2007; Wu et al., 2015), for instance,

China’s grasslands play crucial roles in livestock production, household livelihoods and global the Yangtze River floods that killed thousands of residents, the Yellow River droughts that cost billions ecosystems (Kang et al., 2007). Northern China contains over 50% of China’s grasslands, which are of US dollars in economic losses, and the frequent dust-storms and sand-storms that blanketed the sky of Beijing and affected the health and economic wellbeing of millions of city-dwellers (Harris, 2010). 5 6

ecosystems (Kangecosystems etal.,2007).Northern China c rolesin live crucial China’s grasslands play 2007). Inneret al., Mongolian Plateau (Kang onthe and mostofthemarelocated grazingeconom forcommonly grasslands and other used of theTibetanPlateau(Kang etal., 2007). Themea distributionranginga combined grasslands fro have including meadow typical steppes, steppes,desert community of plant analysis the major on types based Plateau.Th Qinghai-Tibetan Loess andthe the Plateau Inner MongoliaPlateau, distributed onthe grasslands mainly Chinaare of As showninFigure 1.1,the interventions implemen have been policy These Construction since2000. Programs Ecological of series and a reform sincethe1980s years, landtenure before undergoing thorough of hundreds pastoralismover areas maintainedtraditional 75% of grasslandsgrazing 70% and of the accommodate autonomous regions: Xinjiang, Tibet,Qinghai, Sic and provinces insix concentrated grasslands extensive ofChinaare The et al., 2014). of China (Li milli ApproximatelyStatistics ofChina,2009). 17 Figure 1.1 The distribution of grasslands in China Figure distributionofgrasslands inChina 1.1The ted widely infavouroftransformin ted widely stock production, household livelihoods and global livelihoodsand global production,household stock on people maintain their livelihoods on the grasslands livelihoods on the maintain their on people ontains over50%ofChina’s 5 huan, Gansu and Innerhuan,Mongolia. Gansu and for account They

m the northeast plain adjacent toMongoliam thenortheastplainadjacent tosouth steppes and alpine steppes.Thesefour of types steppesandalpine in phytology (Wu et al., 1980; Kang et al., 2007), et 1980; al., Kang al., in phytology (Wu et dow steppesandtypical steppesare themost e extensive grasslands are classified intofour are grasslands classified e extensive livestock of China (Suttie etal.,2005).These ofChina livestock ic activities related tolivestock production, related ic activities g traditional pastoral practices. pastoralg practices. traditional grasslands, are which

existing as traditional self-sufficient enterprises a enterprises astraditionalself-sufficient existing liveli husbandry provides animal fur. local Moreover, productioninChina,with major areasfor livestock and Ho Council, 1992; NationalYang, Research 1990; population ofsheepandgoats andthefourth largest 2010),an and Azadi, (Ho pastoralareas called thetraditional of Beijing and affected the health and economic we healthandeconomic of Beijing the andaffected of USdollarsineconomiclosses, andthefrequent River floodsthatkilledthousandsofresithe Yangtze China’s grasslandsdegraded were to etal., 2010;Waldron regional (Harris, variation scholars andgovernments that grasslands the of Chin of China’s grassland during degradation se the last a documentation clear (Li lacking etal.,2013).Although fertility, un presence of increased soil compaction, manifestitselfinthe can Grassland degradation lo par problem, worldwide is Grassland a degradation, 1.2.2 Grassland degradation inChina addition toprotecting productivity of interventions toadoptpolicy forgovernments the sustaining local and of central environmentalThese socio-economic functions and of gr China alsohasasocialfunction ofmaintaining cultural minorities, suchasMongolians by ethnic are populated areas In the majorityaddition, pastoral (Ni, of 2002). cycle the regionalcarbon globalboth and climates 2005).Itprotection (Suttieetal., was shownthatth biodiversity soilre carbonsequestration, conservation, 2014). of China’s function The ecological grasslands and desertification of the grassla ofthe and desertification inChinathelate1990s areA seriesofnaturaldisasters tosomeextent responsible for degradation Cao etal.,2013b). compar 100 km boundary eastward andabout the western thenorthern In particular, century. twenty-first 2007; Harris, 2010;Cao ar et al.,2013b).Thegrassland ra at a that degradation increasing was and et al., 2010), the grassland ecosystem. grassland ecosystem. —the nds in northern China (Kang etal., (Kang nds innorthernChina various extents by extents the2000s(Unkovich various grassland boundary had moved about 200 kmsouthward about200 moved grassland boundary had nd a few commercial grazing businesses (Li et al.,et (Li businesses grazing few commercial nd a 6 2010). An often-cited figure is that about 90% of figure isthatabout90% 2010). Anoften-cited palatable species of grass, or a combinationgrass, thereofpalatable species of dust-storms and sand-storms that blanketed the sky dust-storms

e grasslands of northern China significantly impact Chinasignificantlye grasslands ofnorthern impact weringgrasslandvity, of producti reduction insoil llbeing of millions of city-dwellers (Harris, 2010). (Harris,2010). llbeingof city-dwellers ofmillions a long history of producing meat, milk,wooland a long meat, history ofproducing dents, theYellow Riverdroughts thatcostbillions ticularly severe in China (Ho and Azadi, 2010). Azadi,2010). and China(Ho in severe ticularly a have experienced degradation, with significant significant with degradation, experienced a have veral decades, it is among it unanimous herders, veral decades, concentration of cattle in the 1990s (Zhang and in theconcentration 1990s (Zhang of cattle tention, soil fertility maintenance, and catchment diversitystability andsocial (Kang etal.,2007). eas by shrunksignificantly thebeginning ofthe and Ewenkis, thus protecting the grasslands ofgrasslands protecting the thus and Ewenkis, hoods to millions of people, with a majority with amajority hoods tomillions of people, te of two million hectares per year (Ren et al., et al., (Ren year million hectares per two of te is being emphasized increasingly, such as such increasingly, isbeing emphasized asslands have invoked the increasing attention asslands invokedtheincreasing attention have Azadi, 2010). They are considered to be the tobe considered are They Azadi, 2010). nd differing in specifics regardingnd differing in specifics theextent ed to the previous hundred years (Liu 2008; hundred totheprevious ed d which were home to the world’slargest home to were which d 2007; Wu et al., 2015), for et al., 2015), instance, 2007; Wu and Nan, 2008; Waldron and Nan, 2008;Waldron in —in

1 It has been noticed that grassland degradation does not only impact herders who rely on the grasslands Tianjin Municipalities) and northeast (Liaoning Province, Jilin Province and Heilongjiang Province) for survival and their livelihood, but also threatens the ecological security of the whole country (Harris, (Cao et al., 2013b). Mixed systems are the interface between traditional grazing and industrial 2010; Huang et al., 2013). production.

Plenty of politicians and scholars have raised concerns over the causes of grassland degradation in China. China is a major player in the international livestock sector and has experienced impressive increases Overgrazing and unsuitable grassland reclamation for cultivation have been accepted as the principal in livestock production 3 but it has been observed that the increases in livestock production are reasons (Ho, 2001). It was indicated that the total number of livestock in China increased by 594 million predominantly associated with industrialised, rather than traditional, grazing systems in China (Cao et from 1981 to 2006 (Cao et al., 2013b), and an estimated 67 million hectares of high-quality grasslands al., 2013b). The intensification of livestock production in China started in the late 1980s and has have been converted to the cultivation of grains since 1949 (Ho and Azadi, 2010). However, other advanced progressively ever since, which resulted in significant changes in livestock production scholars have suggested that overgrazing only exists and leads to degradation in specific areas, with structures as well as the geographical distribution and composition of the livestock commodities of little evidence from national data to suggest that overgrazing per se has been responsible for widespread China (FAO, 2016). In addition, with the increasing environmental concerns about the ecosystem of grassland degradation across China’s pastoral grazing areas (Cao et al., 2013b). As such, one of the natural grasslands, livestock production increasingly relies on the utilisation of crop stalks, bran, and most controversial and debated aspects of research about grasslands is the existence and extent of other byproducts of grain, instead of natural pastures (FAO, 2015). In the 1990s, the central government overgrazing and its impact on grassland degradation (Suttie et al., 2005). Other anthropogenic called for moving cattle and sheep production from the traditional grazing region to the grain-producing interventions, such as over-excavation of medicinal herbs and mining, are thought to be responsible for provinces, because it was believed that overgrazing of pastoral areas was the main reason for grassland at least some of the grassland degradation (Ren et al., 2007). Climate-related factors are an ongoing degradation. This policy strategy was strengthened after a series of environmental disasters had occurred concern as a significant driver of grassland degradation in arid and semi-arid areas. For example, the in the 2000s. What followed has been a steady increase in the total livestock outputs in China but a distribution of precipitation is discussed as a principle factor of the changes in grassland condition decrease in the share of livestock production in the pastoral areas (Li, 2009). In short, the market share (Harris, 2010; Cao et al., 2013b), while temperature is concluded to be a driver for changes in the of livestock products in China has experienced a geographic shift from pastoral areas to crop-farming biomass production of grasslands (Piao et al., 2006). Last but not least, the policy interventions that areas, in accordance with changes in production technology and national policy strategies in recent years attempt to transform traditional pastoralism have been suspected to be significant causes of grassland (Li et al., 2008). degradation in China (e.g. Li and Zhang, 2009; Harris, 2010; Cao et al., 2013b). Livestock production in the pastoral areas provides livelihoods to millions of people and has 1.2.3 Livestock production in the pastoral areas of China considerable capacity to alleviate the increasing stress of livestock product demands that have risen with growing population, urbanisation and rising disposable incomes in China (FAO, 2015). Nevertheless, China’s livestock systems can be classified as traditional grazing, industrial production or mixed few studies have discussed the livestock production of the pastoral areas clearly and thoroughly. One systems (Cao et al., 2013b). Traditional grazing systems are characterised by the direct consumption of important reason is the questionable accuracy of Chinese livestock statistics, especially the data about permanent pastures (Waldron et al., 2007). The majority of production in traditional grazing systems the pastoral areas (Waldron et al., 2007). Statistical data collection on the livestock population is provides subsistence to local residents, and therefore their livestock production usually includes several extremely difficult in China because small rural households raise livestock at a small scale, and supply species of multipurpose animals—producing meat, milk, hides, fibre, transport, and manure for fuel chains are dominated by countless small traders and processors, unlike the centralised slaughter and (Suttie et al., 2005). These regions are called pastoral areas, which are geographically vast but sparse in auction systems that facilitate statistical data collection in developed countries (Waldron et al., 2007). population. They are mainly located in the west, north and northwest of China, including the Mongolian China’s method of collecting these data has been a system of bottom-up reporting, which is vulnerable and Tibetan plateaus and Xinjiang Uighur Autonomous Region (Cao et al., 2013b). Industrial production systems aim to produce commodities for market sale, such as milk and meat. Their livestock production usually concentrates on one or two species of animals and relies also on crop products and sown pastures. Industrial production systems are mainly located in the southeast (Guangdong Province 3 Based on the FAOSTAT database, China’s meat output accounted for 10.8% of the world total in 1980, and the share rose to 28.7% in 2004 (Li, 2009). The average growth rate in meat production between 1980 and 2004 was 7% in China, compared and Guangxi Zhuang Autonomous Region), north (Shanxi Province, Hebei Province and Beijing and with only 3% for the world (Waldron et al., 2007). 7 8

and Guangxi Zhuangand Guangxi Region), (Sha Autonomous north Industrialsown pastures. productionsystems located aremainly inthesoutheast (Guangdong Province two species or on one production concentrates usually production systems market aimtoproducecommoditiesfor Xinjiangand Tibetanplateaus Uighur Au includingtheMongolian of China, andnorthwest west,population. They north inthe are located mainly ar regions These (Suttie etal.,2005). animals species ofmultipurpose several includes usually e livestockproduction therefor their residents, and provides to subsistence local majority The al., 2007). (Waldron et permanent pastures grazing Traditional systems systemset al.,2013b). (Cao China’s livestock systems can be classified as China1.2.3 Livestockproductioninthepastoralareas of LiZhang, 2009;Harris, 2010; Cao etal.,2013b). degradationand inChina(e.g. pastoralism haveb attempt totransform traditional biomass productionofgrasslandsLa (Piaoetal.,2006). 2010;Caoetal.,(Harris, 2013b),while temperature is distribution ofprecipitation isdiscussedasaprinc concern as a driver significant of grassland degradati at least someofthegrassland (Renet degradation ofmedicinalhe interventions, suchasover-excavation grassland degradat itsimpacton and overgrazing deba most controversialand grazi grassland China’s pastoral degradation across little evidence from nationaldatatosuggest thatoverg that overgrazing a only scholars havesuggested exists grai the cultivationof to converted have been from 1981 to 2006 (Cao 67 et al., 2013b), and an estimated reasons It (Ho, 2001). wasindicatedthatthetotalnum for grasslandOvergrazing reclamation andunsuitable have Plentyand scholars of politicians 2010; Huanget al.,2013). the butalsothreatens livelihood, and their for survival It noticedthatgrassland hasbeen degradation does ted aspects of research about grasslands is the existence and extent of extent and existence grasslands isthe about research of ted aspects —producing meat, milk, hides, fibre, transport, fuel and manure for e called pastoral areas, which are geographically vast but sparse ingeographically are vastbutsparse which areas, pastoral e called raised concerns over the causes of grassland degradation inChina. of causes overthe concerns raised ns since 1949 (Ho and Azadi, 2010). However, other other Azadi, 2010). However, and (Ho ns since 1949 traditional grazing, industrialproductionormixed tonomous RegionIndustrial (Caoetal.,2013b). 7

een suspected to be significant causes of grassland tobesignificantcausesof een suspected not only whorely impactherders onthegrasslands al., 2007). Climate-related factorsal., 2007).Climate-related are anongoing ng areas (Cao et al., 2013b). As such, one of the ofthe Assuch,one et al.,2013b). ng areas(Cao iple factor of the changes in grassland condition condition grassland changes in iple factorofthe ion (Suttie et al., 2005). Other anthropogenic anthropogenic Other al., 2005). ion (Suttieet ber of livestock in byChina increased 594 million razing per responsiblese has been for widespread on in arid and semi-arid areas. For example, the the For areas. example, on in arid and semi-arid rbs andmining,for are toberesponsible thought ecological security of the whole country (Harris, country the whole of security ecological of animals and relies also on crop products and and products crop alsoon and relies animals of cultivation have been accepted as the principal as theprincipal accepted cultivation havebeen nd leads to degradation in specificareas, with nxi Province, Hebei Province and Beijing and Beijing Province,Hebeiand Province and nxi concludedtobeadriverfor changes inthe are characterised by the direct consumption of st butnotleast, thepolicy interventions that of production in traditional grazing systems grazing systems intraditional production of sale, such as milkandmeat. Theirlivestock sale, such million hectares of high-quality grasslands ofhigh-quality million hectares few studies have discussed thelivestockproducti few studies havediscussed China’s method of collecting these data has been a systemofbottom a hasbeen collecting thesedata China’s methodof coll statisticaldata facilitate auction systemsthat and slaughter unlikethecentralised and processors, countlesssmalltraders dominatedby chains are and supply livestockatasmallscale, raise smallruralhouseholds inChinabecause extremely difficult the pastoralareas(Waldronetal Chin of accuracy important reason isthequestionable growing population,urbanisationandrising disposab stre theincreasing toalleviate capacity considerable provi Livestock in the pastoral areas production (Liet al.,2008). tec changes inproduction with inaccordance areas, ageo productsin of livestock China hasexperienced inth production oflivestock intheshare decrease increa a steady been followed has in the2000s.What degradation.af was Thispolicy strengthened strategy overgrazing believed that it was provinces, because grain-producing grazing tothe region thetraditional from production sheep and cattle moving called for pastur insteadofnatural grain, of other byproducts grasslands, livest natural withtheincreasin China (FAO, Inaddition, 2016). structures as well as the geographicalstructures aswell asthe distributi result ever since,which progressively advanced oflivestockproducti al., 2013b).Theintensification withindustrialised,ratherpredominantly th associated 3 in livestock production increases impressive experienced has and livestocksector theinternational player in China isamajor production. systems(Cao etal.,2013b).Mixed are theinterfa Province and Heilongjiangnce, Jilin northeast(Liaoning Tianjin Municipalities)and Provi Province) withfor world only the 3% etal.,2007). (Waldron to 28.7% in 2004 (Li, 2009). wasgrowthTheproductionmeat average inChina, rate in between compared 7% 1980 and 2004

Based on the FAOSTAT database, China’s meat China’s database, Based onforworld outputFAOSTAT the 10.8%of accounted totalin the 1980,and share the rose 3 ock production increasingly relies on the utilisation of crop stalks, bran, and and crop stalks,bran, theutilisationof increasinglyock production on relies livestockproductionare in thattheincreases but ithasbeenobserved ., 2007). Statistical datacollectiononthelivestockpopulation is., 2007).Statistical on andcompositionofthe li ection in developed countries (Waldron et al., 2007). et al., 2007). (Waldron countries ection in developed e pastoral areas (Li, 2009). In short, the market share share market In short,the 2009). areas (Li, pastoral e on of the pastoral areas clearly and thoroughly. One clearly andthoroughly. One on ofthepastoralareas es (FAO, 2015). In the 1990s, the central governmentIn thecentral the1990s, es (FAO,2015). hnology and national policy strategies in recent years in recent strategies hnology nationalpolicy and ed in significant in livestock changes production 8 ss of livestock product demands that have risen with with risen thathave demands ss oflivestockproduct

of was main areas of the pastoral for grassland reason g environmental concerns about the g environmental of ecosystem concerns ter a series of environmental disasters had occurred disasters hadoccurred environmental a seriesof ter des livelihoods to milli graphic shift from pastoral areas to crop-farming graphic shiftfrom areas pastoral tocrop-farming le incomesinChina(F ese livestock statistics, especially the data about about thedata ese livestockstatistics, especially an traditional,grazing systems inChina(Caoet on in China started in the late 1980s and has on inChinastartedthelate1980sand ce between traditional grazing traditional ce between and industrial se in the total livestock outputs in China but a a outputsinChinabut thetotallivestock in se -up reporting, which is vulnerable reporting, isvulnerable which -up AO, 2015). Nevertheless, AO, 2015).Nevertheless, ons of people and has vestock commoditiesof

1 to misreporting from herders and local officials hoping to respond to the policy orientation of higher- emphasised the specific allocation of the contracted items (e.g. grassland plots, areas, issuing user-right level governments (Waldron et al., 2007). certificates) to households to ensure the long-term stability of contract relations in 2011 (Li, 2012). According to the current Law on the Contracting of Rural Land, grassland use rights can be contracted 1.2.4 The overview of China’s grassland policies to individual households for 30 to 50 years. For a long time, China had no comprehensive long-term policy for grasslands (Ho, 2000). In the past, Since the severe natural disasters of the late 1990s (Harris, 2010), the central government has hastened grasslands were owned by princes, lamaseries, landlords, or clans, and remained in common use by to engage in ecosystem protection and therefore a series of Ecological Construction Programs have been tenant herders. After that, in the collectivist period of New China, the grassland was owned by the initiated, especially after 2000 (Wu et al., 2015). For example, the Conversion of Cropland to Forest production team or People’s commune and the grassland was still managed and used communally by and Grassland Program (also known as the Grain for Green Program) was mainly implemented between all local herders (Hua and Squires, 2015), while local communities had their own institutional structures 2001 and 2010, and covered 25 provinces of China with a total budget of 225 billion RMB. It involved for the use and management of grasslands (Ho, 2000). When the People’s commune was abolished and returning sloping or deserted cropland to forest or artificial grassland (Wang and Bennett, 2008; Liu, grassland degradation was impending, there was an urgent need to fill the institutional vacuum for Lu and Yin, 2010). The Returning Grazing to Grassland Program was launched in 2000. It is one of the grassland use and management (Ho, 2000). Subsequently, there are two major policies that have been largest environmental programs on the grasslands of China. This program aimed at grassland widely employed for the grassland use and management of China, including the land tenure reform conservation by sowing grass on those grasslands experiencing serious degradation, as well as during the 1980s-1990s and a series of Ecological Construction Programs after 2000. The former aims restricting grazing in all program areas, involving grazing bans (either permanently or seasonally), at privatising use rights of grasslands to individual households and displacing traditional common use rotational grazing and setting a deterministic stocking rate (Li et al., 2014; Hua and Squires, 2015). (Banks et al., 2003) and the latter mainly concerns the grassland conservation as well as the maintenance Apart from these two major programs in the 2000s, there were other programs that employed similar of local household livelihoods (Liu et. al., 2016). policy measures for grassland conservation, such as the Program to Combat Desertification around More specifically, land tenure reform started with the Household Production Responsibility System in Beijing and Tianjin. These Ecological Construction Programs were implemented at household-level, the cropland areas of China in the late 1970s and spread to the pastoral areas in the early 1980s. The with or without some compensation in cash, grain or grass seeds for the herders’ economic loss (Suttie land tenure reform in the pastoral areas was implemented by local governments through local et al., 2005). Since 2011, another successive Ecological Construction Program, the Subsidy and regulations, such as the Double Contracting System (“cao xu shuang cheng bao”) in Inner Mongolia. Incentive System for Grassland Conservation (SISGC), has been initiated in eight pastoral provinces of The ownership of livestock was assigned to households on the basis of the size of the household and China, including Inner Mongolia, Xinjiang, Tibet, Qinghai, Sichuan, Gansu, Ningxia and Yunnan. It labour force (Banks et al., 2003). The use rights to grasslands were informally allocated to small groups aims at conserving the condition of grassland and meanwhile improving local herders’ livelihoods and of households, and use rights to hayfields and croplands were allocated to individual households. livestock production through intensification (State Council of the People's Republic of China, 2010). Meanwhile, the marketing system for most livestock products was simultaneously liberalised and only The central government promised a budget of 13.4 billion RMB per year for its implementation between wool remained under state monopoly control (Banks et al., 2003). The first Grassland Law of China, 2011 and 2015. The major regulations on grazing still include the permanent grazing ban and the initiated in 1985, stipulated that the property rights over grasslands are owned either by the state or the deterministic stocking rate, but they are supported with more subsidies or incentives to compensate the collective but that use rights to the grasslands could be contracted out to either households or collectives. herders’ economic loss. Moreover, the users (individual households) of the grasslands are regulated to control their livestock 1.2.5 Research region: the pastoral areas of Inner Mongolia numbers according to a stocking rate in order to avoid overuse of grasslands (Ho, 2000). Then, the assignment of grassland use rights to individual households was strengthened by Chinese governments Inner Mongolia, a province4 located in the arid and semi-arid areas of northern China, has 118.3 million in the 1990s, which greatly fuelled the progress of grassland privatisation and was regarded as the hectares of land with a permanent population of 25 million in 2014. It contains 21.7% of the area of second round of grassland privatisation in China (Li and Zhang 2009). In 2002, the revised Grassland China’s natural grasslands. Approximately 67% of the total land in Inner Mongolia is classified as

Law reaffirmed the shift from state- and collective-oriented land use rights to those of individual 4 The full name is Inner Mongolia Autonomous Region. households (National People's Congress of the People's Republic of China, 2002). The State Council 9 10

for the use and management of grasslandsof (Ho,for 20 and management the use households (National Congress thePeopl People's of and collective from state- Law theshift reaffirmed grassland privatisationinChina(L second roundof of in the1990s,whichgreatly theprogress fuelled use rightsassignmentgrassland toindividualhouseholdswas of strengthened by governments Chinese numbers according to a stocking rate in order to gra ofthe households) (individual theusers Moreover, collective butthatuserights tothegrasslands couldbe rights propertyinitiated in1985,stipulatedthatthe wool remainedmonopoly et control under (Banks state mostlivestock Meanwhile, themarketing system for of households,anduserights andc tohayfields gr to rights use 2003).The et al., (Banks labour force The ownership oflivestockwas assigned tohouseholdsonthebasisof ofthehouseholdand size regulations, such as the Double Contracting shuang (“cao System Inner xu cheng in bao”) M im land tenure was reform inthepastoralareas in the the croplandareasofChina landMore tenure with reform started specifically, of localhouseholdlivelihoods(Liuet. al.,2016). (Bankset al., 2003) and the latter the mainly concerns of grasslands toindividu rights at privatising use a series of duringEcological and the 1980s-1990s widely grassland employed for the useandmanag 2000).Subsequently management(Ho, and grassland use grassland impending, was degradation an therewas Squires, 2015), while local all local herders (Hua and commune production team orPeople’s tenant inthecollectivist herders. that, peri After grasslands were by owned princes, landl lamaseries, Ingrasslands thepast, policy (Ho, for 2000). long-term For long nocomprehensive a time,Chinahad China’s1.2.4 Theoverviewof grassland policies et al.,2007). level governments (Waldron tothepolicy ofhigher- torespond hoping officials orientation local and to misreportingfrom herders late 1970s andspread tothepastoralareas intheearly 1980s.The and the grassland the was s and al households and displacing traditional common use common al householdsanddisplacing traditional avoidoveruse ofgrassla od ofNew China,thegrassland was ownedthe by une was thePeople’scomm abolishedand 00). When Construction Programsaims former 2000.The after the Household Production Responsibility inthe HouseholdProduction System 9 roplands were allocated toindividualroplands were allocated households.

over grasslands are owned either by the state or the state or by the either are owned grasslandsover ement of China, including the landtenure reform ofChina,including ement the i and Zhang 2009). In 2002, the revised Grassland Grassland In therevised 2002, Zhang 2009). i and asslands were informally allocated to small groups tosmall allocated informally asslands were plemented by governmentsthrough local local communities had their own communities had their products was simultaneouslyproducts was liberalised andonly grassland conservation aswell asthemaintenance contracted out to either -oriented land use rights to those of individual of rights tothose landuse -oriented ords, orclans,andremained incommonuse by grassland privatisation and was regarded as the andwas as regarded grassland privatisation e's Republic of China, 2002). The State Council The State 2002). China, e's Republicof urgent needtofillthein sslands are regulated to control regulated their are livestock sslands al., 2003). The first Grassland Law of China, al., 2003). The firstLaw Grassland of China, , there are two major policies that have been been have policies two that major are , there till managed and used communally by by till managedcommunally andused nds (Ho, 2000).Then,the households or collectives. householdsorcollectives. stitutional vacuum for stitutional institutional structures ongolia. 4 and Grassland known Programas the Grain (also for 2000(Wuetal.,initiated, especiallyF 2015). after a se to andengage ecosystem therefore protection in late1990s disasters ofthe Since thesevere natural to individualhouseholds years. for 30to50 According tothecurrent Law ontheContracting ofR tohouseholdsensurecertificates) thelong-term stability in2011(Li, of contractrelations 2012). it ofthecontracted allocation emphasised thespecific Inner Mongolia, a province Inner1.2.5 Research Mongolia region:thepastoral areas of herders’ economic loss. deterministic stockingwith butthey rate, supported are 2011 and2015.Themajor regulations ongrazing the stillincludethepermanentgrazing banand bi 13.4 abudget of government promised The central livestock productionthrough intensification (State meanwhile ofgrassland and condition aims atconserving the Mongolia, Tibet, Inner Xinjiang, China, including Conservation Grassland (SISGC), Incentive for System et al., 2005). Since Ecolog 2011, another successive with orwithout somecompensation incash,grain orgr Beijing Constructi and Tianjin. These Ecological astheProgram around toCombatDesertification such conservation, grassland policyfor measures Apart from these two major programs in the 2000s, there were other that employed programs similar ade rotational grazing andsetting seasonally), or (either permanently grazing bans involving areas, all program grazingrestricting in grasslands those grass on sowing conservation by largest environmentalprograms onthegrasslands Lu and Yin, 2010). The to ReturningGrassl Grazing returning croplandtoforest sloping ordeserted 2001 and2010,covered 25provinces ofChinawith China’s natural grasslands. Approx China’s natural hectares of land with a permanent population of 25 The full name is Inner Mongolia Autonomous Region. full The Autonomous Inner name Region. is Mongolia

4 located semi-arid in the arid and terministic stocking rate (Li et al., 2014; Hua and Squires, 2015). et al.,2014;Huaterministic stocking andSquires,2015). (Li rate imately 67% of the total land in I imatelyof thetotallandin 67% or or artificial grasslandand Bennett, (Wang2008; Liu, on Programs were implemented at household-level, on Programs were athousehold-level, implemented 10 10 (Harris, 2010), the central government has hastened hashastened government central 2010),the (Harris, ries of Ecological Construction Programs have been Programs Ecological been Construction have ries of

Council ofthePeople's Qinghai, Sichuan, Gansu, Ningxia and Yunnan. It Yunnan. Gansu,Ningxia and Qinghai, Sichuan, or example, theConversionofCroplandtoForestor example, and ProgramIt was in2000. launched isoneof the llion RMB per yearits implementationbetween llion RMB for per Green Program) was implementedbetween mainly million in 2014.It 21.7%ofthearea of contains ems (e.g. grassland plots, areas, issuing user-right emsgrassland (e.g. areas, plots, issuing user-right ural Land,ural grassland use rights can be contracted more subsidiesorincentives tocompensatethe of China. This programof China. aimedatgrassland experiencing serious degradation, as well as well as degradation, serious experiencing ass seedsfor theherders’economicloss(Suttie has been initiated in eight pastoral provinces of pastoral eight provinces initiatedhas been in atotalbudget of225billionIt RMB. involved ical Constructionical Program, the Subsidyand areas of northern China,has118.3million ofnorthern areas improving local herders’ livelihoods and improving livelihoodsand local herders’ nner Mongolia is classified as Mongoliaas nner classified is Republic ofChina,2010).

1 grassland (Angerer et al., 2008). (Angerer et al., 2008). It has played an important role in the supply of livestock products and the ecosystem of China. Official statistics showed that the animal husbandry of Inner Mongolia accounted for 45% of the province’s gross output value of agriculture in 2013. Inner Mongolia accounted for 18% of the population of sheep in China in 2013; the mutton output accounted for 22% of total output in China; 6% of cattle were raised in Inner Mongolia, and 8% of beef was produced there.

As a traditionally pastoral region populated by Mongolians, the vast majority of local people maintain their livelihoods by grazing on the grasslands (Angerer et al., 2008). Inner Mongolia has been the major and prior region for implementing the grassland policies in China because of its typical position in China’s grassland resource and livestock production (Li and Huntsinger, 2011). However, it was reported that 90% of the natural grasslands of Inner Mongolia had been degraded to some extent by 2000 and grassland degradation remains problematic since (Waldron et al., 2010; Briske et al., 2015). With society’s growing attention to the grassland ecosystem and continuing grassland degradation, Inner Mongolia has been increasingly emphasised as an ecological barrier of northern China instead of the previous status it had as the main production area of livestock products (Wu et al., 2015). Moreover, Figure 1.2 Inner Mongolia and its 102 counties the provincial government of Inner Mongolia has put emphasis on the development of livestock production in crop farming areas, rather than the pastoral areas, since 2001. For instance, 70% of 1.3 Objectives and research questions livestock production in Inner Mongolia was located in the crop farming area in 2007 (Su, 2010). Given that China’s grasslands are confronted with the challenge of grassland degradation, the increasing According to the current administrative divisions of Inner Mongolia, there are 102 counties, including demand for livestock products and the livelihoods of millions of households, an investigation into the 33 pastoral counties, 21 semi-pastoral counties and 48 counties dominated by crop farming or urban sustainability of grassland use and management is urgently needed. Plenty of policies devoted to districts. Pastoral counties are characterised by traditional grazing systems and natural grassland is the transforming traditional pastoralism have already been implemented to balance the productive and dominant land type. In semi-pastoral counties, both natural grassland and cropland are the dominant ecological functions of grasslands. However, grassland degradation and rural poverty in some pastoral land types, while traditional grazing systems exist and mixed crop-livestock systems also occur. The areas are still worsening in China and there is no consistency within academic circles regarding how to pastoral counties and semi-pastoral counties are usually regarded as the pastoral areas of Inner Mongolia transform traditional pastoralism under contemporary social, economic and political settings. As such, and represent a typical region to investigate the grasslands of China (e.g. Taylor, 2006; Li and Zhang, the objective of this research is to dissect the outcomes of China’s policy interventions on its grassland 2009). Figure 1.2 illustrates the location of Inner Mongolia in China and its three types of counties. system and further discuss the government-driven transformation of traditional pastoralism.

To fulfil this research objective, four research questions will be addressed in four chapters of this dissertation:

Question 1: What are the individual and social optima of rural land allocation by stakeholders: among cultivated land, grassland, forest and other rural land? This question is answered in Chapter 2. This chapter explores the different desires of stakeholders shaping grassland areas, which mirrors the different perceptions of grassland use under government-driven transformation of traditional pastoralism. 11 12

2009). FigureInnerM 1.2illustratesthelocationof and represent a region typical to Taylor, investigateZhang, Ligr the and 2006; asslands ofChina(e.g. c semi-pastoral and pastoral counties The also occur. systems crop-livestock while traditionalgrazing andmixed land types, systems exist counties, both In semi-pastoral dominant landtype. districts. Pastoral counties are characterised by trad 33 pastoralcounties,21semi-past administ current the According to Innerlivestock productionin Mongolia was lo thanth rather production incropfarming areas, Inner of the provincialgovernment the previous statusithadasthemainproductionar as an northern ecologicalof barrier China of instead increasingly emphasised Inner Mongolia has been With society’sgrowingattention tothegra problematic remains grassland degradation 2000 and grasslandsInne of natural the reported that90%of China’s grassland andlivestock product resource and prior region for implementing the Mongoliamajor has Inner been grasslandstheir livelihoodsbygrazing et al., (Angerer 2008). onthe region populatedby Mong pastoral As atraditionally produced there. were cattle of in China;6% output for oftotal 22% shee18% ofthepopulation Mongolia accountedfor grossInner agriculture for outputvalueof in2013. Inner 45%oftheprovince’s accounted Mongolia of husbandry thattheanimal Officialstatisticsshowed ecosystem ofChina. and the livestock products of thesupply rolein animportant It played has et al.,2008). 2008).(Angerer et al., grassland (Angerer rative divisions of Inner Mongolia, there are 102 counties, including counties,including areInnerrative 102 divisionsof Mongolia,there oral counties and 48countiesdomina oral countiesand the grassland policies inChinab the grassland ounties are usually regarded as the as the regarded usually ounties are Mongolia hasputemphasison Mongolia ssland ecosystem and continuinggrassland ecosystemand de ssland cated in the crop farming area in 2007 (Su, 2010). in2007(Su,2010). inthecropcated farming area e pastoral areas, since 2001. For instance, 70% of For instance,70% since2001. e pastoralareas, ea of livestock (Wu et products al., 2015). Moreover, 11 11

ongolia in China and its three types of counties. counties. ongolia of threetypes inChinaand its itional grazinggrasslanditional systems andnatural is the ion (Li andHuntsinger, 2011).However,itwas r Mongolia had been degraded to some extent by extent degraded tosome hadbeen r Mongolia natural grassland and cropland are the dominant thedominant are cropland and grassland natural olians, the vast majority of local people maintain people majority oflocal olians, thevast since (Waldron et al., 2010; Briske et al., 2015). al.,2015). 2010;Briskeet et al., since (Waldron p inChina2013;the muttonoutputaccounted raised in Inner Mongolia, and 8% of beef was beef of Innerand 8% Mongolia, in raised ecause ofitstypical position in pastoral areas of Inner Mongolia Inner Mongolia of areas pastoral the developmentoflivestock ted by urbancrop farming or gradation, gradation, demand for livestock products andth demand forlivestock products Given that China’s grasslands are 1.3 Objectives questions andresearch pastoralism. pastoralism. under grassland use of perceptions different stakeh of desires different the chapter explores grassland, ruralland? forestcultivated land, andother Question 1: dissertation: ques research To fulfil this research objective, four pastoralism. transformation oftraditional systemfurther thegovernment-driven and discuss ofChina’spolicy istodissecttheoutcomes onitsgrasslandthe objectiveofthisresearch interventions undercontemporary transform traditionalpastoralism areas Chinaand are stillworsening thereisnoconsistency in circlesregarding to how withinacademic grassla However, grasslands. functions ecological of transforming traditionalpastorali is sustainability useandmanagement grassland of What are the individual and social optima of rural land allocation by stakeholders: among by among stakeholders: allocation individualandsocialoptimaofruralWhat are land the Figure 1.2 Inner Mongolia and its 102 counties FigureInner andits102counties Mongolia 1.2 confrontedthe challenge of with sm have already been implemented to balance the implemented productive and been sm have already e livelihoods ofmillions ofhouseholds,aninvestigation intothe olders shaping grassland areas, which mirrors the which mirrors areas, grassland olders shaping 12 12 government-driven transformation of traditional of traditional transformation government-driven

nd degradation and rural poverty in some pastoral some pastoral rural povertynd degradation in and urgently needed. Plenty of policies devoted to policiesdevoted Plenty of urgentlyneeded. tions will be addressed in four chapters of this ofthis chapters infourtions willbeaddressed social, economic and political settings. As such, Assuch, politicalsettings. and social, economic This question is answered in Chapter 2. This This 2.This questionisanswered inChapter grassland degradation, the increasing degradation,grassland theincreasing

1 Question 2: What is the impact of land tenure reform on the grassland condition in the pastoral areas of grassland system (e.g. grazing area, water systems, and animal husbandry); (2) grassland resource units China? This question is answered in Chapter 3. Chapter 3 investigates the ecological effect of land (e.g. forage, water, and animals); (3) users (e.g. herders, farmers and others); and (4) governance system tenure reform on the grassland system, which presents the ecological outcome of the transformation of (e.g. government and policy interventions). These four subsystems are relatively separable but interact pastoralism with the abandonment of communal tenure. to produce outcomes at the SES level, which in turn affect the subsystems (Ostrom, 2009). This thesis aims to reveal the impacts of governance systems upon the grassland systems, resource units and users. Question 3: What is the impact of land tenure reform on livestock production in the pastoral areas of One of the broad questions this SES framework is used to solve is what impacts are likely to result from China? This question is answered in Chapter 4. This chapter explores the economic effect of land tenure using particular rules for the governance, ownership, and use of a resource system and resource units in reform on the grassland system, and presents the economic outcome of transforming pastoral areas, in a specific technological, socioeconomic, and political environment (Ostrom, 2007). Long-term terms of livestock changes in population as well as productivity. sustainability depends on rules matching the attributes of the resource system, resource units, and users

Question 4: What is the impact of Ecological Construction Programs on the grassland condition in the (Ostrom, 2009). This study is devoted to examining whether the policy interventions of China are pastoral areas of China? This question is answered in Chapter 5. This chapter investigates the effect of beneficial to the grassland system, grassland resource units and users in the long term. Ecological Construction Programs on the grassland system, which mirrors the ecological outcome of the transformation of pastoralism, through the use of deterministic stocking rates and grazing ban.

To summarise, the research objective is specifically addressed by evaluating the ecological and economic effects of the land tenure reform and Ecological Construction Programs that have been employed to intervene in the grassland use and management in China. The aim of this research is to contribute to the approach toward sustainable use and management of grasslands in China or other regions that have similar circumstances to China’s, by providing empirical evidence for the academic arguments on the transformation of traditional pastoralism.

1.4 Materials and methods Figure 1.3 Conceptual framework for analysing social-ecological systems of grasslands 1.4.1 Conceptual framework (Adapted from Ostrom’s framework for analysing social-ecological systems (Ostrom, 2009))

All resources used by humans are embedded in complex social-ecological systems (SESs). Ostrom More specifically, this study focuses on evaluating the ecological and economic impacts of China’s (2009) introduced a general framework for analysing sustainability of SESs, which integrates the policy interventions on grasslands, which essentially reflects the outcomes of governance systems on different frameworks of different disciplines and provides a common method for dissecting the complex grassland systems under the social, economic and political settings in which said systems are embedded. SESs. It helps to identify the potentially relevant variables and their subcomponents to design the data The outcomes for grassland systems are specifically measured by grassland condition and livestock collection, to conduct the fieldwork, and to perform the analysis of findings about the sustainability of production. The social, economic and political settings considered refer to market forces, technological complex SESs (Ostrom, 2009). As such, this research employs Ostrom’s framework as the starting point development and climate changes among other factors. The governance system relates to the specific for conducting our study of linked SESs of grasslands. rules for the use and management of China’s grasslands, including the land tenure reform studied in Chapters 3 and 4 and the Ecological Construction Programs studied in Chapter 5. In addition, the Figure 1.3 illustrates the conceptual framework adapted from Ostrom’s framework for analysing the conflicts of different users over resource units, specifically the grassland area, are investigated in SESs of grasslands. It shows the four first-level core subsystems of the SESs of grasslands, the Chapter 2. relationships between subsystems and the link to social, economic, and political settings and related ecosystems. The four first-level core subsystems in the SESs of grasslands are identified as: (1)

13 14

ecosystems. The four first-level core subsystems in the SESs of grasslands are identified as: (1) as: (1) identified in the SESs of grasslands are subsystems core first-level four The ecosystems. relationships link to soc subsystems and the between the grasslands, SESsof the core of subsystems It the four first-level SESs of grasslands. shows Figure 1.3 illustrates the ad conceptual framework of linkedSESsgrasslands. for conducting ourstudy SESs(Ostrom, complex asthesta 2009). Assuch,thisresearch employs Ostrom’sframework collection, to conduct the fieldwork, SESs. It vari thepotentially helpstoidentify relevant different frameworksofdifferent whichintegrates the SESs, sustainability analysing of for framework general a (2009) introduced Ostrom systems (SESs). social-ecological in complex All resources are embedded used by humans 1.4.1 Conceptual framework methods 1.4 Materials and arguments on the of transformation traditional pastoralism. circumstances regions similar thathave sustainableuse approach toward contribute tothe to managementgrassland aimofthisresearch is and the inChina.The employed use to intervene in Ecol and landtenurereform ofthe economic effects researchTo summarise, is specifi the objective through theuse the transformationofpastoralism, Ecological Construction Programs onthegrassland system, whichmirrorstheecological outcomeof pastoral ofThis China? question areas is answered in Question is 4: What the impact ofEcological Construc productivity. inpopulationaswell terms oflivestockchanges in areas, pastoral oftransforming outcome economic the and presents system,grassland reform the on inChapter4.This Thisquestionisanswered China? Question 3: ofcommunaltenure. with theabandonment pastoralism tenure reform onthegrassland thetransformation system, of whichpresents theecological outcomeof land effect of ecological Chapter Thisquestionis China? 3investigatesthe inChapteranswered 3. Question is the impact 2:What on of th land tenure reform What istheimpactoflandte What disciplines andprovidesacommonmethod for dissecting thecomplex and to perform the analysis of fi to China’s, by providing empirical evidence for the academic academic the to China’s, byproviding empirical evidence for nure reform onlivestockproduct nure reform 13 13 of deterministic stocking rates and grazing ban. deterministicof stockingratesgrazing and ban. cally addressed by evaluating the ecological and and ecological the evaluating by addressed cally

apted from the Ostrom’sframework for analysing chapter explores the economic effect of land tenure tenure land of effect economic the explores chapter and management of grasslands in China or other inChinaor ofgrasslands and management ables and to theirdesign subcomponents the data Chapter 5. This chapter of investigates the effect ial, economic, and political settings and related settings and andpolitical ial, economic, tion Programs onthegrassland conditioninthe ogical Construction Programs that have been ogical Construction thathavebeen Programs e grassland in the pastoralareas of condition ndings about the sustainability of ion in the pastoral areas of ion inthepastoralareas rting point Chapter 2. Chapter 2. units,spec resource users over of different conflicts Ecological Construction Chapters 3and4the rules forgrassla the use of China’s and management climate changes development and andpolitical settingsproduction. Thesocial,economic arespecifical systems grassland The outcomesfor politi economic and social, grassland the systems under policy interventions whichessentia ongrasslands, evaluating focuses on More thisstudy specifically, beneficialgrasslandre system,grassland to the (Ostrom, 2009).Thisstudy isdevotedtoexaminin sustainability rulesmatching theattrib on depends socioeconomic, and technological, a specific in units and resource system a resource of ownership,anduse the governance, rules for using particular used to is thisSESframework questions broad One ofthe aims to reveal the impacts of governanceupon systems SESlevel, whichinturn af to produceoutcomesatthe interact government(e.g.four but relatively and policy separable subsystems interventions). are These (3)animals); (e.g. users(e.g. forage, herde and water, grassland system (e.g. grazing area, water systems, grassland systemgrazing water (e.g. systems, area, (Adapted from Ostrom’s framework for analysing social (Adapted framework foranalysing from Ostrom’s Figure 1.3 Conceptual framework for analysing analysing for Figure framework 1.3Conceptual among other factors. The governance system relates to the specific relates tothespecific system governance factors. The otheramong source units and users in the long term. inthelongsource unitsandusers term. political environment (Ostrom, 2007). Long-term Long-term (Ostrom, 2007). political environment utes oftheresource resource system, units,andusers 14 14 lly of governancereflects the outcomes systemson and animal husbandry); (2) grassland(2)and animal resource husbandry); units

ly measured by grassland conditionandlivestock ly by measured rs, farmers and others); and (4) governance system system governance and(4) and others); rs, farmers g thepolicy whether interventions ofChina are the ecological and economic impacts of China’s economic and ecological the Programs studied in Chapter 5. In inChapter5. addition,the Programs studied nds, including thelandtenurereform studiedin considered refer to market forces, technological market forces,technological refer to considered cal settings in which said systems are embedded.said systems settings cal inwhich ifically the grassland in ifically area, are the investigated fect thesubsystems (Ostrom, 2009).Thisthesis the grassland resource systems, unitsandusers. social-ecological systems of grasslands ofgrasslands systems social-ecological solve is what ar impacts -ecological systems (Ostrom, 2009)) (Ostrom,2009)) systems -ecological e likely to result from to e result likely

1 1.4.2 Sampling and data collection reports from the provincial government as well as through questionnaires. The questionnaires were sent to the Animal Husbandry Bureau of each county and answered by local experts. In the second phase, In Chapter 2, the analysis of optimal land allocation is based on a case study. The sample county, Tai interviews were conducted by telephone to confirm the answers that were received. Apart from the Pusi County5, is a semi-pastoral county located in the eco-fragile region of Inner Mongolia, which has questionnaires, interviews were conducted with the officers working in the provincial institutes of been characterised by economic backwardness and ecological degradation. The county covers different Animal Husbandry of Inner Mongolia and with local herders to verify the information thoroughly. types of rural land and involves a variety of different stakeholders with divergent interests in rural land Remote sensing and GIS (geographic information system) were employed to obtain data about the use. It is a typical region to investigate the conflicts of rural land allocation in China, reflected in the grassland condition across extensive grasslands over a long period. The data on livestock production different preferences for grasslands by different stakeholders. For the data collection, interviews with and socio-economic and climate factors was collected based on existing statistical data gathered by the representatives of different types of stakeholders were conducted face to face to reveal their governments. preference on land use, involving cultivated land, grassland, forest land and other rural land. The secondary data on socio-economic indicators was collected from statistics and reports from local 1.4.3 Model selection governments. In Chapter 2, the model for investigating the optimal land allocation is based on welfare economics, Chapters 3, 4 and 5 focus on policy evaluation. Panel data at county-level is used to answer the research specifically utilising the utility function and social welfare function. The utility function presents the questions. Policy interventions have occurred over different years and there are differences in the utilities of different stakeholders for land allocation based on their preference. The Analytic Hierarchy process of implementation among counties. These differences within and between counties are Process (AHP) is used to quantify the preference of stakeholders for different types of land. The social presented by the panel data. For the analysis of land tenure reform in Chapters 3 and 4, all of the counties welfare function is derived based on the utility functions and the weights of stakeholders in land of Inner Mongolia were considered as possible sample counties. After excluding the counties with allocation, in which three scenarios are employed to estimate the weights. Ultimately, the divergences invalid survey feedback, and after dropping the urbanised counties where changes in the grassland area among individual optima derived from the utility functions, the social optimum derived from the social can be attributed mainly to changes in land use types 60 counties are retained as the research sample, welfare function, and actual rural land allocation, illustrate the conflicts in preferences of different including 29 pastoral counties, 18 semi-pastoral counties and 13 crop farming counties6. The research stakeholders regarding rural land use. period covers 1985-2008, spanning the main period for implementing land tenure reform on the Chapters 3, 4 and 5 employ panel data to dissect the effectiveness of policy interventions. The fixed grasslands of Inner Mongolia. For the analysis of Ecological Construction Programs in Chapter 5, all of effects model, which is used to study the causes of changes within an entity over time, is employed to the counties in the pastoral areas, including 33 pastoral counties and 21 semi-pastoral counties, were conduct an empirical estimation. The fixed effects model has been widely used in economic research used in the research sample. It should be noted that the crop-farming counties and urban districts are (e.g., Fergusson et al., 2002; Huan et al., 2006). Compared with the multiple regression model, the excluded from the research sample because they are not the typical regions included for the Ecological advantage of the fixed effects model is the control over unobserved fixed factors that could affect the Construction Programs. This research period spans 14 years from 2001 to 2014, involving the years estimation (Liu et al., 2010; Verbeek, 2012). Theoretically, the fixed effects model employs within before and after the implementation of the specific Ecological Construction Program studied (i.e. the transformation to remove all time-invariant (fixed) explanatory variables, i.e. the model is performed in Subsidy and Incentive System for Grassland Conservation). deviations from individual means (Verbeek, 2012). As such, it provides a method that takes observable In terms of the data collection for the policy evaluation, the required dataset includes four categories of explanatory variables as well as unobservable time-invariant variables into account, but the estimation information: the policy interventions, the grassland condition, the livestock production, and the socio- does not depend on the value of time-invariant (fixed) variables, such as soil nutrient levels, soil depth, economic and climate conditions. The information about policy interventions, such as the progress of slope steepness, geographical positions of grasslands, in the case of this study (Verbeek, 2012). In land tenure reform and Ecological Construction Programs, was gathered based on the statistics and addition, the feasibility of the fixed effects model has been tested using statistical tests based on the specific dataset used in each chapter.

5 Also named as Tai Pusi Qi in Chinese. 6 The crop farming counties considered also have grasslands and experienced the same land tenure reform as the pastoral area. 15 16

including 18 semi-pas counties, 29 pastoral sample, research asthe are retained 60counties types use to changes inland attributedmainly can be surveyand after invalid the dropping urbani feedback, sa Inneras possible consideredof Mongolia were presented For by the panel data. the analysis of land These among counties.process implementation of questions. Policy over interventionshaveoccurred on policyeval and 5 focus Chapters 3, 4 governments. indicators was on socio-economic data secondary involving cultivated land, preference onlanduse, typesthe representativesdifferent of stakeholders of grasslands different by for different preferences Ituse. conflicts typical is a to the investigate region types of land rural and involves a variety of different st and backwardness by economic been characterised 6 5 Pusi County In of optimal Chapter 2, the analysis land allocation 1.4.2 Sampling anddata collection area. area. land tenure reform and Ecological Constructionland tenure reform Prog and about The information and conditions. climate economic grasslainformation: thepolicy the nd condition, the livestock production, interventions, and the socio- collectionevaluati for thepolicy data In ofthe terms SubsidyIncentive forand System Grassland Conservation). Ec implementation the of thebefore after specific and Construction Programs. Thisresearch periodsp are they sample because research from the excluded used in Itthe research sample. should be noted the countiesinpastora Innergrasslands For of theanalysis Mongolia. ofEcol spanning themainpe period covers1985-2008, The crop farming The an also considered have grasslands counties Chinese. Qiin Pusi asTai Also named 5 , is a semi-pastoral county intheeco located , isasemi-pastoral l areas,including counties 33pastoral uation. Panel data at county-level is used to answer the research research the isusedtoanswer at county-level Paneldata uation. toral counties and 13 counties toral that the crop-farming counties and urban districtsare counties thatthecrop-farming ans 14 years 14 years 2001to2014,involvingans from the stakeholders. For the data collection, interviewsFor data with the stakeholders. 15 15 riod for implementinglandtenure reform on the tenure reform inChapters tenure reform ecological degradation. The county covers different Thecountyecological degradation. covers different

uding the counties with excl mple counties.After on, the required dataset includes four categories of includescategories on, the four required dataset different years and there are differences in the areyears in the differences anddifferent there not the typical regions included for the Ecological Ecological included forthe regions typical not the of rural land allocation in China, reflected in the in inChina,reflected allocation ofrural land d experienced the same land tenure pastoral reform the as land d experiencedsame the grassland, forest land and other rural land. The land.The andotherrural grassland, forest land collected collected from statistics and reports from local is based on a case study. Tai Thesamplecounty, is basedonacase ogical Construction Programs inChapter5,all of in the grassland area changes inthe sed countieswhere rams, was gatheredbased onthestatisticsand was rams, akeholders withdivergent interestsinruralland ological Construction Program studied (i.e. the ological Program the Construction studied (i.e. differences within and between counties are within andbetween differences were conducted face to face to reveal their tofacereveal face were conducted policy interventions, such as the progress of progress as the such policy interventions, -fragile regionInner of Mongolia,whichhas crop farming counties farming crop and 21 semi-pastoral counties,were and 21semi-pastoral 3 and4,allofthecounties 6 . The research . The reports from the provincial government as well as t wellgovernment as as provincial reports the from specific dataset used in each chapter. chapter. used ineach specific dataset effectsmodel of thefixed addition, thefeasibility geographicalgrasslands positionsof slope steepness, (fixed) time-invariant of the value does on not depend aswell asunobservabletime explanatory variables deviations from individualmeans (Verbeek,2012). in isperformed themodel variables,i.e. explanatory (fixed) alltime-invariant remove transformation to estimation (Liu Theore Verbeek, et al., 2010; 2012). effectsfixed contro modeladvantage of is the the model,the regression with themultiple Compared 2006). et al., Huan al., 2002; Fergusson et (e.g., effects The fixed mod empirical estimation. conduct an causesof the tostudy which isused effects model, todissect data panel employ and 5 Chapters 3,4 stakeholders regarding rural land use. illu welfare function, ruralactual land and allocation, among individualoptimaderivedfromf theutility toestimatetheweights. employed allocation, inwhichthreescenariosare Ultimately, the divergences functionwelfare isderivedbased ontheutility func Process (AHP) is used to quantify preference of the allocation stakeholders forland utilities ofdifferent utilisingspecifically function. theutility Theutility welfare function the function presents andsocial In Chapter 2,themodelfor investigating the optimallandallocationisbased onwelfare economics, 1.4.3 Model selection governments. collectedwas climate factors and and socio-economic onlivestockproduction grasslands period.Thedata er along extensive ov grassland conditionacross (geographic sy Remote sensing information andGIS Animal HusbandryInner Mongolia of andwithloca institutes of the provincial in working e officers with th were conducted questionnaires, interviews by telephon conducted interviews were to and the Animal Husbandry county Bureau of each e to confirm the answers that were received. Apart from the Apart from received. were that the answers confirm e to l over unobserved fixed factors that could affect the the could affect that factors fixed l overunobserved unctions, thesocialoptimum 16 16 hrough questionnaires. The questionnaires were sent were questionnaires hrough The questionnaires. the effectiveness of policy interventions. The fixed fixed The ofpolicy interventions. the effectiveness -invariant variables into account, but-invariant into the estimation account, variables

As such, it provides a method thattakes observable As such,itprovidesa has been tested using statistical tests based on the statistical testsbased on testedusing has been changes within an entity over time, is employed to to time,isemployed entityover within an changes stakeholders for different types of land. The social stakeholders for types different ofland.The social based on their preference. The Analytic Hierarchy Hierarchy Analytic The theirpreference. based on tically, the fixed effects modelemploys within effects thefixed tically, answered by local experts. In thesecondphase, local by experts. answered l herders to verify the information thoroughly. l herderstoverify information the variables, such as soil nutrient levels, soil depth, nutrientlevels, soildepth, as soil such variables, , in the case of this study (Verbeek, 2012). In (Verbeek, study 2012). this of case , in the stem) were employed toobtaindataaboutthe stem) were employed tions andtheweights ofstakeholdersinland based on existing statistical data gathered bygathered statistical data existing on based strate theconflictsinpreferences ofdifferent el has been widely used in economic research widely el has been derived from the social

1 1.5 Contribution of the dissertation factors are controlled for. These chapters provide some references to academic debates over the transformation of pastoralism, or grassland privatisation instead of communal tenure. The policy In general, given the challenge of grassland degradation and increasing demand on livestock production, evaluation is conducted based on panel data which considers the spatial and temporal differences of this research makes a contribution to providing references for improving the policy interventions toward policy implementation. Moreover, the existing literature about grassland privatisation concentrates on more sustainable use and management of grasslands. Moreover, this study provides an empirical impacts on the grassland ecosystem, but there is a lack of research referring to the impact on local analysis of SESs of grasslands in China based on the common framework introduced by Ostrom. The livestock production, examined here in Chapter 4. In this regard, despite the unreliability of official research findings can be used in other studies that employ the same framework regarding different statistics on livestock, these research findings are useful for detecting general trends, if not specific countries and different disciplines. This research hopes to increase the knowledge needed to enhance numbers. the sustainability of complex SESs of grasslands, as well as the understanding of the transformation of traditional pastoralism. Chapter 5 summarises the Ecological Construction Programs in pastoral areas of China, and explains why the actual grassland degradation has not been prevented as effectively as expected. Based on the This research systematically discusses the major policy interventions for the use and management of research results, this chapter points out the potential for improving grassland conservation policies. grasslands in China, and further evaluates their effectiveness for the grassland condition and livestock Moreover, it provides the empirical evidence for the scientific debate about the equilibrium model and production. The policy evaluation essentially presents the outcomes of government-driven the non-equilibrium model regarding deterministic or stochastic stocking rates. transformation of traditional pastoralism in China. Although the transformation of pastoralism and the specific effects of the policy interventions on grassland systems have been investigated by governments 1.6 Organization of the dissertation and scholars (e.g. Li and Zhang, 2009; Li et al., 2014; Hua and Squires, 2015), there is a lack of studies This dissertation has six chapters, including a general introduction (Chapter 1), four research chapters providing quantitative and empirical analysis based on large-scale areas and long-term observation. As that focus on the research objective of this dissertation (Chapter 2, 3, 4 and 5), and a general discussion such, remote sensing and county-level statistics are combined for this research, the dataset of which (Chapter 6). The research scheme of the four research chapters is presented in Figure 1.4. involves the extensive grasslands of Inner Mongolia and spans decades. Economic models are employed to evaluate changes in grassland systems, targeting the specific policy interventions. This makes it possible to disentangle the policy effectiveness and control for the potential impacts of climate, socio- economic factors and time trend on the grassland system. This provides an example of the combined use of remote sensing and economics for academic research. In addition, the different preferences of different stakeholders for grassland use under the existing policy intervention is qualified based on utility functions. The specific relevance of each chapter is explained next.

Chapter 2 presents the preference of different stakeholders for land allocation, and thus mirrors the different perceptions on grassland use. The novel contribution is the methodology, which combines welfare economics, the method Analytic Hierarchy Process (AHP) and scenario analysis. Although the outcomes are based on a specific case, the methodology is generic and likely to be valid for measuring conflicts between stakeholders over land use in other regions. The results provide interesting insights Figure 1.4 The research scheme of the four research chapters that can help to steer future land-use policies considering the different economic and ecological interests of different stakeholders. Under the conceptual framework for analysing SESs of grasslands adapted from Ostrom’s framework, the four research chapters elaborate on the outcomes of the governance system over the grassland system Chapter 3 and 4 illustrate the progress of land tenure reform in the pastoral areas of China and investigate and its users. Specifically, Chapters 3 and 4 explore the impacts of land tenure reform (grassland the results of privatisation of grasslands on grassland condition and livestock production when other privatisation) on grassland degradation and livestock production, which presents the outcomes of policy 17 18

the results ofprivatisationgra Chapter 3and4illustrate theprogress oflandtenurerefo of different stakeholders. interests and ecological economic ing the different policiesconsider tosteerfutureland-use that canhelp useinot overland stakeholders conflicts between themethodology specific case, outcomes are basedona economics, themethodAnalyticwelfare Hierarchy novel grassland use.The on different perceptions different preference of st the Chapter 2presents utility next. is explained chapter The of functions. each specific relevance different grassl for stakeholders re economicsfor academic use ofremote and sensing grassland trend oneconomic the factors sy and time possible todisentangle an thepolicy effectiveness ingrassland targeting systems, to evaluatechanges employed modelsare Economic decades. andspans Inner Mongolia of grasslands theextensive involves such, remotestatistics county-level are sensingand providing empiricalanalysis quantitativeand based LiZhang,Li etal.,2014; and scholars(e.g. 2009; and ofthepolicyspecific effects ongrassla interventions pastoralisminChina. transformation oftraditional production. The policy essentiallyevaluation pre their grasslands in China,and further evaluates This research major systematically discussesthe traditional pastoralism. SESsthe sustainability ofgrasslands, ofcomplex as Thisresearch countries anddifferent disciplines. usedinotherstudiesthatresearch canbe findings employ the same regarding framework different analysis ofSESsgrasslands inChinabase grassla management of more sustainableuseand tothis providing researcha contribution makes refe Ingeneral,given ofgrassland degra thechallenge the 1.5 Contributionof dissertation and use under the existing policyand useundertheexisting isqualified basedon intervention sslands on grassland conditionand sslands ongrassland d on the common frameworkintroducedd on bythe Ostrom. The effectiveness for thegrassland conditionandlivestock dation and increasing demand on livestock production, onlivestockproduction, demand andincreasing dation hopes to increase the knowledge needed to enhance toenhance needed theknowledge hopes toincrease akeholders for land allocation, and thus mirrors the andthusmirrors allocation, land for akeholders policyof useandmanagement for interventions the d control for the potential impactsofclimate,socio- potential the d controlfor 17 17 her regions. The results provide interestingher regions. results insights The rences for improving the policy rences forimproving toward interventions the Although the transformation of pastoralism and the the ofpastoralismand Although thetransformation

Process (AHP) and scenario analysis. Althoughthe analysis. Process (AHP)andscenario contribution isthemethodology,combines which nds. Moreover, this study provides an empirical nds. Moreover,empirical thisstudy providesan well as the understanding of the transformation of well ofthetransformation astheunderstanding on large-scale areas and long-term and areas observation.on large-scale As nd systems havebeeninvestigatedgovernments by the specific policy thespecific interventions. Thismakes it stem. This provides an example ofthecombined an example stem. Thisprovides combined for thisresearch,combined for ofwhich thedataset Hua and Squires,2015),thereisalack Hua ofstudies search. Insearch. addition, the ofdifferent preferences rm in the pastoral of China and areas investigate sents the outcomes of government-driven theoutcomesof government-driven sents is genericis andlikely validfor tobe measuring livestock production when other livestock productionwhen other privatisation) on grassland degradati privatisation) ongrassland and its users. Specifically, ex 3 and 4 Chapters system overthegrassland s of the governance system elaborateontheoutcome chapters the four research from Ostrom’sframework, ofgrasslands adapted SESs foranalysing framework conceptual Under the (Chapter 6). The research scheme of the four research chapters is presented in Figure 1.4. ispresented chapters four research schemeofthe research (Chapter 6).The objective ofthisdisserta that focus ontheresearch ge a chapters,This dissertation including has six the 1.6 Organizationof dissertation the non-equilibriummodelregardingdetermi evaluation isconductedbasedonpa orgrassland priva pastoralism, transformation of provid chapters These for. are controlled factors Moreover, it provides the empirical evidence for the empiricalMoreover, for evidence it provides the poten pointsoutthe chapter results,research this pre notbeen has grassland degradation actual why the Chapter 5 summarises the Ecological Construction P numbers. statistics onlivestock,these findings use research are livestock production,examinedhere4. inChapter grassland ecosystem, impacts on the policy literatexisting implementation.Moreover, the Figure 1.4 The research scheme of the four research chapters chapters four research ofthe scheme research Figure 1.4 The on andlivestockproduction,which nel data which considersthespat nel datawhich but there is a lack of research research of a lack but thereis nistic or stochastic stocking rates. nistic orstochasticstockingrates. neral introduction (Chapter 1), four research chapters chapters research 1),four neral introduction (Chapter plore the impacts of land tenure reform (grassland tenurereform land plore theimpactsof tion (Chapter 2, 3, 4 and 5), and a general and a discussion tion (Chapter2,3,4and5), 18 18 tial for improving grassland conservation policies. grassland policies. conservation tial for improving

e some references to academic debates over the over debates academic to e somereferences tisation instead of communal tenure. The policy Thepolicy tenure. communal tisation insteadof In thisregard, theunreliability despite ofofficial scientific debate about th scientific about debate ure about grasslandon privatisation concentrates rograms inrograms pastoral areas of China, and explains vented as effectively as expected. Based on the Based expected. as as effectively vented ful for detecting general trends, if not specific referring to the impact on local thereferring impact on local to ial and temporal differences of andtemporaldifferences ial presents the outcomes of policy theoutcomesof policy presents e equilibriummodeland

1 interventions for the grassland system. As previously defined, the grassland system in this research includes grassland conditions as well as livestock production. Grassland degradation is presented by grassland area and quality. Livestock production is indicated by livestock population and meat output. Chapter 2 Chapter 5 examines the effectiveness of Ecological Construction Programs on grassland conservation, which presents the outcome of particular rules for the use and management of grassland resource units. In addition, Chapter 2 focuses on investigating the preference over land use by different users under existing social, economic, and political settings. The preference over grassland use is influenced by policies as well as by the outcomes of policy interventions over grassland systems, which is indicated by dashed lines in Figure 1.4. Another significant question marked by the long dashed lines is the effect Rural land allocation in eco-fragile regions of Ecological Construction Programs on livestock production, which will not be covered by this research but which is an appropriate candidate for future research. It should be noted that the main period for the of northern China implementation of land tenure reform in China is between the 1980s and 1990s, which predates the wide implementation of the Ecological Construction Programs since around 2000. It is assumed that this indication allows us to distinguish these two major policies as well as their main impacts by focusing on different research periods.

19 20 interventions for the grassland system. As previously defined, the grassland system in this research includes grassland conditions as well as livestock production. Grassland degradation is presented by grassland area and quality. Livestock production is indicated by livestock population and meat output. Chapter 2 Chapter 5 examines the effectiveness of Ecological Construction Programs on grassland conservation, which presents the outcome of particular rules for the use and management of grassland resource units. In addition, Chapter 2 focuses on investigating the preference over land use by different users under existing social, economic, and political settings. The preference over grassland use is influenced by policies as well as by the outcomes of policy interventions over grassland systems, which is indicated by dashed lines in Figure 1.4. Another significant question marked by the long dashed lines is the effect Rural land allocation in eco-fragile regions of Ecological Construction Programs on livestock production, which will not be covered by this research but which is an appropriate candidate for future research. It should be noted that the main period for the of northern China implementation of land tenure reform in China is between the 1980s and 1990s, which predates the wide implementation of the Ecological Construction Programs since around 2000. It is assumed that this indication allows us to distinguish these two major policies as well as their main impacts by focusing on different research periods.

19 20

2. RURAL LAND ALLOCATION IN ECO-FRAGILE REGIONS OF NORTHERN CHINA7 2.1 Introduction

There is increased pressure on rural land use in many parts of the world (Petit, 2009; Williams and ABSTRACT: This paper takes a welfare economics perspective to study rural land allocation in eco- Schirmer, 2012). As a result of economic development, technological progress, environmental change fragile regions of northern China and explores the trade-off between stakeholder preferences over the and policy and market forces, the total area of rural land is decreasing worldwide (Liu et al., 2010; economic and ecological benefits of rural land use. The preference of different stakeholder groups for Verburg et al., 2008). This raises the question of how to efficiently allocate the decreasing amount of various types of land is quantified by the Analytic Hierarchy Process, and the individual and social rural land, particularly for a populous country such as China. As a basic production factor, rural land is optimum of the allocation of land are investigated using stakeholder utility functions and the social used for production and subsistence to satisfy immediate human needs for food, fuel, and ecosystem welfare function. Our results show the conflicting views of different stakeholder groups and the services (DeFries et al., 2007; Millennium Ecosystem Assessment, 2003). On the other hand, its potential equilibrium in the allocation of rural land under different scenarios. multifunctional role has also triggered land use divergence and conflicts because of differing land use targets among stakeholders (Barker and Selman, 1990; Rambonilaza and Dachary-Bernard, 2007; Yang KEYWORDS: rural land allocation, individual optima, social optima et al., 2012; Pacione, 2013). This paper investigates the optimal allocation of rural land taking into account the ecological and economic benefits for different stakeholders in society.

Changes in rural land use are not only attributed to micro-level household decisions on land use, but are also affected by national or macro-level land use strategies. For example, in China, several national programs have been introduced since the 1990s that attempt to steer rural land allocation in order to satisfy not only human needs for economic development but also the provision of ecosystem services. One of the main programs is the Sloping Land Conversion Program (SLCP, also known as the ‘Grain for Green’ program), initiated by the Chinese central government in 2000 in order to convert sloping cropland into forest or grassland (Liu et al., 2010). Such national programs that are directed at ecological conservation, may constrain the economic activities of local residents, while the majority of rural households in China rely on rural land for their livelihoods and economic benefits. As such, rural land use change associated with economic development and ecological conservation inevitably results in land use conflicts among different stakeholders.

These problems are particularly severe in eco-fragile regions where ecological issues and collective impoverishment are causing serious land use conflicts in the process of rural land use change. This is reflected in overgrazing, poor implementation of eco-environmental policies, deforestation, soil erosion and ecosystem deterioration. One fifth of the territory of China is identified as eco-fragile regions facing such problems (Yuan et al., 2011). In this paper, we therefore focus on land use change and conflicts in the farming-grazing transitional zone in the eco-fragile regions in northern China.

Optimal rural land allocation can be studied either from the macro perspective referring to the global or regional social-economic situation, or from the perspective of land use decisions at the rural household level. The scientific research on land use and land cover change has received increased attention since

7 Paper by Min Liu, Wim Heijman, Xueqin Zhu, Liesbeth Dries, and Jikun Huang. Rangeland Journal. doi: the 1990s because of prominent problems due to population growth, overuse of natural resources and 10.1071/rj15069 21 22

10.1071/rj15069 7 ECO-FRAGILE REGIONSOFNORTHERNALLOCATION IN 2. RURALLAND CHINA economic and ecological benefitsof ecological economic and fragile Chinaandexplores regions the ofnorthern ABSTRACT: Thispapertakes awelfare pe economics optima, socialoptima rural landallocation,individual KEYWORDS: underdifferentpotential equilibriuminthe scenarios. allocationofrural land welfare function. Our results show the conflicting utilityoptimum oftheallocationfunctions oflandare using andthesocial investigated stakeholder Analytic quantified by the landis of various types Paper by Min Liu, Wim Heijman, Xueqin Zhu, andJikun by Huang. Liesbeth Paper Xueqin Liu, Dries, Min Wim Heijman, rural land use. The preference of different stakeholder groupsfor stakeholder ofdifferent Thepreference rural landuse. 21 21

trade-off between stakeholder preferences over the stakeholder preferences over the between trade-off Hierarchy Process, and the individual and social theindividualand and Process, Hierarchy views of different stakeholder groups and the groups andthe viewsofdifferent stakeholder rspective to study rural land allocation ineco- rspective tostudyrural land allocation Rangeland Journal. doi: 7

Schirmer, 2012). As a result of ec onruralThere landusein pressure isincreased 2.1 Introduction the 1990sbecauseofprominent problemsdueto populati level. cove Theand scientific land on research land use situation,orfrom attheruralregional theperspective household social-economic oflandusedecisions fr either studied canbe allocation ruralland Optimal inregions the eco-fragile zone the farming-grazing transitional China. in northern Insuch problems(Yuan 2011). etal., thispaper,we One fifthoftheterritoryand ecosystem deterioration. of policies, o-environmental ec of poor implementation reflected inovergrazing, conflicts landuse serious impoverishment arecausing ineco-fragile These problemsare severe particularly different stakeholders. land useconflictsamong economic development associated with use change households inChinarely onrural landfor theirlivelihoodsandeconomicbene ec conservation, the constrain may (Liucropland into forest or grassland et al., 2010). Su program),for initiatedby Green’ Chinesec the Conve istheSlopingLand mainprograms One ofthe satisfy notonly needsfor economic devel human programs introduced havebeen sincethe1990sthat atte also affected by land use st or macro-level national to attributed arenotonly use Changes inrural land economicbenefits ecological and account the investig paper This 2013). et al.,2012;Pacione, Selman, and (Barker stakeholders targets among usedive also triggered land multifunctional rolehas 2007; Millennium Ec al., et (DeFries services andecosystem food,fuel, for subsistence tosatisfy immediatehumanneeds used for and production rural land, a particularly for populous country as such how questionof raises the 2008).This Verburg etal., rural area of and policy and market forces, the total onomic development, technologicalonomic development, progress, environmental change onomic activities of onomic activities localthe residents, while majority of rural for different stakeholders in society. insociety. stakeholders for different entral government in2000 entral in order toconvertsloping 1990; Rambonilaza and Dachary-Bernard, 2007; Yang 2007; Yang RambonilazaandDachary-Bernard, 1990; osystem Assessment, 2003). On the other hand, its hand, other Onthe 2003). Assessment, osystem opment butalso the provisionofecosystem services. ates the optimal allocation of rural land taking into ates the optimal allocation of rural land taking many parts of the world (Petit, 2009; Williams and many of the world parts micro-level household decisions on land use, but are butare onlanduse, decisions micro-level household 22 22

om the macro perspective referring to the global or global referring tothe perspective om themacro therefore focus on land use change and conflicts in onlanduse change and focus therefore rgence and conflicts because of differing land use land ofdiffering conflictsbecause and rgence ch national programs that are directed atecological ch nationalprograms are directed that rategies. in China, severalFor example, national and ecological conservation inevitablyand ecologicalin resultsconservation China. As a basic production factor, land rural is rsion Program (SLCP, also known as the ‘Grain alsoknownasthe rsion Program (SLCP, to efficiently allocate the decreasing amount ofamount thedecreasing allocate toefficiently regions where issuesandcollective ecological land is decreasing worldwide (Liu et al., 2010; et al.,2010; (Liu worldwide land isdecreasing in the process of rural land use change. This is This change. rural landuse in theprocessof r change has received increased attention has r change since China is identified as eco-fragile regionsChina isidentifiedaseco-fragile facing mpt to steer rural land allocation inorder to mpt tosteerrural landallocation on growth, of na overuse deforestation, soil erosion deforestation, soil erosion fits. As such, rural land fits. Assuch,rural land tural resources and tural resources and

2 environmental deterioration (Turner Ii et al., 1993; Qasim et al., 2013). Most research has focused on analysing the land cover dynamics using remote sensing data (Turner Ii et al., 1994; Zhan et al., 2007) and the driving forces of land use change based on qualitative analysis, econometrics or game theory (Angelsen, 2001). Others have tried to determine an optimal landscape (Heijman and Mouche, 2013). The current paper will add to the existing literature by using a welfare economics perspective to investigate the equilibrium of rural land allocation, taking into account the interplay between ecological and economic demand. In other words, we will study the optimal rural land allocation based on the trade-offs between economic development and ecological conservation as seen by different stakeholders.

The remainder of this paper is structured as follows. First, we present a conceptual framework of stakeholder interaction in rural land use decisions. Different stakeholder groups are identified based on various land use targets. Next, we derive the theoretical model starting with the utility function and preferences over desired rural land allocation for each stakeholder group. The optimisation of rural land allocation in terms of the benefits to society is explored by deriving the welfare function. This approach Figure 2.1 The framework for analysing the interaction between different stakeholders8 is in line with other studies that have modelled environmental problems through welfare analysis (Gerlagh and Keyzer, 2004; Gerlagha and Keyzer, 2003; Zhu, 2004). In the empirical part of the paper, In the process of rural land allocation, the governance system is represented by the rules and regulations we first describe the data collection methodology. The Analytic Hierarchy Process (AHP) is used to set at the level of the government to regulate rural land management for ecological conservation and quantify the preference of each stakeholder group for different types of land. The optimal land allocation economic development. Public authorities implement the specific rules related to rural land use. Rural is derived based on these preferences and different scenarios concerning the weights that are assigned households, as the users of the rural land, are affected directly by these rules in terms of land use to each stakeholder group in the social welfare function. The paper concludes with a discussion on the decisions for subsistence, production or recreation. In this paper, we assume that public authorities and differences between existing rural land use policies and the socially optimal allocation as studied in this rural households are the social agents in the social-ecological systems of rural land use. Public paper. authorities are divided into public actors with an economic objective (focusing on economic development) and public actors with an ecological objective (focusing on the conservation of natural 2.2 Conceptual framework resources for current and future generations). This is a simplified representation of reality which

To understand the complex relationships in the process of rural land use allocation, we use the concept assumes that these public objectives can be separated. Among the private stakeholders we distinguish of social-ecological systems (SESs). SESs represent complex adaptive systems in which social and between farmers, who engage in both crop production and animal husbandry, and herders, who are biophysical agents are interacting at multiple temporal and spatial scales (Levin et al., 2013). Ostrom solely dependent on the use of grasslands for livestock production. As such, we divide the social agents (2007; 2010) introduced a diagnostic framework for the study of complex social-ecological systems, into four stakeholder groups, namely herders, farmers, ecological public authority and economic public considering the resource units, resource system, governance system and users as four core subsystems authority. to link social, economic and regulatory settings and related ecosystems (Ostrom, 2009). This has In the next sections, we investigate the optimal land allocation based on the trade-offs between enabled researchers to develop a common language that crosses social and ecological disciplines to ecological conservation and economic development among different stakeholders. Specifically, we analyse how interactions among a variety of factors affect outcomes. On the basis of Ostrom’s analyse the individual and social optimum of rural land allocation. The optimum is based on the different framework, this paper focuses attention on the influence of the relationship between the governance system and users on the outcome of rural land use. Figure 2.1 demonstrates the analytical framework in 8 We follow the rural land classification by AQSIQ (General Administration of Quality Supervision, Inspection and this context. Quarantine of the People's Republic of China) whereby rural land is divided into cultivated land, forest land, rangeland and other rural land (such as land used for raising animals, agricultural facilities, agricultural roads, pit-ponds, fishponds, irrigation, drying grains and forming ridges among croplands) according to the functions of land use in 2002. 23 24

this context. this context. Figu use. rural land of ontheoutcome system users and atte focuses framework, this paper f of a variety among howinteractions analyse enabled researchers to develop a common language to linksocial,economicandregulatory settings subsystems four core as and users governance system units, resource system, resource considering the (2007; 2010)introduced adiagnosticfor the framework biophysical interacting are agents at multiple temporal (SESs). SESs systems representof social-ecological concept the allocation, we use relationships use rural intheprocess land of complex To understandthe 2.2 Conceptualframework paper. a rural existing landusepolicies between differences group in thesocialwelfarefunc to eachstakeholder and different preferences is derivedon these based allocation The optimalland land. types of different groupfor stakeholder each of quantifythe preference collection methodology. data we Th the first describe Gerlagha(Gerlaghand Keyzer, 2004; andKeyzer, is inlinewithother studiesthathavem is ex benefits tosociety ofthe allocation interms preferences over desired rural land allocation for e various we land use derive targets. the theoreNext, stakeholder interactioninrurallandusedecisions. struc The remainder ofthispaperis ecologica developmentand economic between trade-offs In other words,and economicdemand. we willst investigateallocation, equilibrium ofruralland the existingadd liter to the will The currentpaper (Angelsen, have2001). tried Others to determine based land use change of forces and the driving dynamicsremote landanalysing cover sensin using the Ii (Turner environmental deterioration ntion on the influence of the relationship between the governance thegovernance relationship between the of ntion on the influence tured as follows. First, we present a conceptual framework of First, wea conceptual framework tured as follows. present et al., 1993; Qasim et al., 2013). Most research has focused on focused has research 2013).Most Qasim etal., 1993; etal., odelled environmental problems through welfare analysis throughwelfareanalysis environmentalproblems odelled plored by deriving the welfare function. This approach approach function. This welfare deriving the plored by on qualitative analysis, econometrics or game theory econometricsor qualitativeanalysis, on an optimal landscape (Heijman and Mouche,(Heijman 2013). an optimal landscape ature by using a welfare economics perspective to perspective economics welfare using a ature by actors affect outcomes. On thebasisofOstrom’s outcomes. actors affect ach stakeholder group. optimisationofrural land The ach stakeholder and related ecosystems (Ostrom, 2009).Thishas andrelated ecosystems 23 23 Different stakeholder groups are identifiedbasedon stakeholder Different 2003; Zhu,In empiricalpartofthepaper, 2004). the taking between ecological into the account interplay udy theoptimalrural landallocationbasedonthe

scenarios concerning the weights that are assigned assigned weights thatare the concerning scenarios tion. The paper concludes with a discussion ontion. The paper concludes the nd thesocially optimal allocation asstudiedin this tical modelstartingwiththeutility function and complex adaptive systems complex in which social and that crosses social and ecological disciplines to social and ecologicalthat crosses disciplines e Analytic to Hierarchy Process (AHP) is used and spatial scales (Levin et al., 2013). Ostrom 2013). Ostrom et al., (Levin and spatial scales g data (Turner Ii et al., 1994; Zhan et al., 2007) 2007) al., et 1994;Zhan al., Ii et (Turner g data re 2.1 demonstrates the analytical framework in in framework theanalytical demonstrates re 2.1 l conservation as seen by different stakeholders. stakeholders. by different asseen conservation l study of complex social-ecological systems, study social-ecological complex of irrigation, dryinggrainsforming and ridges among croplands) according tothe functions of land use in 2002. other rural land (such asland used for raising agr animals, of Republic Chin People's the Quarantine of 8 analyse the individual and socialoptimumofrural la theindividualand analyse and economic ecological conservation Insections, we the next investigate the optimal authority. public economic authority and public ecological farmers, namely herders, groups, into four stakeholder grassl of theuse solely on dependent crop produc in both engage farmers, who between assumes that these public objectives separate can be resources for current and future generations). This ofnatural the conservation (focusing on objective an ecological actorswith public development) and into publicauthorities are divided actors with rural social households agents are the in the soc In productionorrecreation. decisions forsubsistence, households, as the users of the rural land, are affec th Public authorities implement economic development. set at the level of the government to regulate rur In theof ruralprocess land allocation, the governance sy followWe the rural by landAQSIQ (General classification Figure 2.1 The framework for analysing the for analysing framework Figure 2.1The ands for livestock production. As suc production. As livestock ands for a) whereby rural land is divided into cultivated land, forest land, rangeland and rangeland land, forest land, intowhereby land is divided cultivated rural a) development among different stakeholders. Specifically, we Specifically, stakeholders. among different development al land management for ecological conservation and 24 24 an economic objective (focusing on economic land allocation based on the trade-offs between between land allocation based on the trade-offs

nd allocation.Theoptimumisbasedonthedifferent icultural facilities, agricultural roads, pit-ponds, fishponds, roads,icultural pit-ponds, facilities, agricultural tion and animal husbandry, and herders, who are who andherders, tion andanimalhusbandry, interaction between stakeholders different ial-ecological systemsof rural Public land use. ial-ecological d. Among the private stakeholders we distinguish d. Among theprivate this paper, we assume that public authorities and thatpublicauthorities thispaper,weassume ted directlyrules in terms of land by use these is asimplifiedrepresentation ofreality which Administration Inspection and of Quality Supervision, stem is representedstem by rules and regulations the e specific rules related to rural land use. Rural use.Rural toruralland rules related e specific h, we social agents divideh, the 8

2 preferences of each stakeholder group as to the rural land value with a combination of diverse land represent the four stakeholder groups: herders, farmers, ecological public

functions, such as various degrees of economic function, ecological function and social function from authority and ൌ ͳǡ economic ʹǡ ͵Ͷ public authority, respectively. is the area of the cultivated land which is the cultivated land, rangeland, forest land and other rural land. We describe the role of each stakeholder allocated by stakeholder group i for crop production. ݈௖௨௟ሺ௜ሻ is the area of the rangeland allocated by group in the allocation of rural land as follows: the local ecological public authority aims at protecting stakeholder group i for grazing animals and fulfilling ݈an௚௥௔ሺ௜ሻ ecological protection function. is the the regional ecosystem on behalf of the national government and social ecological demand, the area of the forest land demanded by stakeholder group i. is the area of the rest of the݈௙௢௥ሺ௜ሻ rural land economic public authority achieves the local government’s targets for local economic development, and arranged by stakeholder group i. are the ݈utility௢௧௛ሺ௜ሻ elasticities of stakeholder group i on farmers and herders rely on the land for their agricultural production. The herders graze their animals land allocation among the cultivatedܽ௜ǡܾ௜ ǡܿland,௜ܽ݊݀݀ rang௜ eland, forest land and other rural land. The land on the rangeland, while the farmers rely primarily on the cultivated land. Section 3 will present the constraint is as follows: theoretical model which formalizes the conceptual framework.

2.3 Model specification ௖௨௟ሺ௜ሻ ௥௔௡ሺ௜ሻ ௙௢௥ሺ௜ሻ ௢௧௛ሺ௜ሻ ݈ ൅݈ ൅݈ ൅݈ ൌ ܮǡሺʹሻ According to welfare economics, different agents have different demands on the consumption of various ݓ݅ݐ݄݈௖௨௟ሺ௜ሻǡ݈௥௔௡ሺ௜ሻǡ݈௙௢௥ሺ௜ሻǡ݈௢௧௛ሺ௜ሻ ൒ ͲǢ Ͳ ൑ ܽ௜ǡܾ௜ǡܿ௜ǡ݀௜ ൑ ͳǢܽ௜ ൅ܾ௜ ൅ܿ௜ ൅݀௜ ൌ ͳǤ goods/services. This has led to the study of individual utility and social welfare optimisation for where L is the total area of the rural land in the research region. Equation (2) illustrates that the sum of combinations of different goods characteristics (Heijman and Mouche, 2013; Perman et al., 2011). In the cultivated land, rangeland, forest land and other rural land equals the total area of rural land. To find our research context, we consider a hypothetical society consisting of four individuals and four goods. the individual optimum of land allocation, we maximise the utility functions (see Appendix 2-A for the More specifically, the four individuals represent the four stakeholder groups (herders, farmers, detailed derivation). With the help of the Lagrange optimisation procedure this gives: ecological public authority and economic public authority), and the four sets of goods are cultivated

land, rangeland, forest land and other rural land. The four stakeholder groups derive utility from ଴ ݈௖௨௟ሺ௜ሻ ൌܽ௜ ൈ ܮǡሺ͵ሻ consumption of the four types of land. A Cobb–Douglas form of utility function is used to investigate ଴ ௥௔௡ሺ௜ሻ ௜ individual utility and social welfare. The Cobb–Douglas utility function has been applied in previous ݈ ൌܾ ൈ ܮǡሺͶሻ ଴ research to study the optimal consumption, leisure, investment and voluntary retirement problem for an ௙௢௥ሺ௜ሻ ௜ ݈ ൌܿ ൈ ܮǡሺͷሻ agent (e.g. Koo et al., 2013), the trade-off between goods and leisure (e.g. Train and McFadden, 1978), ଴ ௢௧௛ሺ௜ሻ ௜ the ecological discount rate (e.g. Gollier, 2010) and a typical low-rent housing lessees' welfare model ݈ are theൌ resulting ݀ ൈ ܮǤ individual optima of stakeholderሺ͸ሻ group i, indicating ଴ ଴ ଴ ଴ (e.g. Wu et al., 2012). It is an appropriate method to investigate the individual and social optimum in the݈௖௨௟ሺ௜ሻ individuallyǡ݈௥௔௡ሺ௜ሻǡ݈௙௢௥ሺ௜ሻ rational݈ܽ݊݀ optimum௢௧௛ሺ௜ሻ depending on the preference of stakeholder group i for rural land use utilitarianism, as studied here. and taking into account land constraints.

2.3.1 The utility function of each stakeholder group 2.3.2 The social welfare function

In this section, the individual optimum is derived through the utility function of each stakeholder group. Bergson and Samuelson introduced the social welfare function, which sums up the utility functions of As with any resource, the allocation of land may be considered optimal when the aggregate social all the individuals in the society (Pollak, 1979). We assume that the four stakeholder groups represent returns from its various uses are maximised (Lopez et al., 1994). Here, we will consider the economic all of the social agents of land use in the eco-fragile regions, therefore our social welfare function is and ecological uses of rural land by each stakeholder group in order to investigate their utility. presented by the configuration of the four stakeholder groups’ utilities, subject to land constraints. Based

Considering that the target of stakeholder group i is to maximise its utility (Ui) subject to land constraints, on the utility functions of each stakeholder group, the Cobb-Douglas form of the social welfare function the utility function in the Cobb-Douglas form is, is:

௔೔ ௕೔ ௖೔ ௗ೔ ܷ௜ ൌ݈௖௨௟ሺ௜ሻ݈௥௔௡ሺ௜ሻ݈௙௢௥ሺ௜ሻ݈௢௧௛ሺ௜ሻǡሺͳሻ 25 26

the utilityform is, intheCobb-Douglas function Considering target that the group ofto stakeholder maximise its utility i is ( byand ecological stakehol usesofrural each land (Lopez maximised are uses returns itsvarious from may land allocationof the resource, As withany isderived In optimum theindividual section, this stakeholder ofeach group 2.3.1 Theutilityfunction utilitarianism, asstudied here. (e.g. et al., 2012). Wu It is an appropriate method to welfare housing lessees' a typical low-rent model and Gollier, 2010)(e.g.the ecological rate discount betwee thetrade-off al., 2013), Kooet (e.g. agent consumption,leisure, i research theoptimal tostudy individual utility welfare. andsocial TheCobb consumption ofthefourA Cobb types ofland. rural land. andother forest land land, rangeland, ecological public authority and economic public authori More individualsrepresent specifically, thefour so hypothetical weour researchconsider a context, (Hei goodscombinations ofdifferent characteristics goods/services. Thishasledtothestudydual utility ofindivi welfare andsocial optimisationfor ha agents different economics, towelfare According 2.3 Modelspecification framework. theoretical theconceptual formalizes modelwhich Section3willpresentthe farmers the cultivated land. rely while on the on therangeland, primarily animals grazeherders their The agricultural production. on thelandforrely their farmers andherders economic publicauthority thelocal governmen achieves the regional ecosystem onbehalfofthenationa authority aims publicecological local at protecting as follows: landthe group rural allocation in of the the cultivatedland,rangeland, rur forest landand other fu economic degrees of functions, suchasvarious group stakeholder each of preferences ܷ ௜ ൌ݈ ௖௨௟ሺ௜ሻ ௔ ೔ as to the rural land value with a combination ofdiverseland combination witha rural landvalue as tothe ݈ ௥௔௡ሺ௜ሻ ௕ ೔ ݈ ௙௢௥ሺ௜ሻ –Douglas ofutility isused toinvestigate form function ௖ –Douglas utility hasbeenappliedinprevious function ೔ through the utility function of each stakeholder group. through ofeachstakeholder the utility function n goods and leisure (e.g. Train and McFadden, 1978), 1978), Train andMcFadden, (e.g.n goodsandleisure 25 25 nction, ecological function and social function from andsocial function nction, ecological ciety consisting offour individualsandfour goods.

l government and social ecological demand, the the ecological demand, l government andsocial ve different demands on the consumption of various various the consumptionof on demands ve different thefour stakeholdergroups farmers, (herders, nvestment and voluntary retirement problem for an nvestment andvoluntary forretirement problem be considered optimal when the aggregate social aggregate social optimalwhen the considered be et al., 1994). Here, we will consider the economic economic will considerthe we Here, 1994). et al., ݈ jman and Mouche, 2013; Perman et al., 2011). In 2011). Permanetal., jman andMouche,2013; The four stakeholder groups derive utility groups derive from stakeholder The four investigate theindividual andsocialoptimum in ௢௧௛ሺ௜ሻ ௗ der groupinorder toinvestigate theirutility. al land. We describe the role of each stakeholder the roleofeachstakeholder describe al land.We ೔ t’s targets for local ec ty), and the four sets of goods are cultivated cultivated goods sets of are four andthe ty), ǡ U i ) subject to land constraints, ) subjecttoland onomic development, and onomic development,and ሺͳሻ

arranged by stakeholder group byarranged stakeholder allocated by stakeholder group i for crop production. group production. iforallocated by crop stakeholder stakeholder group authority and economic public authority, respectively. authority, public and economic authority is: group, th each stakeholder on theutility of functions g presented ofthefour configuration stakeholder by the eco-frag agents landuseinthe of all ofthesocial all theindividualsin the We 1979). society (Pollak, Bergson andSamuelsonintroduced thesocialwelfare whichsums function, uptheutility functions of 2.3.2 Thesocialwelfarefunction constraints. and taking land intoaccount the individually pre optimum depending rational onthe constraint isasfollows: land allocation among the cultivatedrang land land, allocation among the area of the forest land demanded by stakeholder group by forest stakeholder area land demanded of the detailed derivation). With the help of the Lagrange optimisation procedure this gives: detailed derivation).Lagrange Withthehelpthis optimisationprocedure ofthe the individualoptimumof landallocation,we maximis rural and other land forest rangeland, the cultivatedland, of illustratesthatthesum region.research (2) L Equation of therurallandin where isthetotalarea ݈ ௖௨௟ሺ௜ሻ ଴ ǡ݈

௥௔௡ሺ௜ሻ ଴ ǡʹ ͵Ͷ ʹǡ ͳǡ ൌ ݓ݅ݐ݄݈ ǡ݈ ௙௢௥ሺ௜ሻ ଴ ௖௨௟ሺ௜ሻ i for grazing animals and fulfillingprotection function. an ecological ݈ܽ݊݀ ǡ݈ ௥௔௡ሺ௜ሻ farmers, ecological publicgroups: herders, four stakeholder represent the ௢௧௛ሺ௜ሻ ଴ ǡ݈ ݈ ௙௢௥ሺ௜ሻ ݈ ݈ ݈ ݈ ௖௨௟ሺ௜ሻ i. ௖௨௟ሺ௜ሻ ଴ are theresulting individualoptimaofstakeholdergroup ௥௔௡ሺ௜ሻ ଴ ௢௧௛ሺ௜ሻ ଴ ௙௢௥ሺ௜ሻ ଴ ܽ ǡ݈ ௜ ǡܾ ൅݈ ൌܽ ௢௧௛ሺ௜ሻ ൌ݀ ൌ ൌܾ ൌܿ ௜ ǡܿ ௥௔௡ሺ௜ሻ ௜ ௜ ௜ ௜ ௜ ܽ݊݀݀ ܮǡ ൈ Ǣ ܽ ൑ Ͳ ͲǢ ൒ ܮǡ ൈ ܮǡ ൈ ܮǤ ൈ ൅݈ ௙௢௥ሺ௜ሻ ௜ are theutility group ion ofstakeholder elasticities 26 26 eland, forest land and other rural land. The land other land rural land. The eland,and forest land ile regions, therefore our thereforeour ile regions,

e Cobb-Douglas form of the social welfare function function welfare thesocial form of e Cobb-Douglas assume that the four stakeholder groups assumethatthefour represent stakeholder roups’ utilities, subject to land constraints. Based roups’ utilities,subjecttolandconstraints. e theutility for the (see functions 2-A Appendix ൅݈ ௜ ference of stakeholder group i for rural land use group ruralland ifor stakeholder of ference ݈ ݈ i. ǡܾ ௚௥௔ሺ௜ሻ ௖௨௟ሺ௜ሻ land equals the total area of rural land. To find find To land. rural of area total equals the land ݈ ௜ ௢௧௛ሺ௜ሻ ௢௧௛ሺ௜ሻ ǡܿ ௜ is the area of the cultivated land which is landwhichis thecultivated of isthearea is the area of the rangeland allocated by ǡ݀ restis the area the rural of the of land ܮǡሺʹሻ ൌ ௜ ͳǢܽ ൑ ௜ ൅ܾ social welfare function is function social welfare ௜ ൅ܿ ሺ͵ሻ ሺͷሻ ሺͶሻ ሺ͸ሻ ௜ ൅݀ ௜ ͳǤ ൌ i , indicating ݈ ௙௢௥ሺ௜ሻ

is the is

2 of local rural households’ income is derived from agricultural production. Its total population is 211,146, ସ ఉ೔ and there are 171,500 and 39,646 residents living in rural and urban areas, respectively. Rural residents ܹ ൌෑܷ௜ ǡሺ͹ሻ include 168,514 farmers and 2,986 herders. Tai Pusi County’s total area measures 341,473 hectares, subject to: ௜ୀଵ including 322,100 hectares of rural land, 14,613 hectares of urban and industrial land and 4,760 hectares of unused land. Figure 2.2 presents the 7 townships of Tai Pusi County. The townships are divided into ௖௨௟ሺ௦ሻ ௥௔௡ሺ௦ሻ ௙௢௥ሺ௦ሻ ௢௧௛ሺ௦ሻ ݈ ൅݈ ൅݈ , ൅݈ ൌ ܮǡሺ, ͺሻ pasture area and agricultural area because of different natural and social factors. In the pasture area, ସ ௜ ௜ ௖௨௟ሺ௦ሻ ௥௔௡ሺ௦ሻ ௙௢௥ሺ௦ሻ ௢௧௛ሺ௦ሻ where the herders are living, the local livelihoods depend primarily on grazing and very few herders ,where W is the socialݓ݅ݐ݄ welfare,σ௜ୀଵ ߚ ൌ ͳܽ݊݀ߚ ൒Ͳ ݈ ǡ݈are theǡ݈ areas ǡ݈of cultivated൒Ͳ land, rangeland engage in crop farming. The farmers living in the agricultural area are involved in both crop farming forestland and other rural land in௖௨௟ሺ௦ሻ the society.௥௔௡ሺ௦ሻ ௙௢௥ሺ௦ሻ ௢௧௛ሺ௦ሻ are the utilities of herders, farmers, ݈ ǡ݈ ǡ݈ ǡ݈ and livestock breeding, but only with pen grazing due to the limitation of a strict policy that bans grazing ecological public authority and economic publicଵ authorityଶ ଷ respectivelyସ based on the social land ܷ ǡܷ ǡܷ ܷܽ݊݀ since 2008. In recent years, with increasing population and eco-environmental policies, conflicts on allocation. is the welfare weight of stakeholder group i in the society. rural land use among local farmers, herders and the government have intensified. ߚ௜ To find the socially optimal land allocation, we maximise the social welfare function (see Appendix 2- B for the detailed derivation). This gives:

ସ כ ݈௖௨௟ሺ௦ሻ ൌܮൈ෍ߚ௜ܽ௜ǡሺͻሻ ௜ୀଵ

ସ כ ݈௥௔௡ሺ௦ሻ ൌܮൈ෍ߚ௜ܾ௜ ǡሺͳͲሻ ௜ୀଵ

ସ כ ݈௙௢௥ሺ௦ሻ ൌܮൈ෍ߚ௜ܿ௜ ǡሺͳͳሻ ௜ୀଵ ସ Figure 2.2 The eco-fragile regions in northern China (left) and Tai Pusi County (right) כ ௢௧௛ሺ௦ሻ ௜ ௜ Source: Ouyang (2013) and Farming and Grazing Bureau, Tai Pusi County. ݈ ൌܮൈ෍ߚ݀ Ǥሺͳʹሻ are the resulting௜ୀଵ socially optimal rural land allocations, which כ כ כ כ illustrates݈௖௨௟ሺ௦ሻǡ݈௥௔௡ሺ௦ሻ theǡ݈ optimum௙௢௥ሺ௦ሻ݈ܽ݊݀ for ௢௧௛ሺ௦ሻsocial welfare based on the preferences on land use and welfare weights of each stakeholder group in the society.

2.4 Research region and data collection

An empirical application of the above theoretical model was conducted in Tai Pusi County, located in the eco-fragile region in the mixed farming zone in northern China. Tai Pusi faces both economic backwardness and ecological degradation (Chen, 2007) and is on the southern edge of Otindag Sandy Land, the nearest crucial sand source of sandstorms in Beijing. Its ecosystem plays a significant role in preventing sandstorms from reaching Beijing and various eco-environmental policies with restrictive land use are implemented here. Tai Pusi is one of the poverty-stricken counties of China and two thirds

27 28

B for the detailed derivation). This gives: Thisgives: Bdetailed derivation). for the 2- landallocation, (seemaximiseAppendix thesocialwelfareTo findthesocially we function optimal where Wisthesocialwelfare, forestland and other rural landinthesociety.forestland andother allocation. land social onthe based respectively authority public economic and authority ecological public subject to: subject to: An empirical application of the a collection data 2.4 Researchregion and thesociety.group in each stakeholder illustrates the based optimum on for social welfare e poverty-stricken counties ofChinaandtwothirds counties here. implemented land Pusi use is are one Tai of th e poverty-stricken va and Beijing reaching preventing sandstormsfrom plays inIts inBeijing.ecosystem sandstorms role sand source Land, of asignificant nearest crucial the 2007) (Chen, degradation and ecological backwardness farming the mixed regionthe eco-fragile in zone ݈ ௖௨௟ሺ௦ሻ כ ǡ݈ ௥௔௡ሺ௦ሻ כ ߚ ௜ is the welfare weight of stakeholder group group isthewelfare weight stakeholder of ǡ݈ ௙௢௥ሺ௦ሻ כ ݓ݅ݐ݄ ݈ܽ݊݀ σ ௜ୀଵ ସ ݈ ௢௧௛ሺ௦ሻ כ ௖௨௟ሺ௦ሻ ߚ ௜ ݈ ͳܽ݊݀ߚ ൌ ௖௨௟ሺ௦ሻ which are allocations, theresulting optimalrural land socially bove theoreticalmodelwas conducted ൅݈ ൌෑܷ ܹ ݈ ݈ ݈ ݈ ௙௢௥ሺ௦ሻ כ ௥௔௡ሺ௦ሻ ǡ݈ ௥௔௡ሺ௦ሻ כ ௢௧௛ሺ௦ሻ ௖௨௟ሺ௦ሻ כ כ ௥௔௡ሺ௦ሻ ௜ୀଵ ସ ൌܮൈ෍ߚ ൌܮൈ෍ߚ ൅݈ ൌܮൈ෍ߚ ൌܮൈ෍ߚ ௜ ǡ݈ ൒Ͳ ௙௢௥ሺ௦ሻ ௙௢௥ሺ௦ሻ ௜ ఉ ܷ ೔ ଵ ǡ ,݈ ǡܷ 27 27 ௜ୀଵ ௜ୀଵ in northern China. Tai Pusi faces both economic Pusifaces innorthernChina.Tai botheconomic ௜ୀଵ ௜ୀଵ

ସ the preferences on land use and welfare weights of weights useandwelfare onland the preferences ସ ସ ସ ǡ݈ ଶ ௖௨௟ሺ௦ሻ ൅݈ rious eco-environmental policies with restrictive eco-environmental rious ǡܷ ௢௧௛ሺ௦ሻ ௜ ௜ ௜ ܿ ௜ ௢௧௛ሺ௦ሻ ଷ and is on the southern edge of Otindag Sandy Otindag Sandy of edge on the southern and is ܽ i ݀ ܾ ௜ in the society. ܷܽ݊݀ ௜ ǡ݈ ௜ ௜ ǡ Ǥ ǡ are land,rangeland, theareasofcultivated ǡ ௥௔௡ሺ௦ሻ ܮǡሺ ൌ ସ ǡ݈ are farmers, theutilitiesofherders, ௙௢௥ሺ௦ሻ in TaiPusi County, in located ǡ݈ ௢௧௛ሺ௦ሻ ൒Ͳ ሺ͹ሻ , ሺͳͳሻ ሺͳʹሻ ሺͳͲሻ ሺͻሻ ͺሻ

government intensified. have and farmers,herders among the rural local landuse years, pop In withincreasing recent since 2008. onlyand livestock breeding, but with due pen grazing livingthe The farmers in farming. engage in crop livelihoods are thelocal living, where herders the ofdifferen because area agricultural area and pasture of presentsof unusedland.Figure the7townships 2.2 4,760hectares industriallandand and urban of hectares land,14,613 rural of hectares including 322,100 Pusi Tai and2,986herders. include 168,514farmers 39,646 residents living 171,500 and in and ruralthere are of local ruralincomeisderivedfrom households’ Figure 2.2 The eco-fragile regionsFigure eco-fragile innorther 2.2The Source: Ouyang (2013) and Farming and Grazing Bureau, Tai Pusi County. Bureau,Source: andFarmingGrazing TaiPusiCounty. Ouyang (2013) and agricultural production. Its totalpopulationis211,146 production. agricultural ulation and eco-environmental policies, conflicts on policies,conflicts ulation andeco-environmental 28 28 agricultural area are involved in both crop farming farming inbothcrop involved are area agricultural

depend primarily on grazingfewdepend and primarily very on herders n China (left) and Tai Pusi County (right) and TaiPusiCounty (right) (left) n China to thelimitationofastrict County’s total area measures 341,473 hectares, measures County’sarea 341,473hectares, total t natural and social factors. In the pasture area, area, In thepasture factors. and social t natural Tai PusiCounty. townshipsare dividedinto The and urban areas, respectively. Rural residents respectively. andurban areas, policygrazing thatbans

2 Figure 2.3 shows the change of the cultivated land, forestland, rangeland and other rural land in Tai Pusi Protection Agency and the Forestry Bureau represent the ecological public authorities. Based on the County from 1995 to 2009. interviews, the Analytic Hierarchy Process (AHP) is employed to evaluate the weight values of (see Appendix 2-C for the different steps in the AHP process). AHP, introduced by 400 Thomas௜ ௜ ௜ Saaty ௜(1980), is one of the most effective tools for dealing with complex decision-making. A 350 ܽ ǡܾǡܿ݀ 300 large number of studies in different fields have been published based on AHP, such as planning, resource Cultivated land 250 Forestland allocation, conflict resolution and optimisation (Vaidya and Kumar, 2006). The values of parameters 200 Rangeland that we derived using AHP are shown in Table 2.1. 150 Other rural land 100 ௜ ௜ ௜ ௜ Total ܽ ǡܾǡܿ݀ 50 Table 2.1 Values of parameters ௜ ௜ ௜ ௜ 0 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 ܽ ǡܾǡܿܽ݊݀݀ Cultivated land Rangeland Forest land Other rural land Total Herders 0.2084 0.5286 0.1606 0.1024 1 Farmers 0.4317 0.2695 0.2202 0.0786 1 9 Figure 2.3 The dynamic change of rural land use in Tai Pusi County Ecological 0.2121 0.3173 0.3746 0.0960 1 Source: Land Resources Bureau, Tai Pusi County. public authority The largest proportion of rural land has always been represented by rangeland, with cultivated land in Economic public 0.4304 0.3134 0.1154 0.1409 1 authority second position and other rural land in the last position. In general, rural land use change has occurred, but has had little effect on the order of the dominant land use type. The total area of rural land is 2.5.2 The welfare weight

࢏ decreasing slightly and transitions of land use between cultivated land, rangeland, forest land and other The social welfare weight ሺࢼwillሻ be calculated for three scenarios. rural land are clearly observable. The area of rangeland decreased rapidly during 2007 and 2008 after a ߚ௜ smooth decrease between 2002 and 2006. The area of cultivated land decreased from 1999 until 2008, Scenario 1: is considered equal for all four stakeholder groups. The prevailing practice in choosing while conversely, the area of forest land area started increasing from 1999 with a sharp growth in 2002 social welfareߚ௜ weights is to pick the utilitarian weights (Yang and Nordhaus, 2006). Following this and 2003. approach we assume that each stakeholder group has the same weight in the social welfare function, i.e. . This scenario is considered the scenario of social fairness in land allocation. 2.5 Estimation of the model parameters ߚଵ ൌߚଶ ൌߚଷ ൌߚସ ൌ ͳȀͶ Scenario 2: is considered as the Negishi weight, introduced by Negishi (1972). In this scenario, the 2.5.1 The parameters social welfareߚ௜ function is maximized using a weighted sum of individual utilities with constant weights ࢇ࢏ǡ࢈࢏ǡࢉ࢏ǡ ܉ܖ܌ࢊ࢏ Parameters are the utility elasticities of stakeholder group i on cultivated land, that add up to one across the population. In other words, everyone in the equilibrium of Negishi format rangeland, forestܽ௜ǡܾ௜ǡܿ land௜ܽ݊݀݀ and௜ other rural land. Interviews with key informants and stakeholder will be at the maximum level of utility subject to their initial endowments (Stanton, 2011). Furthermore, representatives were done to assess their preferences. Interviewees included 15 herders, 15 farmers, 6 Manne and Rutherford (1994) interpreted the equilibrium values of the Negishi weights as the shares in 10 representatives for the economic public authority and 6 for the ecological public authority . From 30 the global endowment of economic resources. Zhu and van Ierland (2006) showed that the Negishi public authorities in charge of local affairs in Tai Pusi County, the Local Finance Bureau and the weight is the income share of each region in the total economy if the Cobb-Douglas utility function is Farming and Grazing bureau were selected as the public authorities for economic affairs, since they pay used. Here, we use the income share of each stakeholder group in the total income as its Negishi weight more attention to local economic development than other departments. Similarly, the Environmental . Specifically, the total income of the farmers/herders is the population of farmers/herders multiplied

ߚby௜ the average net income of farmers/herders in 2012. Similarly, depend on the income of

9 Note that garden plots are included in the category of cultivated land as their area is relatively small in the research the economic public authority and ecological public authority, respectively.ߚଷܽ݊݀ߚସ In terms of the income of region. 10 We believe that a limited number of interviewees per stakeholder group is sufficient to determine the parameter estimates the public authorities, a proxy income is employed. That is, assuming that the local government balances because the variance in interviewee responses was very small (see Appendix 2-C for details). 29 30

because the variancebecause thewas interviewee in responses region. representatives for the economic publicauthority representatives theeconomic for Interviewees included15herders, 6 representatives 15farmers, toassesstheirpreferences. were done stakeholder and with keyInterviews informants ruralland. and other rangeland, forestland 2.5.1 Theparameters ofthe 2.5 Estimation model parameters and 2003. while conversely, offorest thearea with asharpgrowth increasing from landareastarted 1999 in2002 2002and2006.Theareasmooth decrease of cultivated landfrom between decreased 1999 until 2008, rural land are clearly observable. The area of rangeland decreased rapidly during after 2007 and a 2008 oflandusebe transitions slightlydecreasing and dominan on theorderof but hashadlittleeffect rural second positionandother landinthelastposition. The largest ruralalways of proportion been land has 10 9 County from 1995to2009. for cultivated land, Figure the change of 2.3showsthe Parameters more attentiontolocal economic authorities wereas affairs, since bureau the forFarming public theyGrazing selected pay economic and public authoritiesincharge oflocal affairs inTa

We believeWe that a limited number of interviewees per stakehol Note that garden thatin plots Note are included the of category cultiv 100 150 200 250 300 350 400

1995 ܽ 1996

௜

ǡܾ ௜ 1997 Figure 2.3 The dynamic change of rural land use in Tai Pusi County in Figurechangerural land use dynamic of 2.3 The ǡܿ

௜

ܽ݊݀݀ 1998

ࢇ Source: Land Resources Bureau, Tai Pusi County. Bureau,LandSource: TaiPusiCounty. Resources 1999 ࢏ ௜ ǡ࢈ are theutilitygroup elasticitiesofstakeholder ioncultivatedland,

࢏ ǡࢉ 2000 ࢏

܉ܖ܌ࢊ ǡ

development thanother departme development 2001 2002

࢏

very small (see Appendix 2-C for details). for details). (seeAppendixsmall very 2-C 2003

tween cultivated land, rangeland, forest land and other and forest other cultivated land land,tween rangeland, and 6 for the ecological public authority and 6fortheecological 2004 29 29 ated land thesmall astheir in ated research areaisrelatively i Pusi County,Local the Finance Bureau and the

der group is sufficient to determine the parameter estimates represented by rangeland, withrangeland, in cultivatedrepresented land by estland, rangeland and other rural land in Tai Pusi Pusi otherrurallandinTai and estland, rangeland

t land use type. The total area of rural land is rural land The totalareaof t landusetype. 2005

In general, rural land use change has occurred, hasoccurred, In usechange general, land rural

2006

2007

nts. Similarly, theEnvironmental 2008 2009 Total Other rural land Rangeland Forestland Cultivated land 9

10 . From 30

ߚ used. Here,stakeholwe each share use the income of weight region istheincomeshare ofeach intheto Zhu resources. economic of endowment the global Manne weights and equilibriumvaluesoftheNegishi Rutherfordin interpreted (1994) astheshares the will beatthemaximum level ofutility subjectto that addupto oneacrossthepopulation.Inword other a weighted using is maximized function social welfare ߚ th group has stakeholder each that assume we approach 2006).Following Nordhaus, this and (Yang social welfare istopicktheutilitarian weights weights ܽ (Vai and optimisationallocation, conflict resolution large numberofstudiesindifferent t most effective one ofthe (1980), is Thomas Saaty Scenario 1: Scenario the public authorities, a proxy income is employed. Th is employed. income aproxy the publicauthorities, publicau the economicpublicauthorityand ecological 2.5.2 Thewelfareweight ሺࢼ ܽ of Hierarchy Process(AHP)isemployedtheinterviews, theAnalytic weightevaluate values to Protection Agencyand theForestryBureau theecological represent public authorities.Based onthe by the average net income of farmers/herders in 2012.Similarly, in offarmers/herders netincome by average the The social welfare weight The socialwelfare Scenario 2: Scenario authority authority Economic public public authority Ecological Farmers Herders ௜ ௜ ௜ ଵ . Specifically, the total income of the farmers/herders is the population of farmers/herders multiplied farmers/herders populationof isthe farmers/herders of the thetotalincome . Specifically, ǡܾ ǡܾ ൌߚ ௜ ௜ ǡܿ ǡܿ ௜ ௜ ݀ ݀ ଶ ൌߚ

ߚ ߚ ௜ ௜ ଷ is considered as the Negishi weight, introduced by Negishi (1972). In the thisscenario, (1972). introduced by weight, Negishi is consideredastheNegishi is consideredequalfor allfour stakehol ௜ ௜ that we derived using AHP are shown in Table 2.1. 2.1. areshowninTable thatwe AHP derivedusing by introduced AHP process).AHP, stepsinthe different forthe 2-C Appendix (see ൌߚ 0.4304 0.2121 0.4317 0.2084 Cultivated land ସ ͳȀͶ ൌ ߚ Table parameters 2.1 Values of ௜ . will be calculated for three scenarios. threescenarios. willbecalculatedfor This scenario is considered thescenar This scenarioisconsidered ࢏ ሻ

fields have been publishedbasedon fields havebeen 0.3134 0.3173 0.2695 0.5286 Rangeland their initial endowments (Stanton, 2011). Furthermore, (Stanton, 2011). Furthermore, their initial endowments 30 30 tal economy if the Cobb-Douglas utility is function tal economy iftheCobb-Douglas

ools for dealing with complex decision-making. A decision-making. ools for dealing with complex dya and Kumar, 2006). Th dya andKumar,2006). der group in the total income as its Negishider group in the total weight and van Ierland (2006) showed that the Negishi theNegishi showedthat (2006) Ierland and van at is, assuming that the local government balances government balances local assuming thatthe at is, sum of individual utilities with constant weights with constant weights sum of individual utilities s, everyone in the equilibriumofNegishis, everyone inthe format e same weight in the social welfare function, i.e. function, social welfare inthe weight e same der groups.prevailing practice The inchoosing thority, respectively. In terms of the incomeofIn thority, respectively. termsofthe 0.2202 0.1606 Forest land 0.1154 0.3746 ܽ ௜ ǡܾ ௜ ǡܿ ߚ io of social fairness in land allocation. allocation. inland fairness social io of ௜ ܽ݊݀݀ ଷ ܽ݊݀ߚ AHP, suchasplanning, resource 0.0786 0.1024 Other rural land 0.1409 0.0960 ௜

ସ depend on the income of on the income depend e values of parameters e valuesofparameters 1 1 Total 1 1

2 its real income and expenditure, the eco-environmental expenditures on rural land are equivalent to the aspirations of stakeholders. In the following section, we explore the equilibrium in land allocation based income of the ecological public authority from the rural land, and the economic expenditures on local on social welfare optimisation under three scenarios. rural residents are equivalent to the income of the economic public authority from the rural land. 2.6.2 The optimal social allocation Scenario 3: is considered as the working population share of stakeholder group i, which is to According to the three scenarios on welfare weight , we identified the optimal rural land allocation in investigate theߚ௜ optimal land allocation from the perspective of the employment structure in the society. eco-fragile regions from the social welfare perspective.ߚ௜ The results are shown in Table 2.4. Values of in the three scenarios are shown in Table 2.2 (see Appendix 2-D for more details). Table 2.4 Socially optimal land allocation under three scenarios and actual allocation in 2009 (103 ߚ௜ Table 2.2 Values of in three scenarios hectares)

௜ Scenario 1 ߚScenario 2 Scenario 3 Land allocation Cultivated Rangeland Forest land Other rural land Total ( ) ( ) ( ) land Herders 0.25 0.01 0.03 Scenario 1 103.3 115.1 70.2 33.7 322.3 Ⱦ୧ Ⱦ୧ Ⱦ୧ ( ) Farmers 0.25 0.53 0.86 Scenario 2 135.4 94.3 58.5 34.1 322.3 ୧ Ecological public authority 0.25 0.04 0.00 (Ⱦ ) Economic public authority 0.25 0.42 0.10 Scenario 3 136.7 90.9 67.1 27.6 322.3 ୧ Total 1 1 1 (Ⱦ ) Optimal range 103.3--136.7 90.9--115.1 58.5--70.2 27.6--34.1 ୧ 2.6 Results ActualȾ allocation in 2009 94.7 158.1 62.4 7.1 322.3

2.6.1 Land use conflicts among stakeholder groups Considering the welfare weight as the income share of each stakeholder group (scenario 2), cultivated land has much more weight than the other three types of land, a result which is similar to the one in We find the local optima of each stakeholder group on land allocation according to the individual scenario 3. However, the demand for rangeland is higher than for cultivated land if we give equal weight preferences and stakeholder utility functions. The actual total area of rural land in 2009 was used as the to every stakeholder group (scenario 1), and also the demand for forest land is larger than in the other maximum rural land availability in the study region (322,261 hectares). scenarios. The optimal range is based on the minimum and maximum in the scenarios as the lower and Table 2.3 Optimal rural land allocation according to each stakeholder group and actual allocation in upper bound, in which the highest demand by the society is on the cultivated land and rangeland, 2009 (103 hectares) followed by forest land, and finally other rural land. Comparing the actual allocation in 2009 with the optimal range, all types of rural land are out of the optimal range except the forest land. In addition, the Cultivated land Rangeland Forest land Other rural land Total mean of the optimal range is highest for cultivated land while the actual allocation tends most towards Herders 67.2 170.3 51.8 33 322.3 rangeland. Farmers 139.1 86.8 71.1 25.3 322.3 Ecological public authority 68.4 102.3 120.7 30.9 322.3 2.6.3 Comparison of individual optimum, social optimum and actual land allocation Economic public authority 138.7 101 37.2 45.4 322.3 Actual allocation in 2009 94.7 158.1 62.4 7.1 322.3 Table 2.5 presents an overview of the above results. It shows the individual optimum for the four

The results in Table 2.3 present the conflicts among stakeholder groups on rural land use. The herders stakeholder groups, the social optimum and the actual situation in rural land allocation, all as the share have a strong preference for rangeland, while farmers have more of an interest in cultivated land and a of each rural land type in total rural land. Comparing the results within each type of land, we see that lesser interest in rangeland. The ecological public authority pays more attention to the rangeland and the individual optimal shares of cultivated land are outside of the socially optimal range. Herders and forest land as they have more of an ecological function. Cultivated land and rangeland are also favoured the ecological public authority wish to allocate less land for cultivation, while farmers and the economic by the economic public authority. Compared to the actual allocation in 2009 we observe large public authority want to allocate more land for cultivation. Preferences for rangeland are largely within differences to the local optima. This indicates the differences between the actual allocation and the the socially optimal range except for farmers who would like to see less rangeland and herders who 31 32

Values of investigate allocationfrom theperspect theoptimalland ive structure of the employment in the society. differences tothelocaldifferences optima.Thisindicates th by authority. the economic public Compared to the actual allocation in 2009 we observe large func more ofanecological forest landasthey have au public The ecological in rangeland. lesser interest ha for while farmers rangeland, have preference a strong The resultspresent inconflicts Table 2.3 amon the (322,261hectares). rurallandavailabilitymaximum region inthestudy preferences utility and stakeholder The act functions. individual to the according group on land allocation stakeholder each of optima find the We local stakeholder 2.6.1 Landuseconflictsamong groups 2.6 Results 3: Scenario to theequivalent incomerural of th are residents th from public authority ecological income of the its real incomeand theeco-environmental expenditure, Total Economic publicauthority Ecological publicauthority Farmers Herders Actual allocation in 2009 Actual allocationin2009 Economic publicauthority Ecological publicauthority Farmers Herders Table 2.3to Optimal rural allocation according land ߚ ௜ in the three scenarios are showninTab are inthethreescenarios ߚ ௜ isto which group i, share stakeholder workingis considered of population asthe

0.25 ( 1 Scenario 1 0.25 0.25 0.25 Ⱦ Table 2.2 Values of Table 2.2Valuesof ୧ 94.7 138.7 68.4 139.1 67.2 Cultivated land 2009 (10 ) e economic public authority from the rural land. economic public the rural authoritye from land. e rural land, and the economic expenditures on local onlocal expenditures economic and the e ruralland, 158.1 101 102.3 86.8 170.3 Rangeland e differences between the actual allocation and the and actual allocation the between e differences tion. Cultivatedlandand rangeland alsofavoured are 31 31 3 g stakeholder groups on rural land use. The herders Theherders rural landuse. groups on g stakeholder

hectares) hectares) ߚ le 2.2 (see Appendix 2-D for more details). for more details). 2-D le 2.2 (see Appendix ual totalarea ofrural landin2009was usedasthe ௜ 1 0.42 0.04 0.53 0.01 ( 2 Scenario thority pays more attention to the rangeland and totherangeland pays attention thority more in three scenarios in three scenarios each stakeholder groupallocationeach and in actual stakeholder Ⱦ expenditures onruralexpenditures tothe landare equivalent ୧ ve more of an interest in cultivated land and a of an inand interest ve cultivated more land 62.4 37.2 120.7 71.1 51.8 Forest land ) 1 0.10 0.00 0.86 0.03 ( 3 Scenario Ⱦ 7.1 45.4 30.9 25.3 33 Other rural land ୧ 322.3 322.3 322.3 322.3 322.3 Total ) According to the three scenarios on welfare weight onwelfare weight According threescenarios tothe 2.6.2 Theoptimalso on social welfare optimisation scenarios. under three mean of the optimal range is highest for cultivated for is highest mean the optimal range of optimal range, ruralall types landare outofthe of ruralfollowed and by other forestfinally land. land, so upper bound,inwhichthe demandby highest the optimalrange ontheminimum scenarios. The isbased grouptothe every (scenario stakeholder 1), and also high is demand forrangeland the However, scenario 3. land has much more weight than the other three type issimilarto a result which theoneins ofland, as theincomesh weight Considering thewelfare Inaspirations ofstakeholders. thefollowing section, the socially optimal range except for farmers who w who farmers for except the socially optimal range Preferencescultivation. public authorityfor are rangelandto allocate more land want for within largely the ecological wishtoallocate le publicauthority the individualoptimalsharesare cultivated land of rural total in rural landtype of each groups,stakeholder thesocialoptimumandactualsi Itresults. individualoptimumfor thefour above showsthe the of overview an Table 2.5presents 2.6.3 Comparison ofindividual optimum, social optimum and actual land allocation rangeland. eco-fragilethe social welfare regions from pers Actual allocation in 2009 Actual allocationin2009 Optimal range ( 3 Scenario ( 2 Scenario ( 1 Scenario Land allocation Ⱦ Ⱦ Ⱦ Table 2.4 Socially optimal land allocation under three scenarios and actual allocation in2009(10 and actual scenarios landallocationunderthree Table 2.4Socially optimal ୧ ୧ ୧ ) ) cial allocation cial allocation ) 94.7 103.3--136.7 136.7 135.4 103.3 land Cultivated land. Comparing the results within each type of land, we see that that wesee land, type each of within land.Comparing theresults hectares) hectares) pective. The results are shown in Table 2.4. inTable2.4. arepective. Theresults shown 158.1 90.9--115.1 90.9 94.3 115.1 Rangeland ss land for cultivation, while farmers and the economic economic land for and the cultivation,whilefarmers ss are of each stakeholder group (scenario 2), cultivated group (scenario2),cultivated ofeachstakeholder are 32 32 optimal range except the forest land. In land.optimal range the forest addition, theexcept land while the actual allocation tends most towards allocation tendsmosttowards actual the land while ߚ

outside of the socially optimal range.and Herders outside ofthesocially we explore theequilibriumwe explore in land allocationbased ௜ Comparing the actual allocation in 2009 with the allocationin2009withthe Comparing actual the , we identified the optimal rural land allocation in allocation optimalrural, we land identifiedthe demand for forest land is larger than in the other is largerforestdemand for thanintheother land ould like to see less rangeland and herders whorangeland and herders less ould liketosee er er than for cultivatedland we giveif equal weight and maximum andmaximum in the scenariosasthelowerand ciety isonthecultivated landandrangeland, tuation in rural landallocation, allastheshare 70.2 Forest land 62.4 58.5--70.2 67.1 58.5 33.7 Other rural land 7.1 27.6--34.1 27.6 34.1 322.3 Total 322.3 322.3 322.3 3

2 prefer a much larger share of rangeland than what is socially optimal. Unsurprisingly, the ecological context, our results could also be driven by the overall pressure on rural land, and cultivated land in public authority would want to allocate much more land for forestry. The actual land allocation in the particular, in China. This pressure is likely to be especially strong in the eco-fragile regions where most research region is outside of the socially optimal range for each of the land types except for forest land. of the local residents depend on farming, but where agricultural productivity falls far behind the national Land allocation based on social optimisation would increase the area of cultivated land and decrease average level. This promotes the demand for more cultivated land and land for agricultural infrastructure. the area of rangeland. Furthermore, our results may also reflect the misalignment between preferences of the national Table 2.5 Shares of rural land allocated to cultivated land, rangeland, forest land and other rural land government – as included in the national policy programs that focus on environmental protection – and in different situations preferences of local governments – whose performance evaluation is based primarily on economic growth indicators (Liu and Diamond, 2005). Moreover, it is worth mentioning that a decrease in the Cultivated land Rangeland Forest land Other rural Total land area of grasslands in favour of cultivated land does not necessarily conflict with environmental Herders optimal 20.9% 52.8% 16.1% 10.2% 100% objectives. First, research on the effectiveness of current grassland protection policies has not yet allocation provided conclusive evidence (Zheng et al., 2006; Cao et al., 2013). Second, a conversion to more Farmers optimal 43.2% 26.9% 22.0% 7.8% 100% allocation cultivated land is likely to lead to a more intensified use of the rural land. As a result, this may reduce Ecological public 21.2% 31.7% 37.4% 9.6% 100% the pressure on the grasslands as rural residents can sustain their livelihoods through cultivation rather authority optimal allocation than by increasing the number of animals on the grasslands. Economic public 43.0% 31.3% 11.5% 14.1% 100% authority optimal Finally, we would like to point out a number of limitations of our research. While the research method allocation The socially optimal 32.1%--42.4% 28.2%--35.7% 18.2%--21.8% 8.6%--10.6% is generic and likely to be valid for other regions facing land use conflicts, the specific outcomes range presented in this paper are based on stakeholder interviews in one Chinese county and are therefore not Actual share in 2009 29.4% 49.1% 19.4% 2.2% 100% generalizable. Furthermore, the application of the model includes only 4 different categories of land and 2.7 Conclusion 4 stakeholder groups which is a simplification of reality. Finally, the analysis only assesses the This paper takes a welfare economics perspective to investigate current land use conflicts and socially distribution of land resources and hence abstracts from the allocation of other relevant resources of rural optimal land allocation. Optimisation based on individual utility functions and AHP reveals land use households. conflicts among stakeholders due to heterogeneous preferences for different types of land. By optimising the social welfare function, using different welfare weights, we derive the socially optimal range of land allocation. Comparing our empirical results with current land allocation over different land types, we find that a socially optimal land allocation would require a shift towards more cultivated land and less rangeland.

These results should be put in the Chinese policy context. Since the 1990s, the Chinese government has shown increased interest in extending the area of forest land and rangeland for ecosystem protection purposes. Programs such as the SLCP that aim to convert sloping cropland into forestland or grassland are witness to this policy direction (Liu et al., 2010). However, our results show that the social optimum land allocation requires conversion in the other direction: from grasslands towards cultivated land, while the area of forest land is already within the socially optimal range. The preference for cultivated land over rangeland may be driven by the policy restrictions on rangeland. Farmers can use cultivated land more freely than rangeland. While this insight might seem controversial in view of the current policy 33 34

more freely than rangeland. While this While more freely insightrangeland. migh than over may rangeland by be driven the policy restri the area offorest landisalready withinthesoc otherdirect inthe requires conversion land allocation are(Liu witness tothispolicy etal., direction c thataimto as theSLCP purposes. Programs such shown increased interest in extending the area intheChinesepolicyThese results shouldbeput Sincecontext. the 1990s,theChinesegovernment has land and less rangeland. land types, find we that asocially optimal land would require allocation a towards shift cultivated more range of land allocation. Comparingempirical re our range allocation. ofland optimal derive thesocially optimising function, thesocialwelfare weights, different welfare using we heterogeneous dueto stakeholders conflicts among optimal landallocation. Optimisation basedonindivi to economicsperspective a welfare This papertakes 2.7 Conclusion the area ofrangeland. socialoptimisationwouldLandallocation basedon research isoutsideofthesocially region optimalra public much more authoritywould to allocate want ecological the issocially thanwhat optimal.Unsurprisingly, rangeland largershare of prefer amuch allocation allocation Farmers optimal allocation Herders optimal Actual share in2009 Actual share range optimal The socially allocation authority optimal Economic public allocation authority optimal Ecological public Table 2.5 Shares of rural landallocatedtocultivat Table 2.5Sharesof 29.4% 32.1%--42.4% 43.0% 21.2% 43.2% 20.9% Cultivated land in different situations 49.1% 28.2%--35.7% 31.3% 31.7% 26.9% 52.8% Rangeland 2010). However, our results show thatthesocialoptimum ourresults However, show 2010). iallyoptimal range. cultivatedland for Thepreference of forest land and rangeland for ecosystem protection forest forof landandrangeland ecosystem ctions on rangeland.Farmers canctions use cultivated land 33 33 ngefor for types each forestexcept of the land. land onvert sloping cropland into forestland or grassland forestlandor into onvert slopingcropland ed land, rangeland, forest land and other rural land landand land otherrural forest ed land,rangeland,

ion: from grasslands towards cultivated land, while while cultivated land, towards grasslands ion: from land for forestry. Theactual land allocation in the t seem controversial in view of thecurrentpolicy in t seemcontroversial investigate current land use conflicts and socially use conflicts and socially investigate land current increase theareacultivated landanddecrease of dual utility and AHP revealsfunctions land use sults with current land sults withcurrent land preferences for different types of land. By land. types different of for preferences 19.4% 18.2%--21.8% 11.5% 37.4% 22.0% 16.1% Forest land allocation over different 2.2% 8.6%--10.6% 14.1% 9.6% 7.8% 10.2% land Other rural 100% 100% 100% 100% 100% Total context, our resultscontext, could also be driven by the government Furthermore, resultsmayalso our reflect the cult more for the demand Thisaverage promotes level. ag onfarming,butwhere residentsof thelocal depend particular, inChina.Thispressure is households. from abstracts and hence distribution oflandresources asimplificationofre groups is which 4 stakeholder and generalizable. ofland ofthemodelincludesonlyFurthermore,categories theapplication 4different in basedonstakeholder are presented inthispaper be va to likely is genericand Finally, wouldliketopointoutanumber we of than bygrasslands. of animals increasing thenumber onthe the pressure on the grasslands as rural residents ca lead to to cultivated landislikely al., 2006; et (Zheng evidence provided conclusive effectiveness onthe research objectives. First, withenvironmental conflict does not necessarily ofcultivatedland grasslands infavour area of Diamond,2005).Moreover,(Liuand it growth indicators governments preferences local of as included in the national policy– as included p lid for other regions facing land use conflicts, the specific outcomes outcomes specific the conflicts, landuse facing regions for other lid a more intensified use of the rural land. As a result, thismay As aresult, reduce therural land. useof moreintensified a whose performance evaluation is based primarily on economic economic on primarily isbased evaluation whose performance – likely to be especially strong in the eco-fragile regions where most most where regions eco-fragile inthe strong especially likely tobe limitations ofourresearch. Whiletheresearch method overall land, and cultivated land in on rural pressure of current grassland protection policies has not yet not policieshas grassland protection current of rograms protection on environmental that focus misalignment between preferences of the national national preferencesmisalignment of the between terviews in one Chinese county and are therefore not arenot terviews therefore in one Chinese county and 34 34 n sustain their ratherlivelihoods through cultivation

Cao et al., 2013). Second, a conversion to more conversion to a Second, 2013). al., Caoet ricultural productivity falls far behind the national farbehindthe falls productivityricultural ivated land and land for agricultural infrastructure. infrastructure. agricultural for land ivated landand the allocation of other relevant resources of rural resources relevant of theallocationofother ality. Finally, the analysis only assesses the analysis Finally,ality. the assesses only isworthmentioning inthe thatadecrease – and

2 ACKNOWLEDGEMENTS:

Funding for this research was supported by the Chinese academy of engineering with the project: Ecological security and food security in the grasslands of China (2012-ZD-7). The authors would like to thank Xiangzheng Deng, Wenhui Kuang and Zeng Tang for their help and suggestions during the Chapter 3 field work in China.

Land tenure reform and grassland degradation in Inner Mongolia, China

35 36

ACKNOWLEDGEMENTS:

Land tenure reform and grassland degradation in Inner Mongolia, China

35 36

3. LAND TENURE REFORM AND GRASSLAND DEGRADATION IN INNER MONGOLIA, CHINA11 3.1 Introduction

The tragedy of the commons, a concept introduced by Hardin (1968), has long been part of the ABSTRACT: Since the start of the land tenure reform in the pastoral areas of China in the 1980s, conventional wisdom in ecology, economics and political science (McEvoy, 1987). Theoretical research grassland use rights have increasingly been assigned to individual households and subsequently more contended that resource degradation was inevitable unless the commons were either privatised or grasslands have been in private use. However, in the same period, most of the grasslands in China have maintained as common land with clear rights of entry and use (Hardin, 1968). Similarly, in the theorem experienced degradation. The question that this paper tries to address is whether the land tenure reform of property rights (Coase, 1960), a clear assignment of property rights is proposed as a precondition for plays a significant role in grassland degradation. It is answered by an empirical analysis of the impact economically efficient resource allocation and environmental sustainability. As a consequence, the of land tenure reform on the changes in grassland condition, using data from 60 counties in Inner privatisation of property rights or use rights over resources has been implemented widely by Mongolia between 1985 and 2008. Grassland condition is presented by grassland quantity and quality governments around the world (Little, 1992; Lesorogol, 2003; Squires et al., 2009; Ybarra, 2009). The using spatial information based on remote sensing. The timing of the assignment of grassland use rights same privatisation trend has been observed in Inner Asia and Africa regarding the reform of grassland and the timing of the actual adoption of private use by households differ among counties. These timing tenure (Sneath, 1998; Lesorogol, 2008). differences and differences in grassland condition among counties allow disentangling the impact of the land tenure reform. A fixed effects model is used to control for climate, agricultural activity and the However, Hardin’s theory of the tragedy of the commons and Coase’s theorem of property rights have time-invariant heterogeneity among counties. The model results show that the private use of grasslands been rejected by pastoral specialists, who found that these theoretical findings provided a poor guide to following the land tenure reform has had significantly negative effects on grassland quality and quantity understanding pastoralism (Feeny et al., 1990; Behnke, 1994; Sneath, 1998). They concluded that in Inner Mongolia. Moreover, the negative effects did not disappear even after several years of exclusionary land tenure is counterproductive to protect the grassland condition in arid and semi-arid experience with private use. In conclusion, our analysis reveals that the land tenure reform, namely areas (Turner, 1999; Galvin, 2009). And some scholars (e.g. Ostrom, 1990; Li and Zhang, 2009) found privatisation of grassland use rights, is a significant driver of grassland degradation in Inner Mongolia that traditional common use systems of grasslands often met the criteria for sustainable Social- in a long term, which presents “a tragedy of privatisation”, as opposed to the well-known “tragedy of Ecological Systems (Ostrom, 2009), namely the exclusion of outsiders and the self-organization of the commons”. resource users (Ostrom, 1990). Therefore, doubts exist about the likelihood that traditional common use systems would induce a tragedy of the commons. Moreover, a growing number of academic studies KEYWORDS: privatisation, formal use rights, private use, grassland degradation, fixed effects model question the effects of the prevailing privatisation of grassland property rights or use rights on the grassland condition. For instance, some studies assert that grassland privatisation puts an end to mobile pastoralism by introducing inflexible boundaries and identify this as a main driver for grassland degradation (e.g. Fernandez-Gimenez and Le Febre, 2006; Li and Zhang, 2009). Mobile pastoralism is embraced because mobility maintains ecosystem stability and biological diversity and conforms with the sustainable use of natural resources that are subject to strong spatial and temporal heterogeneity, as is the case in arid and semi-arid areas (Fernandez-Gimenez, 2002; Fernandez-Gimenez and Le Febre, 2006; Tessema et al., 2014; Fan et al., 2015). Empirical evidence was provided by Sneath (1998) who used satellite images to show that grassland degradation was much less severe in Mongolia – which kept the common use system for grasslands – than in the Russian and Chinese parts of Inner Asia where the privatisation of grassland property rights or use rights was implemented. Furthermore, Guelke (2003) showed that grassland degradation in South Africa was primarily due to the introduction of privatisation.

In China, land tenure reform started with the Household Production Responsibility System in the 11 Paper by Min Liu, Liesbeth Dries, Wim Heijman, Jikun Huang, Xueqin Zhu, and Xiangzheng Deng, submitted to a cropland areas in the late 1970s and spread to the pastoral areas by the early 1980s. The land tenure peer-reviewed journal. 37 38

11 DE 3. LANDTENUREREFORMANDGRASSLAND differences and differences in grassland condition in anddifferences differences us private adoptionof actual timing the and the of Th remote sensing. based on information using spatial inprivateus grasslands havebeen grassland use rights assign increasingly been have 1980s, Chinainthe of areas pastoral reform inthe landtenure start ofthe ABSTRACT: Since the following the land tenure reformfollowing hadsignificantly thelandtenure has time-invariant heterogeneity counties. The among modelresults showthatthegrasslands private useof effectsland tenure Afixed modelisusedto reform. and 2008.Grassl 1985 Mongolia between grassland in thechanges reform on of landtenure It grassland degradation. role in plays a significant tr Thequestionthatthispaper degradation. experienced KEYWORDS: privatisation, formal use rights, private KEYWORDS: privatisation, private formaluserights, the commons”. the well asopposedto tragedyof privatisation”, “a presents in along which term, Inner Mongolia in grasslandprivatisationgrassland use degradation rights,t driverof of is a significan In with private use. experience conclusion, our anal Innereffect in thenegative Mongolia. Moreover, peer-reviewed journal peer-reviewed Paper by Min Liu, Liesbeth Dries, Wim Heijman, Jikun Huang, Xueqin Zhu, and Xiangzheng Deng, submitted to submitted Zhu, Deng, Jikun Huang, Xiangzheng and Xueqin by Wim Paper Heijman, Liesbeth Liu, Dries, Min

. e. However, in the same period, most in thee. However, same period, most and condition is presentedis by condition and e by households differ among counties. These timing timing counties. These among byhouseholds differ e among counties allow disentangling the impact of the ofthe theimpact allow disentangling among counties 37 37 ed toindividualhouseholds andsubsequently more

is answered by an empirical analysis oftheimpact analysis an empirical is answered by s did not disappear even after several yearsof afterseveral s didnotdisappeareven condition, using data from condition, usingdata controlfor climate, agriculturaland the activity ysis reveals thattheland tenurereform, namely GRADATION IN INNER MONGOLIA, CHINA MONGOLIA, GRADATION ININNER e timing of the assignment of grassland use rights rights grassland use of assignment e timing the of negative on grassl effects use, grassland degradation, fixed effects model grassland effects fixed use, degradation, model ies to address is whether the land tenure reform land tenure the is whether address ies to of the grasslands inChinahave grassland quantitygrassland andquality and quality and quantity and quality andquantity 60 counties in Inner Inner counties in 60 - known “tragedy of “tragedy known a 11

conventional wisdom in ecology, economics and poli and economics ecology, conventional wisdomin introduced ofhas athe concept long part The tragedy commons, by(1968), Hardin been of the 3.1 Introduction used satellite images to show that grassland de show that used satelliteimages to Empi Fanetal.,2015). et al.,2014; 2006; Tessema LeFebre, and 2002;Fernandez-Gimenez (Fernandez-Gimenez, areas arid andsemi-arid is thecasein the sustainableare use that of subject natural resources stability mobilityembraced because maintainsecosystem Zhang, is Mobile pastoralism 2009). Li and Febre,Le 2006; and Fernandez-Gimenez (e.g. degradation pastoralism by introducing boundaries inflexible a so grassland condition.For instance, question theeffects oftheprevailing privatisation systems wouldtragedy induce of the commons. a M resource users(Ostrom, 1990).Therefore, doubts exist excl the 2009),namely (Ostrom, Systems Ecological that traditionalcommon usesystemsgrasslands of areas 2009).Andsomescholar (Turner, 1999;Galvin, protec to iscounterproductive tenure exclusionary land understanding(Feeny et al., 1990; pastoralism specialists,whofoundth pastoral by been rejected Hardin’sHowever, theoryof thetragedy of Lesorogol, 1998; tenure (Sneath, 2008). samehas privatisation been obs trend world (Little,governmentsLeso aroundthe 1992; ove rights use or rightsprivatisation ofproperty e allocationand resource efficient economically of propertyrights 1960), a clear (Coase, assignment Similarly, thetheorem (Hardin, 1968). in use and entry of rights withclear commonland maintained as degradationcontended inevitable was thatresource the privatisation of grassland property rights or use or use rightsgrassland property the privatisationof grasslands for system kept thecommonuse cropland areas in the late 1970s and spreadtoth 1970s and areas inthelate cropland In startedwiththe reform China,landtenure wa inSouth Africa grassland degradation showed that me studies assert that grassland pri erved in Inner Asia and Africa regarding the reform of grassland grassland thereform of regarding and Africa Inner Asia erved in than in the Russian and Chinese parts of Innerwhere Asia partsof intheRussianandChinese – than the commons and Coase’sthe theorem commons of property r gradation was much less severe inMongolia gradation muchlesssevere was Household ProductionHousehold Responsibility System inthe Behnke, 1994; Sneath, 1998). They concluded that 1998). TheyBehnke, concluded 1994;Sneath, e pastoral areas by the early 1980s. The land tenure 1980s. The landtenure early the by areas e pastoral nvironmental sustainability. As a consequence, the the nvironmental sustainability. consequence, Asa rogol, 2003; Squires et al., 2009; Ybarra, rogol, 2009).The 2003;Squiresetal., at thesetheoreticalguideto findingsapoor provided 38 38 rights was implemented. Furthermore, Guelke (2003) (2003) Guelke Furthermore, implemented. rights was

rical evidence was by who provided (1998) rical evidence Sneath of property rights isproposedasaprecondition for tical science (McEvoy, 1987). Theoretical tical research science (McEvoy, of grassland property rights oruserights on the nd identify this as a main driver for grassland s primarily due to the introduction of privatisation. ofprivatisation. totheintroduction due s primarily r resources has been implemented widely by implementedwidely hasbeen r resources oreover, a growing number of academic studies growing ofacademic oreover, number a often met the criteria for sustainable Social- s (e.g. Ostrom, 1990; Lis (e.g. Zhang, and 2009) found unless the commons were either privatised or wereeither privatised commons unless the about the about likelihood that traditionalcommonuse usion of outsiders and the self-organization of usion ofoutsidersandtheself-organization to strong spatial and temporal heterogeneity, as temporalheterogeneity, and to strong spatial t the grassland condition in arid and semi-arid inaridand semi-arid grassland condition t the and biological diversity and conforms with with and conforms andbiological diversity vatisation putsanendtomobile ights have ights have which – which

3 reform in the pastoral areas aims at privatising use rights of grasslands to individual households and 3.2 Materials displacing traditional common use (Banks et al., 2003). Some differences have been observed in the 3.2.1 Study area tenure reform of grasslands among the provinces. For instance, the grasslands tenure reform started in To explore the effect of land tenure reform on the grassland condition in the pastoral areas of China, we 1984 in Inner Mongolia and Heilongjiang while it had started already in 1983 in Ningxia (Ho, 2000). selected Inner Mongolia, a province located in northern China, to conduct our empirical study. Inner Furthermore, the privatisation of grassland use rights in some provinces was ahead of the national Mongolia belongs to the arid and semi-arid areas of China that is vulnerable to degradation, decision about the tenure reform of grasslands, as the first Grassland Law of China was introduced only desertification and salinization (Feng et al., 2009). It contains 21.7% of the area of China’s natural in 1985. Moreover, it is reported that most of the grasslands in China have experienced various degrees grasslands. Approximately 67% of the total land in Inner Mongolia is classified as grassland, the of degradation since the 1980s (Li et al., 2014). The manifestation of the grassland degradation involves majority of which is temperate grassland (Angerer et al., 2008). As a traditionally pastoral region a reduction in the extent of grass cover, a reduction in the density of grass cover, a reduction in the populated by Mongolians, the vast majority of local people maintain their livelihoods upon the output of forage material, an increase in unpalatable grass species, an increase in soil compaction, permanent grasslands and the region has a relatively low population density in spite of rapid economic changes in plant functional group or a combination of all of them (Feng et al., 2009; Li et al., 2013; Lin growth and a soaring population growth elsewhere in China (Angerer et al., 2008). The grassland et al., 2015). According to Chinese governmental reports, 10% of the total area of grasslands was condition in Inner Mongolia has been a major concern because the region is not only one of the main degraded in the 1970s, increasing to 30% in the 1980s and 50% in the middle of the 1990s (Meng and production regions for animal products of China (Li et al., 2016), it also functions as an ecological Gao, 2002). By the 2000s, about 90% of the grasslands were degraded to various extents, with barrier for northern China. For instance, the colossal dust storms which rumbled through hundreds of significant regional variation (Unkovich and Nan, 2008; Waldron et al., 2010). It is widely agreed that cities and villages of northern China and blanketed the sky of Beijing between 1998 and 2001, are said grassland degradation is one of the most important environmental issues confronting China presently to have originated from dryland areas and degraded grasslands mainly in Inner Mongolia (Wu et al., (Han et al., 2008; Feng et al., 2009). As efficient policy interventions derive from correct understandings 2015). In terms of its status as main region for livestock production in China, the mutton output of Inner about the causes of grassland degradation (Brown et al., 2008), and good management practices are Mongolia accounted for 22% of the total output of China in 2013, while 8% of all Chinese beef was capable of improving the grassland condition significantly (Parras-Alcántara et al. 2015), a relevant produced here (China Animal Industry Yearbook, 2014). Therefore, continued grassland degradation in question is what role the land tenure reform has played in the process of grassland degradation. Inner Mongolia is supposed to harm economic development as well as cause ecosystem instability in Although some studies exist that discuss the impact of the land tenure reform on the grassland condition China (Meyer, 2006). By the end of the twentieth century, 90% of the grassland condition of Inner in China (e.g. Ho, 2001; Li and Zhang, 2009), there is a lack of empirical analysis based on long-term Mongolia had been degraded to various extents (Mei et al., 2013), which has stimulated growing data on a large-scale (Lu et al., 2015). Therefore, this paper uses information on the changes in grassland attention to the protection of grasslands in recent decades. condition and progress of the land tenure reform in a typical pastoral area in China over a relatively Inner Mongolia was the first region for implementation of the land tenure reform in the grassland area long period (1985-2008) to estimate the effect of land tenure reform on grassland condition. In the (Li and Huntsinger, 2011). According to provincial government reports for Inner Mongolia, the following sections, we first provide some background information about the grassland tenure reform in privatisation of grassland use rights began officially through the Double Contracts System in 1982. China, including the progress in the assignment of grassland use rights and the actual use of grasslands. Because of its limited uptake, the implementation of grassland privatisation was strengthened by the Next, the materials of our empirical research are described and the fixed effects model in our case is Two Rights and One System policy (Shuang Quan Yi Zhi) in 1996 (Bureau of Animal husbandry of presented. We conclude with some remarks about the effects of the land tenure reform on the grassland Inner Mongolia, 2000; Li and Zhang, 2009). This is regarded as the second round of privatisation of condition in the research region and implications of our findings for future research. grassland use rights. To this date, local governments continue to work on clarifying the boundaries of grassland use rights for each household and on issuing certificates (Bureau of Animal husbandry of Inner Mongolia, 2015). In practice, the specific timing and extent of conducting the land tenure reform differed among counties in Inner Mongolia. This paper is therefore based on county-level data with the

39 40

condition in the research region andimplica condition intheresearch presented. with some remarks conclude We about desc are research empirical of our thematerials Next, gra assignment China, including progress inthe of the provide some backgro following we sections, first long to estimate the effect period (1985-2008) of condition and progress of the land tenure reform in a thisdataet al., on a large-scale 2015). (Lu Therefore, Zhang, is 2009),there in China(e.g.Ho, 2001;Liand Although thatsome studies exist discuss the impact land tenurequestion iswhat pl reform rolethe has thegrasslandcapable ofimproving conditionsignifican degradation grassland (Brow about the causes of (Han et 2008; Fengal., etal., 2009). As poli efficient grassland degradation is one of It that agreed iswidely 2010). etal., 2008;Waldron Nan, regionaland significant (Unkovich variation grasslands ofthe about90% the 2000s, ByGao, 2002). degraded in the 1970s, increasing to 30% in the et al., 2015). According to Chinese governmental re changes in plant functional group them(Feng of all of or a combination etal.,2009;Li et al.,2013;Lin an increaseoutput of forage in material, unpalatable ofgrass intheextent a reduction areduction cover, al., 2014).Th (Li et sincethe1980s of degradation in 1985.Moreover, itis reportedmost ofthegrassl that variousdegrees ands inChinahaveexperienced only introduced Law Chinawas of Grassland first grasslands,reformas the decision tenure about the of ofgrassland ri Furthermore, use theprivatisation Innerand Heilongjiang whileit 1984 in Mongolia tenure reform of grasslands For among the provinces. use displacing traditionalcommon reform in the pastoral us aims at areas privatising the most important environmental issues confronting China presently confronting Chinapresently the mostimportantenvironmentalissues (Banks et al., 2003). Some differences have been observed inthe havebeenobserved 2003).Somedifferences (Banks etal., tions of our findings for future research. forfutureresearch. tions ofourfindings 1980s and 50% in the middle of the 1990s (Meng and and 1980s and50%inthemiddle ofthe1990s(Meng und information about the grasslandreform in tenure und information about the e manifestation of the grassland degradation involves grassland degradation involves the e manifestationof n et al., 2008), and good management practices are good management practices and n etal.,2008), ayed in the process of grassland degradation. ayedgrassland degradation. intheprocessof the effects ofthelandtenurereform onthegrassland 39 39 e rights of grasslands toindividualhouseholds and land tenure reform on grassland condition. In the land grasslandtenure condition. reform on ghts in some provinces was ahead of the national ghts national insomeprovinceswas ofthe ahead of the land tenure reform on the grassland condition had started already in 1983 in Ningxia (Ho, 1983 2000). alreadyhad in started

cy interventions derive from correct understandings understandings cy correct from interventions derive paper uses information on the changes in grassland grassland onthechanges in paperusesinformation ssland use rights and the actual use of grasslands. useof actual andthe rights ssland use in thedensity ofgrass a lack of empirical analysis based on long-term on long-term empiricalanalysis alack based of ports, 10% of the total areagrasslandswas of ribed and the fixed effects model in our case is is case modelinour effects fixed and the ribed instance, the grasslands tenure reform started in reform started tenure grasslands the instance, typical pastoral area inChinaoverarelatively typical pastoralarea grass species, an increase in soil compaction, insoilcompaction, an increase grass species, tly (Parras-Alcántara et al. 2015), a relevant etal.2015),arelevant tly (Parras-Alcántara were degraded to various extents, with extents, to various degraded were cover, areduction inthe differed among counties in Inner Mongolia. This pape counties in differed among timi specific In practice, the Inner Mongolia, 2015). and on is each household grassland for use rights grassland userights. Tothisdate,local governments Liand Zhang, is Inner 2009).This Mongolia,2000; Two Rights andOne System policy husbandry YiZhi)of (Shuang Quan in1996(Bureau ofAnimal Because ofitslimiteduptake, theimplementationof userightsprivatisation ofgrassland began official According toprovincial 2011). and Huntsinger, (Li Inner region Mongoliawas thefirst for implementat decades. inrecent ofgrasslands protection tothe attention extents degraded tovarious Mongolia been had China (Meyer, 2006).By endofthetwentiet the as Inner Mongoliasupposed toharmeconomicdevelopmentaswell is Industry2014). Animal produced here (ChinaYearbook, of totaloutput 22% ofthe for Mongoliaaccounted In2015). terms ofitsstatusasmainregion for livest drylandto haveoriginated fromareas anddegraded Chinaandcities andvillages blanke ofnorthern co For instance,the for northernChina. barrier production regions for productsofChina(Li animal et a majorconce Inner hasbeen condition in Mongolia growth populationgrowth andasoaring inChina(Angerer elsewhere etal.,2008).Thegrassland grasslands theregionpermanent and hasarelative populated by thevastmajority Mongolians, of grassland (Angerer majoritywhich istemperate of in land thetotal of grasslands. 67% Approximately desertification and salinization (Feng et al., 2009 ar Mongolia aridandsemi-arid tothe belongs Innerselected Mongolia, a province located in northe theeffect oflandtenurereformTo explore onthegra 3.2.1 Studyarea 3.2 Materials

lossal duststormswhichrumbledof through hundreds ted the sky of Beijing between 1998 and 2001, are sky said 1998and2001, Beijing the of between ted local peoplemaintaintheirlivelihoodsuponthe h century,Innergrassland conditionof 90%ofthe 40 40 ly lowpopulationdensity inspiteofrapid economic ). It contains 21.7% of the (Mei et al., 2013), which has stimulated growing growing which hasstimulated al., 2013), (Mei et suing certificates (Bureau of Animal husbandry of ofAnimalhusbandry suing (Bureau certificates eas of China that is vulnerable to degradation,

ock productioninChina,Inner themuttonoutputof ly the Double Contracts System through in 1982. ng of conductingreform and the extent land tenure ion of the land tenure reform in the grassland area China in 2013, while 8% of all Chinese beef was all Chinesebeef 8% of while China in2013, et al., 2008). As a traditionallyAs a et region al., 2008). pastoral ssland condition in the past rn because the is region not onlyrn because of the main one continueto of workon theboundaries clarifying regarded roundofprivatisation of asthesecond grasslandsInner mainly Mongolia(Wuetal., in r is therefore based on county-level data with the withthe data county-level on based r istherefore grassland was privatisation strengthened bythe rn China, to conduct our empirical study. Innerrn China,toconductourempiricalstudy. Inner Mongolia is classified as grassland, the classified as Inner is Mongolia government reports for Inner Mongolia, the the Inner forgovernment reports Mongolia, al., 2016), it also functions as an ecological al., 2016),italsofunctions asanecological Therefore, continued grassland in Therefore, degradation cause ecosystem instability in area of China’s natural area ofChina’s natural oral areas of China, we

3 aim of disentangling the effects of privatisation on the grassland condition. Figure 3.1 presents the temporal differences in grassland condition across extensive grasslands during a long period. In this region of Inner Mongolia and marks the counties that are included in our empirical analysis. regard, our information on grasslands and climate conditions relies on a land use database developed by the Chinese Academy of Sciences with original data from Landsat Thematic Mapper / Enhanced Thematic Mapper (Plus) (TM/ETM+) images which have a spatial resolution of 30*30m (Deng et al. 2011). Visual interpretation and digitization of Landsat TM/ETM+ images were conducted to generate land use maps for the years 1985, 1995, 2000, 2005 and 2008. Land cover classification system was applied to distinguish grasslands from cultivated land, forestry area, water area, built-up area, and unused land. According to the grass cover, grasslands were further classified into three subclasses which were adopted from the research by Feng et al. (2009) and Deng et al. (2011), including dense grassland, moderate grassland and sparse grassland. In addition, socio-economic information was derived from statistical data collected by local governments. After dropping the urbanised counties - where changes in the grassland area can be attributed mainly to changes in land use types - and those counties that experienced changes in administrative regions, 60 counties out of 74 counties with valid questionnaires are retained as the research sample for this study, as marked in Figure 3.1.

3.2.3 Land tenure reform for the grasslands 3.2.3.1 A review of the legislation on grassland tenure of China Figure 3.1 Inner Mongolia and the selected counties in this study

In the collectivist period of China, grasslands were assigned to production teams or People’s communes 3.2.2 Data sources and the grasslands were managed and used communally by all households in the same production team County-level data were collected for the period 1985-2008, the main period during which the land tenure (Hua and Squires, 2015; Taylor, 2006). After that period, given the successful experience of privatising reform was implemented across Inner Mongolia. Information about the land tenure reform was collected land use rights to individual households in the cropland regions in the late 1970s, the Chinese central through questionnaires. The questionnaires were sent to the Animal Husbandry Bureau of each county government introduced the household-based privatisation of land use rights into the pastoral areas in and answered by local experts on the grassland tenure reform. In a second phase, interviews were the early 1980s. Since then, the management of grasslands has increasingly become the responsibility conducted by telephone to confirm the answers that were received. A total of 74 valid questionnaires of individual households (Banks et al., 2003). were obtained out of 102 counties of Inner Mongolia. Besides the questionnaires, we had interviews with the officers who are working in the provincial institutes of Animal Husbandry of Inner Mongolia The first Grassland Law of China was initiated in 1985 and stipulated that the property rights over and with local herders to verify the progress of land tenure reform on local grasslands. grasslands are owned either by the state or the collective but that use rights to the grasslands can be contracted out to either households or collectives (Ho, 2000). According to the current Law on the In the existing literature, grassland condition is indicated by grass cover, height, density, biomass Contracting of Rural Land, the grassland use rights can be contracted to individual households for a production or density of perennial vegetation (Gu and Li, 2013; Yu and Farrell, 2013). Commonly used duration of 30 to 50 years. In practice, local governments started to assign grassland use rights to methods to measure these indicators include small-scale sampling tests in field studies and data analysis households through a Double Contracting System (“cao xu shuang cheng bao”) in the 1980s. In 2002, with remote sensing and GIS (Geographic Information System) over large areas. The former focuses on the revised Grassland Law reaffirmed the devolution of land use rights from the state and collectives to measuring the specific indices of grass production and vegetation diversity directly, while the latter individual households (National People's Congress of the People's Republic of China, 2002). The State estimates the general grassland condition based on satellite images (Gu and Li, 2013). In this study, the Council emphasised the specific allocation of the contracted items (e.g. grassland plots, areas, issuing method with remote sensing and GIS is more appropriate because we aim to quantify the spatial and user-right certificates etc.) to households to ensure the long-term stability of contract relations in a

41 42

method with remote sensing and GIS appropr ismore and method withremotesensing estimates thegeneral condi grassland measuring of grass production thespecificindices GISwith remoteInformation and sensing (Geographic analysis in field studies and data sampling tests small-scale indicators include methods to measure these Liperennialproduction ordensityand vegetation(Gu of In literature,grassland height,existing conditionisindicatedby biomassgrass the density, cover, verify tenure theand to progress reform of with land local herders on local grasslands. workingwith the officers who are in the provincia were obtained outof102countiesInner Mongolia that answers confirmthe to telephone conducted by te onthegrassland experts by local and answered sent were The questionnaires through questionnaires. reform was implemented across Inner Mongolia. InInner Mongolia. across implemented reform was period 19 forCounty-levelcollected the were data 3.2.2Data sources regionInner andmarksthecounties Mongolia of aim ofdisentangling theeffects ofprivatisation Figure 3.1 Inner Mongolia and the selected counties inthisstudy counties FigureInner andtheselected Mongolia 3.1 tion based on satellite images (Gu and Li, Intion based on satellite imagesand 2013). (Guthis thestudy, 85-2008, the main period during which the land tenure during thelandtenure themainperiod which 85-2008, that are included in our empirical analysis. inourempiricalanalysis. are included that on the grassland condition. Figure 3.1 presents the presents the on thegrassland Figure condition. 3.1 41 41 formation about the land tenure reform was collected reform was tenure land formation aboutthe l institutes of Animal Husbandry of Innerl institutesofAnimalHusbandry Mongolia of

nure reform. In a second phase, interviews were were phase,interviews Ina second nure reform. were received. A total of 74 valid questionnaires questionnaires of74valid Atotal received. were and vegetationdiversity whilethelatter directly, to the Animal Husbandry Bureau of each county county each of Bureau totheAnimalHusbandry . Besidesthequestionnair iate because we aim to quantify the spatial and and to quantifyspatial aim the we iate because System) over large areas. The former focuses on focuses TheSystem)areas. former overlarge , 2013; Yu and Farrell, 2013). Commonly used es, we hadinterviews

user-right certificates etc.) tohouseholdsensure etc.) certificates user-right Council emphasised the contra specific allocation of the individual households(National People's Congress Lawthe revised reaffirmed Grassland of thedevolution In 2002, the 1980s. in bao”) shuang cheng xu (“cao Contracting System Double households through a duration of30to50years.In local governm practice, individual households for a contracted to can be rights thegrasslandLand, use Contracting Rural of households orcollectiv either contracted outto grasslands are eitherby owned thestateor coll initiatedin was ofChina Law Grassland The first of individualhouseholds(Banksetal.,2003). the early 1980s.Since then,themanagement ofgrassla privatisation government household-based introduced the land userights toindividual householdsinthec Taylor,(Hua 2006) andSquires,2015; managed usedcommuna grasslands and were and the In thecollectivistperiod ofC thelegislation ofChina ongrassland tenure 3.2.3.1 Areviewof forthegrasslands 3.2.3 Landtenurereform Figure 3.1. sampleforas markedin thisstudy, as theresearch are retained administra in changes experienced in the grassland be attributed mainly can area to governments.bystatistical Af datalocal collected moderate grassland and sparse gr were the adopted research from by Feng et al. (2009 grasslands we thegrassunused land.According cover, to applied todistinguish grasslands from cultivate years Landland usemapsfor classificationsystem was 1985,1995, 2000, 2005and2008. the cover Landsat generate of anddigitization 2011). Visualinterpretation TM/ETM+ to imageswereconducted (Plus)(TM/ETM+) ha imagesThematic Mapper which byof Scienceswithorigina theChinese Academy regard, ongrasslandsand climatec ourinformation ex in grassland condition across temporal differences hina, grasslands wereassigned to assland. In socio-economic addition, assland. tive regions, 60 counties out of 74 tive regions,counties outof74 60 . After that period, given the successful experience of privatising privatising of experience given thesuccessful period, that . After es (Ho, 2000). According to the current Law on the Law onthe tothecurrent (Ho, 2000).According es ropland regions in the late 1970s, the Chinese central d land, forestry area, water area, built-up area, and d land,forestry area,built-up water area,

ter dropping the urbanised counties - where changes changes - where counties ter theurbanised dropping 42 42 changes in land use types - andthosecountiesthat changes inlandusetypes ofthePeople's Republic of China,2002).TheState

l data from Landsatl data from Mapper / Enhanced Thematic ) andDeng including etal.(2011), densegrassland, ective but that use rights to the long-term stabilityof thelong-term onditions relies on a land use database developed developed onditions relies onalandusedatabase 1985 and stipulated that the property rights over rights over theproperty stipulated that 1985 and lly by all households in the same production team all householdsinthesameproductionteam lly by tensive grasslandsInthisduring period. along nds has increasingly become theresponsibility become nds hasincreasingly land use rights from thestateandcollectivesto land userights from ve a spatial resolution of 30*30m (Deng et al. resolution of30*30m(Deng et ve aspatial re further classified into three subclasses which re classified further ents startedtoassign grassland userights to cted items (e.g. plots, areas, grassland issuing of land use rights into the pastoral areas in intothepastoralareas oflanduserights productionteams orPeople’scommunes counties with valid questionnaires withvalidquestionnaires counties information was derivedfrom information the grasslands be can contract relations ina contract relations

3 government report in 2011 (Li, 2012). With these legislative efforts, the privatisation of the grasslands Zhang, 2009; Li, 2012) have identified a number of possible barriers to the completion of the land has been conducted gradually in the extensive pastoral areas of China, and increasingly private use by tenure reform in the pastoral areas of China: (1) the inflexible boundaries stemming from the households has replaced traditional open access to land (Hua and Squires, 2015; Yang et al., 2004). By privatisation of grassland use rights is incompatible with the need for livestock mobility due to the 2014, the total area of grasslands that have been assigned to individual households reached 223 million heterogeneity of resources in arid and semi-arid areas (Li and Zhang, 2009); (2) there is a lack of hectares, out of 400 million hectares of grasslands in China (Ministry of Agriculture of China, 2015). financial resources for protecting private property rights, e.g. to build fences (Ho, 1996; Li et al., 2007); (3) individual household ownership is inconsistent with local or traditional knowledge about property 3.2.3.2 The implementation of the land tenure reform on grasslands or use rights (Richard et al, 2006); and (4) private use by individual households destroys the traditional Table 3.1 provides a typology of the assignment of use rights and the actual ways of grassland use in institutions of grassland use and management (Li, 2012). In what follows, we will investigate another the pastoral areas of China since the start of the land tenure reform. This typology is in line with the possible explanation for why the adoption of private use is lagging, namely that privatisation negatively investigation of land tenure reform on the Tibetan plateau by Banks (2003). Three types of assignment affects the grassland condition. of grassland use rights existed during the last three decades, namely, formal use rights assigned to 3.2.3.3 Land tenure reform in sample counties individual households, to a group of households or to a collective. Correspondingly, there were three ways of grassland use, including private use, joint use and common use. Private use indicates that a Table 3.2 gives indications of the land tenure reform progress of grasslands in the selected counties certain area of grasslands is managed and used by an individual household, while other users are between 1985 and 2008. In 1985, only 6.67% of the 60 counties had assigned grassland use rights to excluded. Joint use means that a group of households manage and use an area of grasslands together, in individual households. By 2008, this proportion had increased to 81.67%, and the average number of which group members are often neighbours or relatives that have small parcels of grasslands and wish years since formal use rights had been contracted to individual households was 11.55. Meanwhile, actual to make their land larger and contiguous. Finally, common use means that the grasslands within a changes in the use of the grasslands followed this institutional evolution. In 1985, only 5% of the collective or administrative village are used by all of the members of the collective or the villagers. selected counties had adopted the practice of private use. In 2008, this percentage had reached to 63.33% and the average number of years since private use had been adopted was 8.72. These observations about Table 3.1 The assignment of formal use rights of grasslands and the actual ways of grassland use the land tenure reform in the sample counties demonstrate that the adoption of private use lagged behind

Actual use of grasslands the privatisation of formal use rights. Formal use rights assigned to Private use Joint use Common use Individual households + + + Table 3.2 Land tenure reform progress in the selected counties, 1985-2008 Group of households na + + Unit 1985 1995 2000 2005 2008 Collective/administrative village na na + The proportion of counties in which Percent 6.67 23.33 76.67 80.00 81.67 Source: adapted from “formal and de facto grassland management units” (Banks et al., 2003) grassland use rights have been assigned to na indicates not available. individual households The average number of years since the use Years 0.18 1.92 5.17 9.13 11.55 When the use rights are assigned to a collective or an administrative village, the actual way of grassland rights have been assigned to households use is always common use. If use rights are assigned to a group of households, actual use can be common The proportion of counties with actual Percent 5.00 15.00 58.33 63.33 63.33 private use of grasslands use or joint use. Private use, joint use and common use can all occur when the use rights are assigned The average number of years since actual Years 0.13 1.22 3.68 6.82 8.72 to individual households. The intention of the land tenure reform in the pastoral areas is, however, to private use has been adopted achieve full private use when use rights are assigned to households. This has spurred criticisms and claims of failure of the privatisation process in China's extensive grasslands, compared to the case of the cropland areas where household-based use rights were assigned overnight and private use was established simultaneously (Ho, 2000; Banks et al., 2003). Scholars (e.g. Richard et al, 2006; Li and

43 44

established simultaneously (Ho, 2000; Banks et al., Bankset al., established simultaneously (Ho, 2000; rights use household-based where areas the cropland inChina' process failure theprivatisation of claims of assign are rights when use use achieve fullprivate to individualhouseholds. te Theintentionoftheland assigned are rights theuse when all occur usecan common use, joint use and Private use or joint use. use is always common Ifuse. use rights are assign or collective toa assigned are theuserights When na indicates not available. managemental.,2003) grassland units” (Banks et anddeSource: facto from adapted “formal collective oradministrative village to make their land larger Finally,and contiguous. or relatives often neighbours which group are members useexcluded. Joint means thatagroup mana ofhouseholds ismanaged grasslands andused of certain area use,join private including grassland use, ways of individual households,to agroup to ofhouseholdsor during lastthre rights existed of grassland the use assignment of types Three Banks (2003). plateau by Tibetan the on investigation reform landtenure of landte ofthe sincethestart the pastoralareasofChina in use grassland of ways actual the and rights of use assignment the of a typology provides Table 3.1 grasslands on reform 3.2.3.2 Theimplementationofthelandtenure hectares, outof400millionhectar assi been grasslands thathave of 2014, thetotalarea tradition households hasreplaced pastora gradually extensive in the conducted has been government in2011(Li, report theselegi 2012).With Collective/administrative village Collective/administrative village Group of households Individual households to rightsFormal use assigned Table 3.1 The assignment of formal use rights of grasslands and the actual ways of grassland use assignmentTablegrasslandsways 3.1 ofgrassland Theformal useactual rights of of use and the

al open access to land (Hua and Squi land(Hua al openaccessto es of grasslands in China (Ministry of Agriculture of China, 2015). inChina(Ministrygrasslands Agriculturees of ofChina,2015). of are used by all of the members of the collective or the villagers. orthevillagers. collective of themembers the all usedby are ed to a group of households, actual use can be common be actualusecan toagroup ofhouseholds, ed na na + Private use grasslandsActual useof an administrative village, the actual way of grasslandan administrative of theactualway village, 43 43 byan individualhousehold,whileother usersare t use and common use. Private use indicates that a that indicates Privateuse commonuse. t useand

ed to households. This has spurred criticisms and hasspurred criticisms This tohouseholds. ed e decades, namely, formal use rights assigned to assigned formal userights namely, e decades, gned to individual households reached 223 million gned reached 223 toindividualhouseholds common the use grasslands means within that a 2003). Scholars (e.g. Richard et al, 2006; Li and Richard2003). Scholars(e.g. etal,2006; nure reform in areas the pastoral is, however, to slative efforts, the privatisation of the grasslands theprivatisationofgrasslands efforts, slative l areas of China, and increasingly private use by by use private increasingly China,and of l areas s extensive grasslands, compared to the case of of case grasslands, compared to the extensive s were assigned overnight and private use was use private assigned overnight were and a collective. Correspondingly, there were three there a collective.Correspondingly, that have small parcels of nure reform. isinlinewiththe Thistypology ge and use an area of grasslands together, inge ofgrasslands together, anduseanarea na + + use Joint res, 2015; Yang et al., 2004). By grasslands and wish + + + Common use

the privatisation of formal use rights. the privatisation userights. offormal the landtenurereform inthesample yearsand theaveragenumberof sinceprivateuseha In to 63.33% reached 2008, had this use. percentage ofprivate thepractice adopted counties had selected grasslandsfollowed this the useof changes inthe hadbeencontracted formalyears userights since individual households.By hadin 2008, thisproportion Inbetween 1985 and 2008. 1985, only of the 60 countie 6.67% tenurereform gives indicationsoftheland Table 3.2 Land insamplecounties reform 3.2.3.3 tenure condition. grassland the affects forpossible explanation whyadoption ofprivateuse the institutions of grasslanduse and management 2012) (Li, or use rights (Richard et al, 2006); and (4) use private ownershipknowledge aboutproperty (3) isinconsistentwith local individualhousehold or traditional al., 2007); Li (Ho,rights, tobuildfences et 1996; property e.g. protecting for financial resources private heterogeneity of resources in arid and semi-arid inaridandsemi-arid ofresources heterogeneity privatisation of grassland use rights is incompatible China: (1) of areas pastoral tenure reform in the Li,Zhang,have identifiedanumberof 2009; 2012) private use has been adopted adopted private usehasbeen yearsactual since of The average number grasslands private useof withactual The proportionofcounties tohouseholds assigned rights been have use years sincethe of number The average individual households to beenassigned grassland have userights inwhich The proportionofcounties Table 3.2 Land tenure reform progress in the selected counties, 1985-2008 Land counties, progress 1985-2008 Table 3.2 intheselected reform tenure counties demonstrate to individualhouseholdswasto 11.55.Meanwhile,actual Percent Percent Unit Years Years Percent Years 44 44 areas (Li and Zhang, 2009); (2) there is a lack of (2)Zhang, thereisalack 2009); areasand (Li

d been adopted was 8.72. These observations about Theseobservationsd beenadoptedwas about 8.72. institutional evolution. In institutionalevolution. the 5%of 1985,only the inflexible boundaries stemming from the from boundaries stemming theinflexible by individual households destroys the traditional destroys the by individualhouseholds with theneedfor livestock mobility duetothe possible barriers to thecompletionoflandpossible barriers to progress of grasslands in the selected counties progress counties in the grasslands selected of creased to 81.67%, and the average number of number and theaverage to81.67%, creased is lagging, namely that is lagging, namely . In. follows,we willinvestigate what another that the adoption of private use lagged behind that theadoptionofprivateuselagged behind 0.13 5.00 0.18 6.67 1985 s had assigned grassland use rights to s hadassigned grassland userights 1.22 15.00 1.92 23.33 1995 3.68 58.33 5.17 76.67 2000 privatisation negatively privatisation negatively 6.82 63.33 9.13 80.00 2005 8.72 63.33 11.55 81.67 2008

3 3.3 Empirical model moderate grassland and sparse grassland. According to the research by Feng et al. (2009) and Deng et

3.3.1 Fixed effects model al. (2011), dense grassland has the highest quality and its grass cover is more than 50 percent. Moderate grassland has grass covering 20-50 percent of the land, and sparse grassland with 5-20 percent. In our The fixed effects model has been widely used in economic research (e.g., Fergusson et al., 2002; Huang case, it is supposed that the grassland degradation is manifested by the decrease of total, dense and et al., 2006). It is used primarily to study the causes of changes within entities over time. The general moderate grassland areas, as well as the increase of sparse grassland area when total grassland area with model is presented as follows: decrease or no changes. The latter considers that the dense and moderate grasslands may degrade into

sparse grassland, which was observed by the research about the moving direction of grassland centre of gravity in China during 1976-1996 by Feng et al. (2009). Hence, four dependent variables are included ܻ௜௧ ൌ ߚܺ௜୲ ൅ܷ௜ ൅߮௜௧ሺͳሻ where i and t present entity i and year t, respectively. is the dependent variable. denotes the vector in the empirical analysis to indicate the grassland condition: total grassland area, dense grassland area, of independent variables. is the specific interceptܻ ௜௧for each entity and representsܺ௜௧ all time-invariant moderate grassland area, and sparse grassland area. The following equations for estimation are derived (fixed) characteristics of entityܷ௜ i that influence the dependent variable, including both observable and from the general fixed effects model (see equation (1)): unobservable factors. is the random error term. The difficulty in estimating the parameters of β is that involves unobserved௜௧ factors. In this regard, a within transformation is used to remove all time- ߮ ଶ ଶ ଶ ଶ ௜௧ ௜ ଵ ௜௧ ଶ ௜௧ ଷ ௜௧ ସ ௜௧ ହ ௜௧ ଺ ௜௧ ଻ ௜௧ ଼ ௜௧ ଽ ௜௧ ௜௧ invariantܷ௜ (fixed) explanatory variables, i.e. a model is performed using deviations from individual ܩ ൌ݃ ൅ܽ ܮ ൅ܽ ܮ ൅ܽ ܶ ൅ܽ ܶ ൅ܽ ܨ ൅ܽ ܨ ൅ܽ ܶܨ ൅ܽ ܥ ൅ܽ ܥ ൅ᖡ ሺʹሻ ଶ ଶ ଶ ଶ means (Verbeek, 2012). Therefore, the fixed effects model is a method of estimating the parameter of ௜௧ ௜ ଵ ௜௧ ଶ ௜௧ ଷ ௜௧ ସ ௜௧ ହ ௜௧ ଺ ௜௧ ଻ ௜௧ ଼ ௜௧ ଽ ௜௧ ௜௧ ܦ ൌ݀ ൅ܾ ܮ ൅ܾ ܮ ൅ܾ ܶ ൅ܾ ܶ ൅ܾ ܨ ൅ܾ ܨ ൅ܾ ܶܨ ൅ܾ ܥ ൅ܾ ܥ ൅Ɋ ሺ͵ሻ β that takes the observable as well as unobservable time-invariant explanatory variables ( ) into ଶ ଶ ଶ ଶ ௜௧ ௜ ଵ ௜௧ ଶ ௜௧ ଷ ௜௧ ସ ௜௧ ହ ௜௧ ଺ ௜௧ ଻ ௜௧ ଼ ௜௧ ଽ ௜௧ ௜௧ ܯ ൌ݉ ൅ܿ ܮ ൅ܿ ܮ ൅ܿ ܶ ൅ܿ ܶ ൅ܿ ܨ ൅ܿ ܨ ൅ܿ ܶܨ ൅ܿ ܥ ൅ܿ ܥ ൅ߛ ሺͶሻ account, but the estimation does not dependent on the value of (Fergusson et al., 2002). ܷ௜ ଶ ଶ ଶ ଶ ௜௧ ௜ ଵ ௜௧ ଶ ௜௧ ଷ ௜௧ ସ ௜௧ ହ ௜௧ ଺ ௜୲ ଻ ௜௧ ଼ ௜௧ ଽ ௜௧ ௜௧ ܷ௜ ܵ ൌݏ ൅݀ ܮ ൅݀ ܮ ൅݀ ܶ ൅݀ ܶ ൅݀ ܨ ൅݀ ܨ ൅݀ ܶܨ ൅݀ ܥ ൅݀ ܥ ൅ߪ ሺͷሻ The fixed effects model is appropriate to solve our research question considering that the results of the where i and t present the ith county and year t. Total grassland area ( ), dense grassland area ( ), land tenure reform within each county can be studied effectively by controlling for the unmeasured moderate grassland area ( ) and sparse grassland area ( ) are the dependentܩ௜௧ variables. Equationsܦ (2),௜௧ heterogeneity among counties. As shown in table 3.2, the progress in land tenure reform changes (3), (4) and (5) will be referredܯ௜௧ to as the total, the dense,ܵ the௜௧ moderate and the sparse grassland model, substantially across our research period and for a considerable proportion of the counties in our dataset, respectively. which satisfies the basic requirement of the fixed effects model (Daun-Barnett, 2011). Moreover, our research sample attempts to include all of the relevant counties in Inner Mongolia, rather than random indicates our main variable of interest, the progress of the land tenure reform in county i until year draws, which preliminarily indicates the fixed effects model more appropriate than the random effects ܮt. ௜௧It is measured as the number of years until year t that private use has been adopted in county i. We model (Verbeek, 2012). Nevertheless, the appropriateness of the fixed effects model will be further focus on the years of actual private use of the grasslands rather than the years of the assignment of examined through statistical testing in section 5. formal use rights to households because it is considered that the actual use will affect the grassland condition. As was shown in table 3.2, the adoption of actual private use lags behind the assignment of 3.3.2 Variable definitions and expected results formal use rights, and the correlation value between the years of assigning formal use rights and actually The dependent variable in the empirical analysis is the grassland condition. The grassland condition adopting private use is 0.825. In this regard, including both the years of formal use rights assignment includes the situations of both quantity and quality of grasslands, which are quantified considering the and the years of actual private use as variables in the model would cause multicollinearity. The quadratic extent and density of grass cover that are suggested as the crucial manifestation of grassland degradation term of , , is included in the model because a non-linear relationship is expected between the land ଶ (Feng et al., 2009; Li et al., 2013). Specifically, the quantity of grasslands is measured by the total area tenure reformܮ௜௧ ܮ௜௧ and the grassland condition. of grasslands in hectares, while the areas with less than 5 percent of grass cover are not counted as To correctly estimate the effect of the land tenure reform, we control for other factors that may affect grasslands. And the quality of grasslands is presented through the areas of grasslands of different the condition of the grasslands, namely, the factors related to climate (Gao et al., 2010; Li et al., 2012; subclasses that are classified according to the different density of grass cover, including dense grassland, 45 46

that of independent variables. of independent variables. substantially across our research period and for a c a andfor period research substantially our across As shownintab counties. among heterogeneity controlling fortheunmeasured by effectively bestudied can county each reform within land tenure effects modelisappropriatetosolveourre The fixed subclasses that are classified according to thediffer to according classified are subclasses that grasslands ispres of thequality grasslands. And while theareasof grasslands inhectares, withless (FengLi et al., 2009; 2013). Specifically, quantity et al., grasslands the of measured by is the total area coverthatare suggeste grass anddensity of extent includes thesituationsof bothquantity andquality theempiricalanalysis variablein The dependent Variable andexpected definitions 3.3.2 results throughexamined statisticaltesting 5. insection model (Verbeek, 2012). Nevertheless, eff preliminarily the draws,fixed indicates which therele toincludeallof research attempts sample require which satisfiesthebasic that takes the observable as well as unobservable time asunobservable aswell β thattakes theobservable means (Verbeek, 2012). Therefore, the fixed ef Therefore,fixed the means (Verbeek, 2012). a mo explanatoryinvariantvariables, i.e. (fixed) model ispresented as follows: Itet al.,2006). isusedprimarily tostudy cause the in been widely used model has effects The fixed effectsmodel 3.3.1 Fixed 3.3 Empirical model where i and t present entity i and yearrespectively. t, and entityi tpresent where iand account, but the estimation does not dependent on thevalue of doesnotdependenton account, buttheestimation unobservable factors. th entity ithatinfluence of characteristics (fixed) ܷ ௜ involves unobserved factors. In this regard, a within transformation is used to remove all time- remove used to is a withintransformation In thisregard, involves unobservedfactors. ߮ ௜௧ is the random difficulty erroresti term. The in ܷ ௜ for the each specific en intercept is ܻ ment of the fixed effects model (Daun-Barnett, 2011). Moreover, our 2011).Moreover, our effects model(Daun-Barnett, fixed ment ofthe ௜௧ ߚܺ ൌ the appropriateness of the fixed of thefixed theappropriateness ௜୲ ൅ܷ ௜ fects modelisamethodof ൅߮ economic research (e.g., Fergusson researchet al.,2002;Huang economic (e.g., d as the crucial manifestation of grassland degradation manifestation as the crucial d ects model more appropriate than the randomappropriate thanects the model more effects is the grassland condition. The grassland condition is thegrassland condition.Thegrassland e dependent variable, including both observable and includingand both observable dependentvariable, e vant counties in Inner Mongolia, ratherInner thanrandom Mongolia, vant countiesin le 3.2, the progress in land tenure reform changes le 3.2,theprogress inlandtenurereform changes onsiderable proportion of the counties in our dataset, thecountiesinourdataset, onsiderable of proportion 45 45 ent density of grass cover, including dense grassland, includingdense grass cover, of ent density del is performed using deviations from individual deviations from individual using del isperformed ented through the areas of grasslands of differentgrasslands of of areas the ented through

s ofchanges withinentiti of grasslands, which are qua grasslands, whichare of ௜௧ ܻ than 5 percent of grass cover are notcountedas coverare than 5percent ofgrass search questionconsideringsearch thattheresults ofthe ௜௧ ሺͳሻ is the dependent variable. variable. the dependent is -invariant explanatory variables ( ܷ ௜ (Fergusson et al., 2002). et al.,2002). (Fergusson tity and represents all time-invariant time-invariant all tity and represents mating of theparameters effects model will be further

estimating the parameterof es overtime.Thegeneral ntified considering the ntified considering ܺ ௜௧ denotes the vector vector denotes the ܷ ௜ ) into β is the condition of the grasslands, namely, relatethe factors re effect ofthelandtenure the To correctlyestimate grassland tenure reform condition. and the ܮ respectively. as thetotal,th referred to (5)willbe (3), and (4) term of generalfrom (1)): equation the effects model(see fixed and the years use as variables in th and the of actual private adoptingIn privateuseis0.825. thisregard,years boththe including use rights of formal assignment the valuebetween correlation formal rights, andthe use ofac adoption intable3.2,the condition. Aswas shown formal use rights to households because it is consid years of grasslands oftheassignment thanthe rather use ofthe yearsfocus private onthe ofactual yearyears until of number Itas the ismeasuredt. grassland area and sparse grasslandmoderate area, la ofthe percent grass 20-50 grassland has covering and it grassland highest quality has the dense al. (2011), According to grassland. sparse and grassland moderate in the empirical analysis to indi analysis in the empirical gravity by inChinaduring Feng etal. 1976-1996 wasresear observed by thegrassland, sparse which The latterthe considers thatchanges. decrease or no moderate asgrassland well areas, as the increase sp of and dense oftotal, decrease the by is manifested grassland degradation the itissupposedthat case, moderate grasslandmoderate area ( where i the iandtpresent ௜௧ of the land tenure reform in county i until year iuntil reform incounty ofthelandtenure the progress ofinterest, indicates ourmainvariable ܩ ܯ ܦ

ܵ ܮ ௜௧ ௜௧ ௜௧ ௜௧ ௜௧ , ܮ ൌݏ ൌ݃ ൌ݀ ൌ݉ ଶ ௜௧ ௜ , is included in the model because anon-linea , isincludedinthemodelbecause ௜ ௜ ൅݀ ௜ ൅ܽ ൅ܾ ൅ܿ ଵ ଵ ଵ ܮ ଵ ܮ ܮ ௜௧ ܮ ௜௧ ௜௧ ௜௧ ൅݀ ൅ܾ ൅ܽ ܯ ൅ܿ th ௜௧ county and year t. Total grassland area ( year grassland county area t.Total and ଶ ଶ ଶ ) andsparse ( grassland area ଶ ܮ ܮ ܮ ܮ ଶ ௜௧ ଶ ௜௧ ଶ ௜௧ ଶ cate the grassland condition: total gr condition: total grassland cate the ௜௧ ൅݀ ൅ܾ ൅ܽ ൅ܿ ଷ ଷ ଷ ଷ ܶ ܶ ܶ ܶ ௜௧ ௜௧ ௜௧ ௜௧ ൅݀ ൅ܾ ൅ܽ ൅ܿ ସ ସ ସ ସ ܶ ܶ ܶ ܶ ௜௧ ௜௧ ௜௧ ଶ ௜௧ ଶ ଶ ଶ e dense, the moderate and the sparse grasslandthe sparse model, moderatedense, and the e ൅ܾ ൅݀ (2009). Hence, four dependent variables are included variablesare dependent (2009). Hence, four ൅ܿ ൅ܽ e modelwouldcausemulticollinearity. Thequadratic 46 46 t that private use has been adopted in county i. We i. in county adopted has been use t that private . Thefollowing for estimationare equations derived ch about the moving direction of grassland centre of about of thech movinggrasslandcentre direction of

nd, and sparse grassland with 5-20 percent. In percent. our with5-20 grassland nd, andsparse form, we control we factors for that may other form, affect ହ ହ ered that the actual use will affect the grassland willaffectthe ered thattheactualuse ହ arse grassland total arsearea when areawith dense and moderate grasslands may degrade dense andmoderate into ହ ܨ ܨ ܨ ܨ years of assigning formal use rights and actually and actually rights use formal years assigning of s grass cover is more than 50 percent. Moderate Moderate percent. grass than 50 is more cover s ௜௧ ௜௧ ௜௧ the research by Feng et al. (2009) and Deng et Deng and (2009) et al. Feng by research the ௜௧ ܵ tual private use lags behind the assignment of theassignment behind uselags tual private d to climate (Gao et al., 2010; Lial., 2012; et al., 2010; et climate (Gao d to ൅ܾ ௜௧ ൅ܿ ൅݀ ൅ܽ ) are the dependent var ) are thedependent r relationship is expected between the land isexpectedr relationship theland between ଺ ଺ ଺ ଺ ܨ ܨ ܨ ܨ ௜௧ ௜௧ ଶ ଶ ௜୲ ௜௧ ଶ ଶ ൅ܿ ൅ܾ ൅݀ ൅ܽ assland area, dense grassland area, grassland area, area,dense assland ଻ ଻ ଻ ଻ ܩ ܶܨ ܶܨ ܶܨ ܶܨ ௜௧ ௜௧ ), dense grassland( ), dense area ௜௧ ௜௧ ௜௧ ൅ܿ ൅ܾ ൅݀ ൅ܽ ଼ ଼ ଼ ܥ ଼ ܥ ܥ ܥ ௜௧ ௜௧ iables. Equations(2), ௜௧ ௜௧ ൅ܿ ൅ܾ ൅݀ ൅ܽ ଽ ଽ ܥ ଽ ଽ ܥ ௜௧ ܥ ܥ ଶ ௜௧ ଶ ௜௧ ௜௧ ଶ ଶ ൅ߛ ൅Ɋ ൅ߪ ൅ᖡ ௜௧ ௜௧ ௜௧ ௜௧ ሺͶሻ ሺ͵ሻ ሺͷሻ ሺʹሻ ܦ ௜௧ ), ),

3 Lu et al., 2015), agricultural activities (He et al., 2014; Hua and Squires, 2015) and heterogeneity among counties (Deng et al. 2011; He et al., 2014). Climate factors are presented by temperature ( ) and precipitation ( ) (Gong et al., 2015). Their quadratic terms ( and ) and cross term ( ܶ௜௧ ) are ଶ ଶ considered (Liܨ et௜௧ al., 2013). is the percentage change in cultivatedܶ௜௧ landܨ௜௧ in county i in the fiveܶܨ௜௧ years Sparse grassland model Dependent variable -/+ +/- -/+ +/- - + -/+ + - preceding year t and is itsܥ௜௧ quadratic term. and are considered as controls for the switch of ଶ ଶ land between grasslandܥ௜௧ and arable land whichܥ ௜௧was shownܥ௜௧ to have a significant and direct effect on grassland quantity in China (Feng et al., 2009; Li et al., 2013). Finally, the factors about the heterogeneity among counties such as elevation, slope, soil type and distance to the provincial capital do not change significantly over time and are therefore treated as time-invariant (fixed) factors in the Moderate Moderate modelgrassland Dependent variable +/- -/+ +/- -/+ + - +/- - + model, denoted by , , and in equations (2) to (5). Variable definitions and the expected impact of independent variables݃௜ ݀௜ on݉ ௜the grasslaݏ௜ nd condition are listed in table 3.3. Expected results

Dense grassland grassland Dense model Dependent variable +/- -/+ +/- -/+ + - +/- - +

48 Total grassland grassland Total model Dependent variable +/- -/+ +/- -/+ + - +/- - +

ଶ

ଶ ଶ

ଶ Unit Hectares Hectares Hectares Hectares Years Celsius degree Millimeter Celsius degree* Millimeter Percentage Table 3.3 Variable definitions and their expected results

௜௧ ܨ and

௜௧ ௜௧ ௜௧ ௜௧ ௜௧ ܶ ܮ ܶ ܨ ܥ Variable definition Variable Total grassland area of i in county t year Dense areagrassland of i in county t year i area of county Moderate grassland tin year i in area of Sparse county grassland t year The number of until t that years year actual private use of grasslands has been adopted in i county term of Quadratic The annual average temperature in i in tcounty year term of Quadratic i county precipitation in annual The tin year term of Quadratic term of Interaction average percentageThe change in land of cultivated in county the area preceding t i in the five years year term of Quadratic

௜௧ ଶ ଶ ௜௧ ଶ ௜௧ ௜௧ ௜௧ ௜௧ ௜௧ ௜௧ ௜௧ ௜௧ ௜௧ ௜௧ ௜௧ ଶ ܵ ܶ ܶ ܨ ܨ ܮ ܮ ܥ ܥ ܩ ܦ ܯ ܶܨ Variables 47

counties (Deng et al. 2011; He et al., 2014). Climate factors are presented by temperature ( by temperature are presented factors 2014). Climate et al., He et al.counties 2011; (Deng Luagriculturalet al., 2015), activit on the grasslaof independent variables preceding yearand t preceding model, denotedby therefore timeandare over do notchange significantly elevation,such as sl counties among heterogeneity grassland quantityLi in China (Feng et al., 2009; andarablegrassland la land between considered(Li et al., 2013). precipitation ( ܨ ௜௧ ) (Gong etal.,2015).Theirquadratic terms( ݃ ௜ , ݀ ܥ ௜௧ ଶ ௜ is its quadratic term. term. quadratic its is , ݉ ௜ ܥ and ௜௧ is the percentage change in cultivated land in county i in the five yearscounty iinthefive cultivated landin changeis thepercentage in ݏ ies (He et al., 2014; Hua and Squires, 2015) and heterogeneity among among andheterogeneity andSquires,2015) Hua al., 2014; ies (Heet ௜ in equations (2) to (5). Variable de (5). (2)equations to in nd which was showntohaveasi was nd which nd condition are listed in table 3.3. are listedintable3.3. nd condition ܥ ௜௧ and 47 47 ope, soil type and distance to the provincial capital totheprovincialcapital ope, soiltype anddistance

etal.,2013).Finally,about the thefactors ܥ ௜௧ ଶ treated as time-invariant (fixed) factors in the inthe factors (fixed) astime-invariant treated of for theswitch areas controls considered ܶ ௜௧ ଶ and finitions and the expected impact impact expected finitions andthe ܨ gnificant and direct effect on and direct effect gnificant ௜௧ ଶ ) and cross term ( ܶܨ ܶ ௜௧ ௜௧ ) and ) and ) are

Table 3.3 Variable definitions and their expected results

Variables Variable definition Unit Expected results Total grassland Dense grassland Moderate Sparse grassland model model grassland model model Total grassland area of county i in Hectares Dependent year t variable ௜௧ ܩ Dense grassland area of county i in Hectares Dependent year t variable ௜௧ ܦ Moderate grassland area of county i Hectares Dependent in year t variable ௜௧ ܯ Sparse grassland area of county i in Hectares Dependent year t variable ௜௧ ܵ The number of years until year t that Years +/- +/- +/- -/+ actual private use of grasslands has ܮ௜௧ been adopted in county i Quadratic term of -/+ -/+ -/+ +/- ଶ ଶ The annual average ௜௧temperature in Celsius degree +/- +/- +/- -/+ ܮ௜௧ ܮ county i in year t ௜௧ ܶ Quadratic term of -/+ -/+ -/+ +/- ଶ ଶ The annual precipitation௜௧ in county i Millimeter + + + - ܶ௜௧ ܶ in year t ௜௧ ܨ Quadratic term of - - - + ଶ ଶ ௜௧ Celsius degree* +/- +/- +/- -/+ ܨ௜௧ Interaction term ofܨ and Millimeter ௜௧ ௜௧ ௜௧ ܶܨ The average percentageܶ changeܨ in Percentage - - - + the area of cultivated land in county ܥ௜௧ i in the five years preceding year t Quadratic term of + + + - ଶ ଶ ܥ௜௧ ܥ௜௧

3 Given the contradicting views in the literature about the effect of privatisation on grassland quality, the expected effect of in our estimations is ambiguous (e.g. Hardin, 1968; Guelke, 2003; Galvin, 2009). Std. Dev. 810,785 513,817 368,679 160,195 7.69 2.58 82.89 5.63 However, we expectܮ௜௧ that the effects of the land tenure reform will be strong at the beginning and reduce in intensity over the years. Hence an opposite sign to that of is expected for the coefficients of in ଶ the four models. It is noted that the expected effects of all explanatoryܮ௜௧ variables in the sparse grasslandܮ௜௧ 2008 Mean 761,399 365,027 272,815 122,126 8.72 5.20 346.65 0.65 model are the opposite of the expectations in the other three models because grassland degradation involves a decrease in total, dense and moderate grassland areas, but an increase in sparse grassland

area. Additionally, the expected sign of temperature ( ) is ambiguous because higher temperatures Std. Dev. 806,050 513,908 368,618 158,800 6.47 2.80 109.74 7.54 facilitate plant growth, but also cause fast evaporationܶ௜௧ from ground surfaces and stimulate the development of pests (Deng et al., 2013; Li et al., 2013). Precipitation ( ) improves the grassland 2005 Mean 759,026 365,300 272,747 121,057 6.82 4.74 273.40 -2.06 condition in the arid and semi-arid areas (Li and Zhang, 2009). Both temperatureܨ௜௧ and precipitation are expected to have nonlinear relationships with the grassland condition (Li et al., 2012; Zhou et al., 2002), and therefore the opposite signs to those of and are expected for the coefficients of and in ଶ ଶ Std. Dev. 808,624 514,489 374,282 162,608 4.69 2.98 80.70 6.61 the four models. The interaction term of ܶ ௜௧and ܨ ௜௧is also included because the effect ofܶ temperature௜௧ ܨ௜௧ depends on the amount of precipitation. Finally,ܶ௜௧ ܨthe௜௧ intensive agricultural activities were suggested to

cause grassland degradation (Zhang et al., 2007; Feng et al., 2009). We therefore expect negative 2000 Mean 762,376 364,280 275,346 123,152 3.68 4.52 260.55 5.52 coefficients for in the total, dense and moderate grassland models, and a positive coefficient in the

sparse grassland model.௜௧ Again a nonlinear relationship is expected for this variable. ܥ 50 Std. Dev. 815,044 520,029 380,816 160,332 3.24 2.55 84.35 5.65 3.3.3 Descriptive statistics on the variables

Table 3.4 shows the mean value and standard deviation of our main variables for the research sample. 1995 Mean 772,780 377,815 277,557 118,894 1.22 4.45 335.78 2.99 In general, the areas of total grassland ( ), dense grassland ( ) and moderate grassland ( ) decreased, and the area of sparse grasslandܩ (௜௧ ) presented fluctuantܦ ௜௧increases between 1985 and 2008.ܯ௜௧ The areas of total, dense and moderate grasslands௜௧ experienced their highest points in 1985, and in 2000 Descriptive variables statistics 3.4 main of Table ܵ Std. Dev. 810,035 517,236 378,658 160,957 0.60 2.86 113.74 8.86 for sparse grassland area. Specifically comparing 1985 and 2008, the total grassland area decreased which indicates the degradation of grassland quantity. Furthermore, given the decrease in total grassland area, the decrease in dense and moderate grassland areas and increase in sparse grassland area 1985 Mean 778,779 380,860 278,432 120,627 0.13 3.31 373.87 2.06 demonstrate the degradation of grasslands with higher grass cover into lower quality grasslands. Moreover, the average number of years with actual private use in the 60 sample counties was increasing from 0.13 in 1985 to 8.72 in 2008. And the standard deviations of indicate that the differences in

adopting private use among the 60 sample counties were increasingܮ from௜௧ 1985 to 2008. The average Unit Hectares Hectares Hectares Hectares Years Celsius degree Millimeter Percent temperature reached its highest point in 2008, while annual precipitation was the highest in 1985 and the lowest in 2000. The average percentage change in cultivated land increased from 1985 to 2000 then Obs 60 60 60 60 60 60 60 60

decreased to become even negative in 2005.

௜௧ ௜௧ ௜௧ ௜௧ ௜௧ ௜௧ ௜௧ ௜௧ ܵ ܶ ܨ ܮ ܥ ܩ ܦ ܯ Variables 49 based on ourand Statistical Source: survey MongoliaYearbooks of Inner

in intensity over the years. to thatof an oppositesign in intensity Hence the over land tenure that theof effects the expect However, we In general, the areas of total grassland ( totalgrassland of general, areas In the deviati and standard meanvalue showsthe Table 3.4 statist Descriptive 3.3.3 for thisvariable.Again isexpected grasslandrelationship anonlinearsparse model. and therefore the opposite signsand therefore the opposite to those of withth relationships tohavenonlinear expected condition in the arid and areasa (Lisemi-arid decreased to become even negative in 2005. negativein2005. even decreased tobecome in change The average percentage the lowest in 2000. reachedtemperature itshighest pointin2008,wh average 1985to2008.The from were increasing counties 60sample the among use adopting private expected effect of of effect expected about intheliterature contradicting views Given the coefficients for coefficients cause grasslandcause degradation (Zhang Fe et al., 2007; precipitation.Finally, thein amountof depends onthe from 0.13 in 1985 to 8.72 in 2008. And the standard deviations of deviations standard And the in2008. 8.72 1985to from 0.13in Moreover, the average number of years of withactual number Moreover, theaverage ofgrademonstrate thedegradation grassla moderate and in dense thedecrease area, which indicatesthedegradationofgr 1985and2008,grassland Specificallyfor thetotal comparing sparse area. grassland decreased area The areasoftotal,denseandmode development of pests (Deng et al., facilitate plant growth, but also cause fast ev decreased, and the area of sparse grassland ( sparse area of decreased, andthe temperature sign ( of expected the area. Additionally, grassl denseandmoderate intotal, involves adecrease model are the opposite of in the expectations the ot Itthe fourmodels. effects isnotedthattheexpected ofallexplanatory inthesparse variables grassland the four of models.Theinteraction term ܥ ௜௧ ܮ in thetotal,grassland denseandmoderate ௜௧ in our estimations isambiguous (e.g ics on the variables rate grasslands experienced their highest points in their highest grasslands 1985, and inexperienced rate 2000 2013; Li 2013).Precipitation2013; ( etal., assland quantity. Furthermore, given the decrease intotalgrassland decrease given the Furthermore, assland quantity. sslands with higher grass cover into lower quality grasslands. grasslands. sslands withhigher lowerquality grass into cover ܶ ܩ ௜௧ ܶ ௜௧ and ܵ ௜௧ ௜௧ ), dense grassland ( ), dense nd Zhang,Both are temperature andprecipitation 2009). and ) presented fluctuant increases between 1985and2008. fluctuantincreases ) presented e grasslandal., condition 2012; Zhou (Lial., 2002), et et ܨ aporation from ground surfaces and stimulate the aporation from andstimulatethe groundsurfaces ile annualprecipitation thehighest was in1985and 49 49 ௜௧ ܨ private use in the 60 samp

is also included because ௜௧ the effect of privatisation on grassland quality, the on privatisation the effect of nd areas and increase in sparse grassland area insparse increase nd areasand cultivated land increased from 1985 to 2000 then to 2000 then 1985 increased from land cultivated on of our main variables for the research sample. sample. research forthe mainvariables on ofour her three models because grassland degradation grassland degradation her threemodelsbecause are expected for the coefficients of thecoefficients for are expected reform will be strong at the beginning and reduce at thebeginningreform and reduce willbestrong ng et al., 2009). We therefore expect negative negative expect therefore al., 2009).We et ng tensive agricultural activities were suggested to activities were suggested tensive agricultural ܶ and areas, but an increase in sparse grassland insparse butanincrease and areas, ௜௧ ) is ambiguous because higher temperatures temperatures higher because ambiguous ) is . Hardin, 1968; Guelke, 2003; Galvin, 2009). 2003;Galvin, 2009). 1968;Guelke, . Hardin, ܮ ௜௧ models, coefficient and a positive in the is expected for the coefficients of for thecoefficients is expected ܦ ௜௧ ܮ ௜௧ ) and moderate grassland ( ) andmoderate in differences indicate that the ܨ ௜௧ le counties was increasing le countieswas increasing ) improves thegrassland the effect of temperature temperature effectof the ܶ ௜௧ ଶ and ܮ ܨ ܯ ଶ ௜௧ ௜௧ ଶ in in in ௜௧ )

Table 3.4 Descriptive statistics of main variables

Variables Obs Unit 1985 1995 2000 2005 2008 Mean Std. Dev. Mean Std. Dev. Mean Std. Dev. Mean Std. Dev. Mean Std. Dev. 60 Hectares 778,779 810,035 772,780 815,044 762,376 808,624 759,026 806,050 761,399 810,785

ܩ௜௧ 60 Hectares 380,860 517,236 377,815 520,029 364,280 514,489 365,300 513,908 365,027 513,817 ܦ௜௧ 60 Hectares 278,432 378,658 277,557 380,816 275,346 374,282 272,747 368,618 272,815 368,679 ܯ௜௧ 60 Hectares 120,627 160,957 118,894 160,332 123,152 162,608 121,057 158,800 122,126 160,195 ܵ௜௧ 60 Years 0.13 0.60 1.22 3.24 3.68 4.69 6.82 6.47 8.72 7.69 ܮ௜௧ 60 Celsius 3.31 2.86 4.45 2.55 4.52 2.98 4.74 2.80 5.20 2.58 degree ௜௧ ܶ 60 Millimeter 373.87 113.74 335.78 84.35 260.55 80.70 273.40 109.74 346.65 82.89

ܨ௜௧ 60 Percent 2.06 8.86 2.99 5.65 5.52 6.61 -2.06 7.54 0.65 5.63 Source:ܥ௜௧ based on our survey and Statistical Yearbooks of Inner Mongolia

3 3.4 Results ) ୧୲

In this section, we first examine whether our data are appropriate for using the fixed effects model t -1.67 3.45 0.70 -1.18 -2.02 2.37 -0.62 2.12 -1.68 12.15 through statistical tests, involving the choice of the fixed effects or random effects model and a test for cross-sectional dependence. Moreover, we examine whether the variable of private use ( ) and its quadratic term ( ) are endogenous variables. This may be the case because the moment at whichܮ௜௧ actual ଶ private use was adoptedܮ௜௧ in a county could be decided by local households rather than by the (exogenous) formal government policy. Sparse grassland model ( Sparse grassland coefficient -569.16* 61.62*** 1,370.21 -181.93 -110.58** 0.16** -2.42 20,782.77** -81,822.8* 140,156.3*** 0.1255 300 Specifically, the Hausman test is employed to determine whether a fixed or a random effects model is ) more appropriate. The test results for all four models reject the null hypothesis at p<0.05, indicating that ୧୲ the fixed effects model is more appropriate. Cross–sectional dependence in panel–data models (also z -2.12 1.46 0.03 -0.81 -0.32 0.06 0.69 -2.86 2.59 13.68 called contemporaneous correlation) is then tested using the Pasaran CD test considering that our panel data includes many more cross-sectional units (60) than time periods (5). Test results show that residuals are not correlated across entities and hence no cross–sectional dependence occurs in any of the four models. Ultimately, the Durbin-Wu-Hausman test is employed to examine endogeneity of the variables of and . The test results reject the null hypothesis of exogeneity of in the total and dense ଶ grassland௜௧ models.௜௧ The null hypothesis of exogeneity of is rejected in the total,௜௧ dense and moderate modelModerate grassland ( coefficient -2,137.97** 84.24 116.18 -227.41 -31.48 0.01 4.87 -50,368.11*** 230,107.50** 287,707.00*** 0.0365 300 ܮ ܮ ܮ ଶ

grassland models. It indicates that endogeneity is found௜௧ in the total, dense and moderate grassland

ܮ 52 ) models, which therefore require the instrumental variables approach. In this regard, the number of years ୧୲ for which formal use rights have been assigned to households is strongly correlated with the number of z -1.78 -0.18 1.31 -1.44 2.37 -1.62 -1.38 -2.10 0.06 18.07 years for which actual private use has been adopted (the correlation coefficient is 0.825). Moreover, the implementation of this formal policy was not decided by the local government or households, but it followed the policy guidelines from higher level governments, which means the variable about formal use rights can be supposed to be exogenous. We therefore consider the number of years with formal use Table 3.5 Estimated results based on the fixed model effects rights and its quadratic term as the instrumental variables of and . The instrumental variables tests ( model Dense grassland coefficient -1,273.95* -6.85 4,097.36 -356.23 207.22** -0.17 -8.60 -33,047.87** 4,989.18 333,232.90*** 0.1548 300 ଶ further prove that they are valid and strong instrumental variables௜௧ for௜௧ and . After these tests, our

ܮ ܮ )

ଶ ୧୲ four models are run using fixed effects and based on instrumentalܮ ௜௧variablesܮ௜௧ in the models with endogeneity. The results are shown in table 3.5. z -3.28 2.24 1.51 -2.56 0.94 -0.42 -0.82 -2.99 1.81 29.27

Total grassland model ( Total grassland coefficient -3,252.50*** 116.66** 6,530.85 -879.51*** 113.83 -0.06 -7.05 -65,265.84*** 195,365.30* 749,546.10*** 0.1668 300

௜௧ ଶ ଶ ଶ ଶ ௜௧ ௜௧ ௜௧ ௜௧ ௜௧ ௜௧ ௜௧ ௜௧ ଶ ܶ ܶ ܨ ܨ ܴ ܮ ܮ ܥ ܥ Obs ܶܨ Constant Variables 51 at 5%; *significant at 10%; **significant ***significant at 1%.

of is test models. Ultimately,Durbin-Wu-Hausman the and hencenocross entities are notcorrelatedacross than units(60) cross-sectional many more data includes correlation) is thencalled tested contemporaneous effects modelisthe fixed moreappropriate.Cross resultsall four models The test for more appropriate. testis theHausman Specifically, employed todeter governmentformal policy. c acounty adopted in private usewas further prove that they are valid and strong instrumental variables for instrumentalvariables valid and strong further are thatthey prove resultsendogeneity. The are shown in table 3.5. with inthemodels oninstrumentalvariables and based effects fixed usingare run four models models, whichtherefore theinstrumentalvari require endogeneityIt that indicates grassland models. grassland hypothesis models. The null of of exogeneity rights astheinstrumentalvariablesof anditsquadratic term th We exogenous. tobe supposed be can use rights g level fromguidelines higher followed thepolicy implementation ofthisformal policy notdecide was adopted usehasbeen years actualprivate for which to assigned have been rights formalfor use which cross-sectional dependence. Moreover, we Moreover, ex dependence. cross-sectional through involving thechoice statistical tests, ofthe whether da examine In our this section, first we 3.4 Results quadratic term ( quadratic term ܮ ௜௧ and ܮ ଶ ௜௧ . The test results reject the null hypothesis of exogeneity of ofexogeneity thenullhypothesis reject . Thetestresults ܮ ଶ ௜௧ ) are endogenous variables. This may be the case because the moment at which actual actual which the moment at because case the be Thisare may variables. endogenous ) ould be decided by local households rather than by the (exogenous) byould bedecided householdsratherthanby the(exogenous) local amine whether the variable ofprivateuse( amine whether thevariable is found in the total, dense and moderate grassland moderate and total,dense foundinthe is ta are appropriate for using the fixed effects model appropriate effects forta fixed are model usingthe households is strongly correlated with the number of withthenumber correlated households isstrongly erefore consider the number of years with formal use use formalyears with of thenumber consider erefore 51 51 overnments, which means the variable about formal aboutformal thevariable whichmeans overnments, using considering thePasaran CDtest thatourpanel

fixed effects or random effects model and a test for effects modelanda test effects orrandom fixed (the correlation coefficient is 0.825). Moreover, the is0.825). Moreover, the (the correlationcoefficient sectional dependence inpanel –sectional dependence mine whether a fixed orarandommine whether modelis effects afixed reject the null hypothesis at p<0.05, indicating that atp<0.05,indicating reject nullhypothesis that the employed to examine endogeneity of the variables variables of the endogeneity examine employed to ables approach. In thisregard,yearsables approach. thenumber of sectional dependence occ –sectional dependence d by the local government or d households, bybut it local the time periods (5). Test results show that residuals residuals results showthat Test (5). time periods ܮ ଶ ௜௧ total, denseandmoderate is rejected inthe ܮ ௜௧ and ܮ ଶ ௜௧ ܮ . Theinstrumental variablestests ௜௧ and ܮ ܮ ௜௧ ଶ ௜௧ in the total and dense urs inany ofthefour . After these tests, our . Aftertests, our these data models (also–data models ܮ ௜௧ ) and its

Table 3.5 Estimated results based on the fixed effects model

Variables Total grassland model ( ) Dense grassland model ( ) Moderate grassland model ( ) Sparse grassland model ( ) coefficient z coefficient z coefficient z coefficient t ୧୲ ୧୲ ୧୲ ୧୲ -3,252.50*** -3.28 -1,273.95* -1.78 -2,137.97** -2.12 -569.16* -1.67 116.66** 2.24 -6.85 -0.18 84.24 1.46 61.62*** 3.45 ௜௧ ܮଶ 6,530.85 1.51 4,097.36 1.31 116.18 0.03 1,370.21 0.70 ௜௧ ܮ -879.51*** -2.56 -356.23 -1.44 -227.41 -0.81 -181.93 -1.18 ௜௧ ܶଶ 113.83 0.94 207.22** 2.37 -31.48 -0.32 -110.58** -2.02 ௜௧ ܶ -0.06 -0.42 -0.17 -1.62 0.01 0.06 0.16** 2.37 ௜௧ ܨଶ -7.05 -0.82 -8.60 -1.38 4.87 0.69 -2.42 -0.62 ௜௧ ܨ -65,265.84*** -2.99 -33,047.87** -2.10 -50,368.11*** -2.86 20,782.77** 2.12 ௜௧ ܶܨ 195,365.30* 1.81 4,989.18 0.06 230,107.50** 2.59 -81,822.8* -1.68 ௜௧ Constantܥଶ 749,546.10*** 29.27 333,232.90*** 18.07 287,707.00*** 13.68 140,156.3*** 12.15 ௜௧ ܥ 0.1668 0.1548 0.0365 0.1255 Obsଶ 300 300 300 300 *significantܴ at 10%; **significant at 5%; ***significant at 1%.

3 As listed in table 3.5, the coefficient of is significant and negative and the coefficient of 4 ଶ 3 is significant and positive in the total grasslan௜௧ d model. It indicates that a possible turning point௜௧ trend of total ܮ ܮ 2 exists at which the decrease in total grassland area may be reversed as the increase in the number grassland ha) 1 4 trend of dense of years with private use. Similarly in the sparse grassland model, the coefficient of is 0 grassland -1 0 2 4 6 8 10 12 14 16 18 20 22 24 266 significant and negative, and the coefficient of is significant and positive as well. Againܮ௜௧ the trend of moderate ଶ -2 grassland possibility of a turning point exists at which theܮ ௜௧decrease in sparse grassland area will reverse. -3 trend of sparse It is noted that the coefficients of are significant and negative in the dense and the moderate grassland areas (10 -4 grassland grassland models, but the coefficients௜௧ of are not significant. It indicates that when other -5 ܮ x= 4.6 x=14 ଶ -6 variables remain constant, the dense (moderaܮ௜௧ te) grassland area decreases by 1274 (2138) hectares if the period with private use increases by one year. In addition, the significant and Figure 3.2 Estimated changes in the areas of different grasslands as the increase in the number negative coefficient of indicates that temperature has a nonlinear relationship with the total of years with private use ଶ grassland area. The significantܶ௜௧ coefficients of show that precipitation is able to increase the dense grassland area significantly, but converseܨ௜௧ly reduce the sparse grassland area. And sparse In figure 3.2, we employ the significant regression coefficients of and to particularly ଶ grassland area has a nonlinear relationship with precipitation as presented by its significant depict the changes in the areas of different grasslands as the number ܮof௜௧ yearsܮ ௜୲with private use coefficient of . Ultimately, the growth of cultivated land ( ) has significant and negative increases. It is interpreted as the estimated effects of the land tenure reform on grassland ଶ condition while all of the other factors are controlled for. Specifically, the x-axis indicates the effects on the total,ܨ௜௧ dense and moderate grassland areas, and a ܥsignificant௜௧ and positive effect on the sparse grassland area. Nonlinear relationships are present in the total, moderate and sparse number of years with private use and the y-axis presents the areas of different grasslands. grassland models as shown by their significant coefficients of . Considering the longest duration with private use till 2008 was 26 years in our original dataset, ଶ we drawn the maximum year of the x-axis at 26. The figure shows that the dense and moderate ܥ௜௧ Comparing the above estimated model results with our expected results in table 3.3, the grassland areas present continuous decreases without reversion as their coefficients for are coefficients of significant variables in the models are almost consistent with our expectations, ଶ not significant in the model. And the moderate grassland area decreases more quicklyܮ ௜௧than but the coefficients of and in the sparse grassland model did not present opposite signs dense grassland. However, both sparse and total grassland areas demonstrate the predicted ଶ with that in the other three௜௧ models.௜௧ This difference indicates that the private use results in the ܮ ܮ upward trends after the initial decreases as their coefficients for and are significant. That decrease of sparse grassland area firstly and then increase, rather than increase from the ଶ is, sparse and total grassland areas decrease at first but then startܮ௜௧ to increaseܮ௜௧ after a certain beginning as we expected. A possible explanation could be that some sparse grassland degrades number of years with private use. The turning point is 4.6 years for the sparse grassland area to such a degree that the grass cover drops from 5-20% to less than 5%. Such severe levels of and 14 years for the total grassland area in our case. degradation would however not be picked up by the variable of sparse grassland because land areas with less than 5% grass cover are no longer counted as grassland areas. And then the As such, the figure indicates that the total grassland area decreases and is not able to recover to decrease of spares grassland area per se is offset by its increase caused by the degradation of the initial level until several decades after the start of the land tenure reform. Moreover, the grassland quality that the dense and moderate grasslands degrade into sparse grassland after continuing decrease in dense and moderate grassland areas indicates that the grasslands with several years of private use. higher grass cover will sustain degradation over time under privatisation. This trend is confirmed by the increase in the sparse grassland area while the reduction in the total grassland area. Subsequently, the increase in the total grassland area may stem from the increase in the sparse grassland area, rather than the improvement of grassland quantity. In short, the empirical

53 54

grassland area. The significant coefficients of coefficients grassland area. Thesignificant grassland models as shown by significant of grassland coefficients modelsasshown their the sparse grassland area. Nonlinear relationships on positive effect and significant a and areas, grassland and moderate total, dense on the effects but the coefficients of but the coefficients variablesinthem ofsignificant coefficients modelresultsestimated Comparing above the As listed in table 3.5, the coefficient of coefficient 3.5,the As listedintable

negative coefficient ofܶ negative coefficient private use increase hectares if the period with variables remain the constant, dense (modera several years of private use. years use. of private several grassland quality that the dense se is of per area grassland spares decrease of grass coverareareas with lessthan5% nolong by pickedwould howevernotbe up degradation to to such a less than grass that degree dropsfromSuch cover 5%. the levelssevere 5-20% of beginning A possiblegrassland degrades explanation couldbethatsomesparse expected. as we and th firstly area grassland sparse of decrease models.Thisdiffer with thatintheother three grassland models,butthe of coefficients It is noted that the coefficients of coefficients It thatthe isnoted grassland possibilitywill areareverse. atwhichthedecrease in sparse exists of aturningpoint coefficient of coefficient relationship with hasanonlinear grassland area converse but area significantly, grassland dense of years with private use. Similarly in the sparse grassland model, the coefficient of coefficient model,the yearsgrassland Similarly sparse of use. inthe withprivate ar intotalgrassland atwhichthedecrease exists turningpoint Ita possible grasslan model. that indicates d positiveinthetotal is significant and significant and negative, and the coefficient of and thecoefficient significant andnegative, ܨ ௜௧ ଶ . Ultimately, thegrowth ofcultivatedland( ܮ ௜௧ ௜௧ ଶ indicates thattemperature hasanonlinearrelationship withthe total and ܮ ଶ ௜௧ and moderate grasslands degrade grasslands degrade and moderate ܮ in the sparse grassland model did not present opposite signs opposite signs modeldidnotpresent grassland in thesparse ௜௧ are significant and negative in the dense and the moderate and themoderate dense negative inthe and significant are ܮ ௜௧ is significant and negati significant is ܮ ଶ ௜௧ odels are almost consistent with our expectations, It significant.are that when not other indicates fset by its increase caused by the degradation of by by caused its increase fset ܨ 53 53 ence indicates that the private use results in the results inthe use thattheprivate indicates ence ea may be reversed as the increase in the number inthenumber increase asthe may reversed ea be ܮ ௜௧ ly reduce the sparse grassland area. And sparse grassland sparse area.And reduce thesparse ly te) grassland area decreases by 1274(2138) grassland areadecreases te) s byyear. one In addi ଶ ௜௧ showthatprecipitation is abletoincrease the the variable of sparse grassland land variablebecause of sparse the er counted as grassland areas. And then the the And then grasslandareas. counted as er with our expected results in table 3.3, the in table 3.3, the results with our expected is significant andpositiveaswell. Again the en increase, rather than increase from the increase from than rather en increase, precipitation as presented by its significant precipitation by aspresented itssignificant are inthetotal, present moderate andsparse ܥ ܥ ௜௧ ௜௧ ଶ ) has significant and negative) has significant . ve and the coefficient of coefficient and the ve into sparse grassland after after grassland sparse into tion, thesignificant and ܮ ௜௧ ܮ is ଶ ௜௧

grassland areas present continuous decreases without reversion as their coefficients foras their reversion grassland decreasescontinuous without areas present atwe 26. The figure maximum drawnyear dense shows and the moderate that the of the x-axis Considering thelongestduration withprivateusetillyears 2008was 26 inour original dataset, y-ax years useandthe number of withprivate indicates the x-axis for. the Specifically, are controlled other factors condition whileallofthe sparse grassland rather area, than the improvemen theincrease areamay inthe from area. inthetotalgrassland Subsequently, stem the increase grassla inthesparse increase the confirmed by higher grass willsustaindegradati cover continuing indenseandmoderategrassl decrease until severalinitial level the As such, the figure indicates that the total grasslan case. inour yearsgrassland area forand 14 thetotal turningpoint years use.The withprivate number of at areasdecrease totalgrassland is, sparse and Ittheincreases. as estimate interpreted is gra different areas inthe of depict thechanges

In 3.2,we figure employ significant regression coefficients of the upward trendsaftertheinitiald grassland. to bothdense sparse However, and And the moderate model. not significant in the Figure 3.2 Estimated changes in the areas of different grasslands as the increase in the number the number Figure grasslands in 3.2 Estimatedareas of as different the changes in increase the grassland areas (104 ha) -6 -5 -4 -3 -2 -1 0 1 2 3 4 01 41 82 22 26 24 22 20 18 16 14 12 10 8 6 4 2 0 =4.6 x= decades afterthestartof ecreases for astheircoefficients of years with private use yearsof with privateuse x=14 d effects ofthelandtenure reform on grassland on overtimeunderprivatisa nd area while reduction in the the total grassland 54 54 sslands as the number of yearssslands asthenumber withprivateuse of tal grasslandareas dem first but then start to increase after a certain certain a after butthenstarttoincrease first grassland area decreases more quickly thangrassland decreases area is presents the areas of grasslands. different t of grassland quantity. Inshort, the empirical grassland quantity. of t d area decreases and is not able to recover to to and isnotableto recover d area decreases and areasindicatesthatthegrasslands with is 4.6 years for the sparse grassland area area grassland years sparse is4.6 for the land tenure reform. Moreover, the land tenurereform. Moreover, ܮ ௜௧ and 6 ܮ ௜௧ ܮ and onstrate the predicted thepredicted onstrate ଶ ௜௧ are significant.That tion. Thistrend is tion. grassland sparse trend of grassland moderate trend of grassland dense trend of grassland total trend of ܮ ଶ ௜୲ to particularly to ܮ ଶ ௜௧ are are

3 analysis confirms the negative effects of privatisation on grassland quantity and quality in the community on the basis of a community-based arrangement at low transaction costs. However, long term. more empirical evidence is needed on a large-scale to allow assessing the benefits of community-based governance over private use, taking into account the conditions of current 3.5 Discussion social, economic and policy incentives in the pastoral areas. In line with the results of Sneath (1998), Li et al. (2007) and Li and Huntsinger (2011), we find 3.6 Conclusions a negative impact of privatisation on the grassland condition. This may have discouraged herders to adopt private use, which could be an inherent explanation for the slow progress in This paper used quantitative analysis to investigate the relationship between grassland the implementation of the land tenure reform on the grasslands of China in recent decades. But privatisation and grassland degradation. First, a review was made of the controversial views on conversely, some scholars (see e.g. Li et al., 2007) have argued that the observed negative the impact of privatisation on the grassland condition. Next, we summarized the progress in the impacts of the privatisation of grassland use rights result from constraints in the supporting land tenure reform in the pastoral areas of China. The evolution of grassland degradation was conditions for a successful privatisation process, rather than from the privatisation per se. demonstrated in a typical pastoral area, using data for 60 counties in Inner Mongolia between Specifically, they point to a lack of resources for pastoral herders for instance to build fences 1985 and 2008. A fixed effects model was employed to disentangle the effect of the land tenure and to effectively protect privatised grasslands from overuse by outsiders. An interesting reform on the grassland condition while controlling for factors related to climate, agricultural avenue for further research would be to explore in more detail whether the so-called tragedy of activity and heterogeneity among counties. The model results show that private use causes the privatisation has its roots in privatised use rights per se or in its failed implementation in practice. reduction of the total grassland area and the degradation of dense and moderate grasslands into sparse grassland in the long run. In conclusion, our analysis suggests that the land tenure reform, Another challenge for future empirical research is to take into account the specific effects of namely the privatisation of grassland use rights, plays a significant role in the degradation of overgrazing and the impacts of ecological reconstruction projects that have been initiated to grassland quantity and quality in Inner Mongolia. In other words, we observe “a tragedy of protect the grassland condition in China since the 2000s (Li et al., 2014). The existing evidence privatisation” (Guelke, 2003; Li and Huntsinger, 2011), as opposed to the well-known “tragedy on the relationship between livestock production and grassland degradation is vague and of the commons”. controversial (Harris, 2010; Cao et al., 2013b). Moreover, Liu et al. (2016) has shown that the impact of ecological projects on grassland condition was offset to some extent by socio- economic and climate factors. While these elements were not directly investigated in this paper because of limitations in the data, they were approximated by including the variables of agricultural activities in the model.

Concerning the governance of China’s grasslands, instead of the private use by households, the community-based governance is fiercely advocated by pastoral specialists (e.g. Fernandez- Gimenez et al., 2012). They claim that community-based governance can facilitate the reciprocity of resource use among households, including capital investment and labour force use (e.g. Vetter, 2005; Li and Zhang, 2009; Cao et al., 2013b). Some empirical evidence exists to support community-based governance as well. For instance, Cao et al. (2013a) observed that the biomass, vegetation cover and species richness of grasslands are significantly higher under multi-household management than under single-household management in Maqu county, Gansu province. Ho (2001) showed that grasslands can be successfully managed by a village

55 56

Gansu province. Ho (2001) showed that grasslands can be successfully managed by a village village a by managed grasslands canbesuccessfully showed that Ho (2001) Gansu province. sing managementmulti-household thanunder richness andthe biomass,vegetation species cover as well. governance to supportcommunity-based Liuse (e.g. andZhang, Vetter, 2005; 2009;Caoetal.,2013b).Someempirical exists evidence in households, among use resource of reciprocity Gimenez etal.,2012).Theygovernance canfacilitatethe claimthatcommunity-based (e.g. Fernandez- advocatedby specialists pastoral governance is fiercely community-based grass ofChina’s governance Concerning the agricultural inthemodel. activities were oflimitationsinthedata,they because economic While and climate factors. ongrassland projects impact ofecological co controversial (Harris, 2010; Cao et al., 2013b). on the relationship between livestock production on therelationship betweenlivestockproduction inprotect thegrassland Chinasince th condition overgrazing of ecological and the impacts reconst empiricalresearch future for Another challenge privatisation hasitsrootsinprivatise would be to explore research further avenue for grasslandsand to effectively protect privatised resources of alack they pointto Specifically, perse. privatisation from the than rather process, privatisation conditions for successful a of theprivatisationgrimpacts Li (see etal e.g. conversely, somescholars gr the implementationoflandtenurereform onthe an could be use, which private to adopt herders on the gra a negative of privatisation impact etLiSneath (1998), results of In line with the 3.5 Discussion long term. analysisconfirms the effectsnegative of privatis

assland use rights resultassland from userights constraints inthesupporting d userights perseorinitsfa these elementswere notdirectly lands, instead of the private use by lands, insteadoftheprivateusehouseholds, ., 2007) that have the argued observed negative al. (2007) Li and and Huntsinger (2011), findwe 55 55 for pastoral herders for instance to build fences pastoral forfor herders to build instance fences ssland condition. This may have discouraged condition.Thismay havediscouraged ssland in more detail whether the so-called tragedy of tragedy of so-called in more detail whether the approximated by including the variables of by variables the including approximated Moreover,Liu etal.(2016) hasshownthat the e 2000s (Li evidence etal.,2014).Theexisting ndition was offset to some by extent ndition socio- inherent explanation for the slow progress in forexplanation the inherent le-household management in Maqu county, county, inMaqu management le-household ation ongrasslandation inthe and quality quantity is to take into of account thespecificeffects totake is from overuse Aninteresting by outsiders. For instance, Cao et al. (2013a) observed that observedet that al.For (2013a) instance, Cao cluding capital investment and labour force andlabourforce capital investment cluding of grasslands are significantly higher under ofgrasslands under higher are significantly ruction projects that have been initiated to beeninitiatedto ruction projectsthathave and grassland degradation is vague and asslands of China in recent decades. But But Chinainrecent decades. of asslands iled implementation in practice. iled implementationinpractice. investigated inthispaper of thecommons”. privatisation” (Guelk Inner Mongolia. grassland in andquality quantity namely theprivatisationofgrassla sparse grassland in the run.Inlong conclusion, our ar grassland reduction ofthetotal counties. Them among heterogeneity activity and reform on the grassland condition while controlling for agricultural related factors to climate, employed was effects model fixed 1985 and2008.A using pastoral area, typical demonstrated in a ofChin pastoralareas reformland tenure inthe on privatisation the impactof degradation. First, grassland privatisation and This paper used quantitative analysis to i 3.6 Conclusions social, economic in areas. and the policy pastoral incentives ofcurrent conditions the intoaccount use,taking private over governance community-based alar isneededon more empiricalevidence arra community-based a basisof community onthe

e, 2003; Li asopposedtothewell- and Huntsinger,2011), e, 2003; the grassland condition. Next, we summarized the progress in the the the progress in grassland we condition.Next, summarized the ea and the of degradation dense nd userights, plays roleinthedegradation of asignificant nvestigate grassland therelationship between a review was made of the controversial views on views controversial the made of was a review 56 56 data for 60 counties in Inner Mongolia between InnerMongolia between counties in for 60 data a. The evolution of grassland degradation The was evolution of a. ge-scale to allow assessing the benefits of benefits the assessing toallow ge-scale analysis suggests that the land tenure reform, landtenurereform, suggeststhatthe analysis odel results show that private use causes the the usecauses odel results showthatprivate ngement at low transaction costs. However, However, costs. transaction atlow ngement In other words, we observe “a tragedy of of “a tragedy In weobserve otherwords, to disentangle the effect of the land tenure tenure of the land effect to disentangle the and moderate grasslands into grasslands into and moderate known “tragedy

3 ACKNOWLEDGEMENTS: Chapter 4 Funding for this research was supported by grant from National Natural Sciences Foundation of China (71333013). The authors would like to thank Xiaoyi Wang, Dan Wang, Yulan Bu, Jiliang Hu, Yangjie Wang and Jie Zhou for their useful suggestions and help. We are also grateful to the anonymous reviewers for their good comments that contributed to improve the quality of this paper. How does land tenure reform impact upon pastoral livestock production? An empirical

study for Inner Mongolia, China

57 58

ACKNOWLEDGEMENTS: Chapter 4 Funding for this research was supported by grant from National Natural Sciences Foundation of China (71333013). The authors would like to thank Xiaoyi Wang, Dan Wang, Yulan Bu, Jiliang Hu, Yangjie Wang and Jie Zhou for their useful suggestions and help. We are also grateful to the anonymous reviewers for their good comments that contributed to improve the quality of this paper. How does land tenure reform impact upon pastoral livestock production? An empirical

study for Inner Mongolia, China

57 58

4. HOW DOES LAND TENURE REFORM IMPACT UPON PASTORAL LIVESTOCK 4.1 Introduction PRODUCTION? AN EMPIRICAL STUDY FOR INNER MONGOLIA, CHINA12 Land tenure reform in China is characterised by assigning long-term land use rights to ABSTRACT: How has land tenure reform affected livestock production in pastoral areas of individual households, involving cropland, grassland, and forest land (see e.g. Hu, 1997; Banks, China? This question is explored by estimating what impact assigning grassland use rights has 2003; Zhang et al., 2012a). According to the current Law of the People's Republic of China on on livestock production based on county-level data for Inner Mongolia between 1985 and 2008. the Contracting of Rural Land, use rights of grassland are contracted to individual households The timing of the introduction of the household-based assignment of grassland use rights for a duration of 30-50 years. At the end of the 1970s, land tenure reform was first implemented differed between counties, enabling a comparison of the effects of the land tenure reform. The through the Household Production Responsibility System in the crop farming areas of China changes in livestock production over time are examined by analysing data on changes in and was completed rapidly and successfully. Research showed that the assignment of individual livestock population and meat output. The descriptive analysis shows that livestock production cropland use rights gave farmer the incentive of improved agricultural production and spurred increased at a higher speed in the crop farming areas, but that the development of livestock the marketing of agricultural goods (see e.g. Ho, 1996; Hu, 1997; Krusekopf, 2002; Banks, productivity was faster in the pastoral areas. In the empirical analysis, we employed a fixed 2003). Based on this outcome, the central government of China continued land tenure reform effects model to disentangle the effects of land tenure reform on livestock production from in pastoral areas in the early 1980s and assigned grassland use rights as well as livestock factors related to market forces, grassland condition, technological development and property rights that had been owned by the communes to individual households. However, the environmental heterogeneity. The model results reveal that the implementation of land tenure assignment of grassland use rights has not been completed as successfully as that of cropland reform had significant and negative effects on the increase in livestock production, although in China (Liu et al., 2015). For instance, compared with the cropland in China, where total livestock production actually increased. It therefore appears that land reform is in itself household-based use rights were assigned overnight, the assignment of grassland use rights is unable to offset the impact of other factors that accelerate the increase in livestock production. still incomplete despite efforts for 30 years. By 2014, around 84% of grasslands had been Moreover, the constraining effect of land tenure reform on the increase in livestock production allocated to individual households in China, and the central government continues to emphasise decreases with the number of years for which land tenure reform has been implemented, and the need for clarifying grassland use rights for individual households (Ministry of Agriculture ultimately disappears. Remarkably, the constraining effect of land tenure reform is stronger on of China, 2015). In practice, in some areas of Inner Mongolia, only the grasslands for mowing the increase of livestock population than on that of meat output. This indicates that land tenure forages have been allocated to individual households while the grazing grasslands are still reform is beneficial in that it improves livestock productivity. In conclusion, land tenure reform, owned and used by all of the local households (Li et al., 2007). Scholars (e.g. Richard et al, namely the privatisation of grassland use rights, puts a ceiling on livestock production, which 2006; Li and Zhang, 2009; Li, 2012) have identified a number of possible barriers to the could be a possible reason as to why it has been difficult to implement the reform on grasslands. completion of the land tenure reform in the pastoral areas of China. This paper will explore However, the reform does prove to be beneficial in improving the livestock productivity of possible reasons by focussing on the aspect of pastoral livestock production. pastoral areas. Land tenure has been the concern of academia for a long time. The privatisation of grassland KEYWORDS: assignment of property rights, use rights, pastoral area, livestock population, property rights or use rights has been regarded as a panacea to avoid overgrazing and protect meat output, fixed effects model the grassland ecosystem, in order to avoid the situation termed 'the tragedy of the commons' (Hardin, 1968; McEvoy, 1987; Ybarra, 2009). Coase’s (1960) theorem of property rights also

argued that a clear assignment of property rights is a precondition for economically efficient resource allocation and environmental sustainability. The followers of the 'tragedy of the

commons' and property rights theories advocate that the grassland resource should be either 12 Paper by Min Liu, Jikun Huang, Liesbeth Dries, Wim Heijman, and Xueqin Zhu, submitted to a peer-reviewed journal. privatised or maintained as common land whilst clearly assigning rights of entry and use to

59 60

journal. model effects meat output,fixed KEYWORDS: assignment of propertyrights, use pastoral areas. prove the reform to be benefic does However, could as to why be a possible reason it has be namely the ofgrassland privatisation use rights reform in that it is improves beneficial livestoc population thanonthatofthe increase oflivestock meatoutput.Thisindicates thatlandtenure ultimately Remarkably, theconstraining isstronger disappears. oflandtenurereform on effect yearsdecreases withthenumberof for which tenur land of effect constraining the Moreover, that factors ofother impact the unable tooffset total livestockproductionIt actually increased. th on negativeeffects and reform hadsignificant reveal thattheimplementationoflandtenure Themodelresults environmental heterogeneity. grassl forces, relatedfactors tomarket effects modeltodisentangle theeffects of In thepastoralareas. fasterproductivity in was of livestock development but that the areas, farming crop in speed the higher a at increased livestock populationand meat output.Thedescriptiv changes over time are in livestock production a comparis enabling counties, between differed ofthehousehol The timing theintroduction of da on livestockproductionbasedcounty-level has rights use assigninggrassland impact bywhat estimating Thisquestionisexplored China? affected reform tenure ABSTRACT: Howhasland PRODUCTION? AN EMPIRICAL STUDY FOR INNER MONGOLIA, CHINA MONGOLIA, PRODUCTION? ANEMPIRICAL STUDYFORINNER REFORMIMPACT UPONPASTORALLIVESTOCK 4. HOWDOESLANDTENURE 12 Paper by Min Liu, Jikun Huang, Liesbeth Dries, Wim Heijman, and Xueqin Zhu, submitted to by Zhu, Xueqin Paper submitted and Heijman, Liesbeth Liu, Min Dries, Wim Jikun Huang,

and condition, development technological and and en difficult toimplementthereformen difficult ongrasslands. land tenure reform on livestock production from k productivity. In landtenurereform,k productivity. conclusion, 59 59 e reform on the increase in livestock production production in livestock theincrease e reform on accelerate the increase in livestock production. production. inlivestock increase the accelerate land tenure reform has been implemented, and implemented, tenure reformland has been on of the effects of the land tenure reform. The tenurereform. The theland of on oftheeffects , puts a ceiling on livestock production, which production,which onlivestock aceiling , puts ial inimprovingial thelivestockproductivity of the increase in livestock production, although inlivestockproduction, increase the ta for Inner Mongolia between 1985 and 2008. Inner 1985and2008. for Mongolia ta between the empirical analysis, we employed a fixed employed afixed we analysis, empirical the examined examined bychanges data in analysing on d-based assignment of grassland use rights rights grasslandassignment use of d-based erefore appears thatlandreformerefore appears is in itself rights, pastoral area, livestock population, livestock area, pastoral rights, livestock production in pastoral areas of of pastoralareas livestockproductionin e analysis shows that livestock production 12

a peer-reviewed privatised ormaintainedascommon landwhilst clea either shouldbe grassland resource the commons' that and rightsproperty advocate theories environmental sustainabi resource allocationand assignmentclearargued of property that a rights 1987; Ybarra,(Hardin, 1968;McEvoy, the inordertoavoid grassland ecosystem, the has been regarded or use rights rights property f Land theconcernofacademia hasbeen tenure livestockproduction. ofpastoral focussingpossible reasonsby ontheaspect completion ofthelandtenure reform inthepastora LiLi,2006; 2009; andZhang, 2012)haveiden owned andusedby allofthelocal households stillwhile the grasslandsgrazing are individual households to allocated forages been have InnerIn in some areas Mongolia,of China, 2015). forof onlygrasslands mowingpractice, the grassland userightsthe needfor for clarifying individualhouseholds(Ministry ofAgriculture allocated toindividualhouseholdsinChina,a still incomplete despite effortsyears. for 30 overni were assigned rights household-based use co instance, For in China (Liu et al., 2015). assignmentgrassland userights of hasnotbeen property rights that had been owned by the 1980sandassigned in pastoralareastheearly g central this outcome,2003). Basedthe on the marketing goods ofagricultural (seee. farmer gave theincent cropland use rights Rese and successfully. rapidly and wascompleted through the HouseholdProduction Responsibility th end of the years. At for of30-50 aduration gra rightsLand, of Rural use the Contracting of Law of China2003; on Zhang Republic et According al., 2012a). current the People's to the of individual households,involvinggrassl cropland, Land land use rights tenure reform in China is characterised toby assigning long-term 4.1 Introduction

2009). Coase’s (1960) theorem of pr theorem of (1960) Coase’s 2009). ive ofimproved agricultural andspurred production communes to individual households. However, the toindividualhouseholds.However,the communes g. 1996; Hu, 1997; Krusekopf, Ho, 2002; Banks, overnment of China continued land tenure reform landtenurereform ofChinacontinued overnment e 1970s, land tenure reform was first implemented reform wasfirstimplemented landtenure e 1970s, nd the central government continues toemphasise government continues nd the central By 2014,around84%ofgrasslands By hadbeen 60 60 situation termed 'the tragedy of the commons' the 'the tragedyof termed situation ssland are contracted to individual households toindividual contracted are ssland as a panacea to avoid overgrazing and protect andprotect toavoidovergrazing a panacea as mpared with the cropland in China, where where croplandinChina, mpared withthe (Li et al., 2007). Scholars (e.g. Richard et al, or along time.Thepri or completed as successfully as that of cropland as that of cropland as successfully completed and, and forest land (see e.g. Hu, 1997; Banks, Hu,1997;Banks, e.g. andforest land(see and, ght, theassignment ofgrasslanduserights is arch showed that the assignment ofindividualassignment showedthatthe arch isaprecondition for economically efficient tified anumberofpossiblebarriers tothe lity. The followers of the 'tragedy of the 'tragedy ofthe lity. followersofthe The System ofChina areas inthecropfarming grassland use rights as well as livestock grassland well userights as aslivestock l areas of China. This rly assigning rights of entry and useto of entry assigningrly rights vatisation of grassland paper will explore paperwillexplore operty rights also operty rights

4 promote the sustainable use of grassland through exclusiveness. Following this conventional academic arguments and government reports on the results of the privatisation of grassland, wisdom, most governments in the world have assigned grassland property rights or use rights there is a lack of quantitative studies, especially based on large-scale areas and long-term in an attempt to ensure the sustainable development of grasslands (Li et al., 2014). In line with observations (e.g. Li and Huntsinger, 2011; Yu and Farrell, 2013; Conte, 2015). Moreover, the observations about the effects of privatisation of cropland, the central government of China existing literature is short of empirical analysis targeting the impact of grassland privatisation believes that the assignment of grassland use rights provides land users with incentives to graze on livestock production, although some studies are concerned with ecological effectiveness. animals within the carrying capacity of the land, as well as to increase land investment in order Nevertheless, animal husbandry provides livelihoods to millions of people in pastoral areas and to conserve their own grasslands (Banks, 2003). has the potential capacity to meet the rapidly increasing global demand for livestock products which is stimulated by growing populations, urbanisation and rising disposable incomes (FAO, However, the findings on the effects of privatising grassland resources are complex (e.g. see Li 2015). In this regard, livestock production in pastoral areas is a very significant area and one and Zhang, 2009). During recent years, some scholars have expressed concern that the which deserves being paid close attention to. traditional common-use system and nomadism have been replaced by private use and settlement due to grassland privatisation (Banks et al., 2003). This transformation of traditional In the following section, we first describe the land tenure reform and the current livestock pastoralism has impeded mobility and flexible grazing (Fernandez-Gimenez, 2000; Wang et al., production in the research region. Next, in section 3, the data collection method is illustrated 2013) and limited herders’ access to emergency pastures and other key productive resources and a descriptive analysis based on the collected data is used to elaborate on the changes in (Niamir-Fuller and Turner, 1999; Fernandez-Gimenez, 2002). This, in turn, has increased livestock production in the pastoral areas compared to crop farming areas. In sections 4 and 5, feeding costs and reduced herders’ ability to withstand natural risks (e.g. Li et al., 2007), we present the empirical model used in the study and the model results on how land tenure causing constraints on the development of animal husbandry in pastoral areas. Some voices reform and other potential factors affect livestock production in the pastoral areas. In section 6, have also stated that grassland privatisation reduces the amount of land available for livestock we discuss the model results and their underlying reasons. We conclude this paper with remarks grazing, further leading to a reduction in the number of livestock that an individual can on the effects of grassland privatisation on livestock production in pastoral areas. potentially own, ultimately resulting in poverty (Mwangi, 2007). In addition, research has 4.2 Research region shown that the carrying capacity of grassland is reduced due to decreased access to heterogeneous landscapes (Boone and Hobbs, 2004; Hobbs et al., 2008; Boone et al., 2005). As China has around 400 million hectares of grassland, accounting for nearly 40% of its total such, the total population of livestock that could be supported by a grassland ecosystem is territory, this being the second largest area of grassland in the world after Australia (Hua and predicted to decline as a result of the spatial and social boundaries stemming from privatisation. Squires, 2015). Inner Mongolia, a province located in the arid and semi-arid areas of northern China, has 118.3 million hectares of land; in 2014, its permanent population was 25 million. It In fact, much of the growth in animal supply has been coming from rapidly expanding intensive accounts for 21.7% of China’s permanent area of grasslands. Approximately 67% of the total breeding systems rather than through traditional grazing systems in pastoral areas (FAO, 2015). land area in Inner Mongolia is classified as grassland, the majority of which can be sub- Similarly, the main production areas for livestock products in China have experienced a classified as temperate grassland (Angerer et al., 2008). Inner Mongolia plays an important role geographic shift from being pastoral areas to being crop farming areas (Li et al., 2008). The in the supply of animal products as well as in the ecosystem of China due to its extensive market share of livestock products from grazing systems is decreasing compared to that of crop- grasslands. In the pastoral areas populated by Mongolians, the vast majority of local people livestock mixed or industrialised systems (Squires et al., 2009). This trend is suggested to be maintain their livelihoods through grazing their livestock on the grasslands (Angerer et al., attributed to the reduction in grass yields due to grassland degradation, and raises environmental 2008). Inner Mongolia was one of the first regions in which land tenure reform was concern about the ecosystem of permanent grassland (Li et al., 2008; Squires et al., 2009). implemented on grasslands in China because of its crucial position in China’s grassland Besides these possible factors, we wish to ascertain whether the privatisation of grasslands has resource and livestock production (Li and Huntsinger, 2011). played a role in the changes in livestock production of pastoral areas. Despite an abundance of

61 62

played a role in the changes inlivestockproduct changes role inthe played a Besides whether ofgrasslands thesepossiblefactors,we theprivatisation has wishtoascertain grass ofpermanent ecosystem aboutthe concern yields to in grassdue attributed to the reduction livestock mixed orindustrialised systems (Squires grazingmarketproducts oftems from livestock is decreasing share to thatcompared of crop- sys geographic to shiftfrom being pastoral areas live areasfor Similarly, themainproduction grazing throughbreedingareas(FAO, traditional 2015). systems systems in than pastoral rather growth in Inanimal supplyfact, much of the ha apredicted of result the spatial as to decline such, thetotalpopulationoflivestockthatcoul and Hobbs,2004; (Boone landscapes heterogeneous grasslan shown thatthecarrying of capacity inpovert potentiallyresulting own,ultimately grazing,further to leading a reduction in th have also stated privatisation that grassland re anim onthedevelopmentof causing constraints feeding costs and reduced herders’ ability to 1999;Fernandez-G (Niamir-Fuller andTurner, accesstoe herders’ 2013) andlimited mobility pastoralism hasimpeded et al (Banks due tograssland privatisation and nomadismhave traditional common-usesystem years, some scholars have expre recent 2009).During and Zhang, theeffectsofprivatisin on However, thefindings grasslands to conserve (Banks, theirown 2003). theland,as animals withinthecarrying of capacity grassland ofbelieves that righ the assignment use observations abouttheeffects ofprivatisation in an attempt sustainable devel to ensure the wisdom, mostgovernments worldhaveas inthe promote thesustainableuseofgrassland through

and flexible grazing (Fernandez-Gimenez, 2000; Wanget al.,grazing (Fernandez-Gimenez, and flexible mergency pastures andotherkey pro mergency and social boundaries stemming from privatisation. from privatisation. boundariesstemming and social opment of grasslands (Li et al., 2014). Inal., 2014). ofgrasslands linewith (Liopment et duces theamountoflandavailablefor livestock 61 61 grassland degradation, and raises environmental raises degradation,grassland and being crop farming areas (Li etal.,2008).The being areas cropfarming s been coming from rapidly expanding intensive intensive expanding rapidly coming from s been e number of livestock that an individual can can an individual livestocknumber that of e ., 2003). This transformation oftraditional 2003). Thistransformation ., stock a products have in China experienced withstand naturalrisks(e.g.Li etal.,2007), of cropland, the central government ofChina government cropland,thecentral of y (Mwangi, 2007). In addition, research has has In research y addition, 2007). (Mwangi, ion of pastoral areas. Despite an abundance of of anabundance areas.Despite ion ofpastoral g grassland resources are complex (e.g. see Li (e.g. see complex are grassland resourcesg ts provides land users with incentives tograze provideslanduserswith ts signedgrassland rights property oruserights imenez, 2002). This, in turn, has increased imenez, 2002).This,inturn,hasincreased d be supported by a grassland ecosystem is grassland is ecosystem a besupportedby d al husbandry in pastoral areas. Some voices areas. in pastoral Some voices husbandry al d isreduced duetodecreased to access land (Li et al., 2008; Squires et al., 2009). et al., 2008;land Squires (Li et well as to increase land investment in order exclusiveness. Followingexclusiveness. this conventional been replaced by private use and settlement and settlement use private by replaced been et al.,2009).This trendissuggested tobe Hobbs et al., 2008; Boone et al., 2005). As etal.,2005).As Hobbsetal.,2008;Boone ssed concern that the concernthatthe ssed ductive resources ductive resources resource and livestock production (Li and Huntsinger, 2011). resourceand Huntsinger, andlivestockproduction(Li 2011). grasslands inChinabecaus implemented on reform was tenure regions land inwhich first the Inner of was Mongolia one 2008). maintain theirlivelihoods throughgrazing theirli by populated areas In thepastoral grasslands. in thesupplyproducts aswell asin of animal et grassland (Angerer classified as temperate land areaInner in Mongolia is classified as areaof ofChina’spermanent accounts for 21.7% land;in hectares of China, has118.3million locate Innera province Mongolia, Squires, 2015). of area largest second territory, this the being total ofits 40% nearly for accounting grassland, of hectares million China hasaround400 region 4.2 Research privatisationon grassland on theeffects of we discussthemodelresults andtheirunderlying affect livestock factors potential reformand other we modelusedin present the theempirical livestockareas production compared in the pastoral basedonthecollected and adescriptiveanalysis production intheresearch region. insect Next, In thefollowing we firstdescribe section, close attentionwhich to. being paid deserves In2015). productioninpastora thisregard, livestock urba growing populations, which isstimulatedby tomeet therapidlygloballivestock products increasing for demand has thepotentialcapacity animalhusbandryNevertheless, provides livelihoods ar although somestudies on livestockproduction, literatureisshortexisting ofempiricalanalysis the Moreover, Conte,2015). Farrell, 2013; and Huntsinger, 2011;Yu and Liobservations (e.g. there is a lack of quantitative studies, espec grassland, of theprivatisation of theresults reports on government and arguments academic

livestock production in pastoral areas. areas. livestock productioninpastoral the land tenure reform and the current livestock al., 2008). Inner Mongoliaplaysal., 2008). an important role 2014, its permanent population was 25 million. It million. populationwas 25 itspermanent 2014, study andthemodelresults on howlandtenure grassland, the majority of which canbesub- which grassland, themajority of ially based on large-scale areas and long-term andially long-term based areas on large-scale 62 62 e of its crucial position in China’s grassland China’s positionin itscrucial e of grassland in the world after Australia (Hua and Australia (Hua after grassland intheworld targeting theimpactof Mongolians, the vast majority of local people people local vast majority of the Mongolians, ion 3,thedatacollection method isillustrated theecosystem ofChinaduetoitsextensive nisation and rising disposable incomes (FAO, (FAO, incomes disposable andrising nisation reasons. We concludeth We reasons. data is used to elaborate on the changes in production in the pastoral areas. In section 6, areas. production in the pastoral d in the arid and semi-arid areas of northern of areas and semi-arid arid d in the grasslands. Approximat e concerned with ecological effectiveness. effectiveness. withecological concerned e vestock onthegrasslands (Angerer etal., to crop farming areas. In sections 4 and 5, Inand 5, farming sections4 to crop areas. to millions ofpeoplein pastoral areasand l areas is a very significant area and one andone area significant a very is l areas grassland privatisationgrassland is paper with remarks ely thetotal 67%of

4 According to the current administrative divisions of Inner Mongolia, there are 102 counties, the majority of the pastoral areas in the 1980s. Particularly, the use of grasslands merely including 33 pastoral counties and 21 semi-pastoral counties. The remaining 48 counties are indicated a rough direction and position to herders, but local herders and their animals still dominated by crop farming or urban districts. Pastoral counties are characterised by traditional roamed wherever they preferred. Hinton (1990) called this phenomenon privately controlled grazing systems and permanent grassland is the dominant land type. In semi-pastoral counties, stock of animals on publicly owned lands in Inner Mongolia. The result was an uncontrolled both permanent grassland and cropland are the dominant land types. Mixed crop-livestock scramble for wherever forage existed, which amounted to a general attack on a range of systems also exist, where intensive animal husbandry and cropping both occur (Waldron et al., vegetation (Hu, 1997). Subsequently, the assignment of grassland use rights to individual 2010). The majority of the pastoral areas of Inner Mongolia are found in the pastoral counties households was strengthened by the Two Rights and One System policy (Shuang Quan Yi Zhi) as well as semi-pastoral counties. Figure 4.1 presents the location of Inner Mongolia in China in 1996 (Bureau of Animal husbandry of Inner Mongolia, 2000; Li and Zhang, 2009). This and its three types of counties. stimulated the progress of grassland privatisation greatly and is regarded as the second round of land tenure reform in the pastoral areas of Inner Mongolia (Yang, 2007). In recent years, growing attention has been paid to ensure the long-term stability of grassland use rights to individual households, such as confirming grassland plots, areas and contracts (Li, 2012). In 2015, local governments in Inner Mongolia are still working on clarifying the boundaries of grassland use rights for each household and issuing certificates (Bureau of Animal husbandry of Inner Mongolia, 2015).

Compared with crop farming areas, the implementation of the land tenure reform on grasslands is relatively complex and has taken a long time. More specifically, the grasslands have been subject to several different types of ownership (Yu and Farrell, 2013) and correspondingly, various types of land use have arisen since the start of the land tenure reform in Inner Mongolia. Table 4.1 provides a typology of the land tenure reform on the changes in formal use rights and actual types of grassland use, which is in line with the investigation of the Tibetan plateau by Banks et al. (2003).

Figure 4.1 Inner Mongolia and its 102 counties

4.2.1 Land tenure reform in the pastoral areas of Inner Mongolia

Land tenure reform started in Inner Mongolia with the Double Contracts System (Cao Xu Shuang Cheng Bao) in 1982. This system aimed at assigning to individual households the livestock and grasslands that were managed by the communes during the collectivist period of China. Meanwhile, the users (individual households) of the grasslands are regulated in order to control their livestock numbers according to a determinant stocking rate with the aim of avoiding overuse of grasslands (Ho, 2000). The assignment of property rights over livestock was completed promptly but grasslands were not strictly assigned to individual households in

63 64

was completed promptly but grasslands werenot but promptly was completed rights livestock property over assignment of The (Ho, 2000). grasslands avoiding of overuse control their livestock numbers according to a China. Meanwhile,theusers(individualhouseholds that werelivestock andgrasslands by managed Shuang Cheng Bao)in1982.Thissystem aimed Inner in Land Mongolia reform started tenure InnerMongolia areas in of thepastoral 4.2.1 Landtenurereform and its three types of counties. counties. of and itsthreetypes counties.Figureas well assemi-pastoral 4.1 Inne of areas pastoral the of 2010). Themajority al., et (Waldron occur cropping both and where intensiveanimal husbandry systemsalso exist, the are cropland grassland and both permanent counties, In landtype.grassland semi-pastoral isthedominant grazing and permanent systems dominated bycrop farming by orurban countiesare districts.Pastoral characterised traditional and 21 includingsemi-pas 33 pastoral counties According administrative tothecurrent divi

Figure 4.1 Inner Mongolia and its 102 counties FigureInner andits102counties Mongolia 4.1 sions of Innersions of Mongolia, there are 102 counties, presents thelocation ofInner Mongolia inChina 63 63 the communes during the collectivist period of communesduring thecollectivist the with the Double Contracts System (Cao Xu theDoubleContractsSystem (CaoXu with toral counties. The remaining 48 counties are r Mongolia are found in the pastoral counties counties pastoral found inthe Mongolia are r dominant land types. Mixed crop-livestock crop-livestock Mixed dominant landtypes. strictly assigned to individual householdsin assignedstrictly toindividual determinant stocking withthe aimof rate ) ofthegrasslands are inorderto regulated at assigning toindividualhouseholdsthe

al. (2003). Banks et linegrassland which is in use, of actual types ofthelandtenurerefor Table 4.1providesatypology Inner Mongolia. reform in ofthelandtenure sincethestart have arisen oflanduse various types correspondingly, and Farrell, 2013) and (Yu ofownership types subject toseveral different a time. long has taken and is relatively complex Compared withcropfarming areas,theimp Innerof Mongolia, 2015). grassland use rights householdand for each of on clarifying theboundaries Inner stillworking in 2015, local Mongoliaare governments individual households,such asconfirminggrassl growing attention has been paid to ensure the years, Inrecent 2007). (Yang, Inner Mongolia of areas the pastoral reform in of landtenure stimulated theprogress grassl of This Zhang, 2009). and Li InnerMongolia, 2000; husbandry of Animal of in 1996(Bureau byhouseholds was One strengthened System and theTwoRights Zhi) policy(Shuang Yi Quan Subsequently vegetation (Hu, 1997). scramble whichamounted toageneral forage for onarange wherever existed, of attack I stock ofanimalsonpublicly landsin owned roamedHinton (1990 they wherever preferred. indicated arough directionandpositiontoherde thein themajority theareas 1980s. pastoral of

and privatisation greatlyand is , the assignment of grassland use rights ofgrassland use toindividual , theassignment lementation of the land tenure reformgrasslands on issuing (Bureau ofAnimalhusbandry certificates 64 64 with the investigation of the Tibetan plateau by by Tibetanplateau the of theinvestigation with nner Mongolia. The result was an uncontrolled Mongolia. resultnner an uncontrolled was The ) called this phenomenon privately controlled long-term stabilitygrassland of userights to and plots, areas andIn (Li, contracts 2012). More specifically, the grasslands have been been grasslands have the More specifically, rs, butlocal herders andtheir animalsstill Particularly, the use of grasslandsParticularly, merely the use of m on the changes in formal use rights and userightsm onthechanges informal and regarded the second round as

4 Table 4.1 Typology of land tenure reform on grasslands of Inner Mongolia of cattle were raised in Inner Mongolia, and 8% of China's total beef output was produced there, in both cases ranking second in China. Figures 4.2 and 4.3 present the livestock production of Actual grassland use Formal use rights owned by Inner Mongolia from 1979 to 2013, based on data from the China Statistical Yearbook (Zhong Private use Joint use Common use Guo Tong Ji Nian Jian). Livestock production is interpreted by the livestock population and Individual households + + + meat output. Sheep and cattle are the dominant animals being raised in Inner Mongolia (Zhang A group of households na + + et al., 2012b). We therefore employ the population of sheep and cattle to represent the livestock Collective/administrative village na na + Source: adapted from “formal and de facto grassland management units” (Banks et al., 2003) population, and the outputs of mutton and beef for meat output. na indicates not available Figure 4.2 shows that the population of sheep decreased in 1982 and then generally presented As presented in Table 4.1, with the implementation of land tenure reform, three types of an increasing trend until 1999, apart from a decrease during 1991-1993. It experienced a sharp ownership of formal grassland use rights have arisen, namely individual household ownership, increase from 2001 to 2004 after a decrease from 1999 to 2001, and remained relatively steady ownership by a group of households and collective ownership. There are also three types of between 2005 and 2013. The population of cattle saw slight fluctuations but remained almost actual grassland use, these being private use by individual households, joint use by a group of unchanged between 1979 and 2000. It reached the lowest point in 2001 after a slight decrease households and common use by all of the villagers. When formal use rights are owned by an from 1999 and then experienced an increase between 2002 and 2010, followed by a slight administrative or natural village, the only type of grassland use is common use. This arose decrease until 2013. It is evident that the number of sheep exceeded cattle numbers by a wide primarily before or at the beginning of the implementation of the land tenure reform when the margin during the whole period. grassland rights had not yet been assigned. If the formal use rights are owned by a group of households, the land use is either common or joint. This situation mostly occurs in areas where 104 heads 6000 the grassland resource is too scarce to be assigned to individual households, or where grassland 5500 use rights have not been assigned. Private use, joint use and common use exist simultaneously 5000 4500 when the formal use rights are owned by individual households. The intention of the land tenure 4000 3500 sheep population reform in the pastoral areas, however, is to achieve full private use when use rights are assigned 3000 2500 cattle population to households. This reflects the gap between the household-based assignment of formal use 2000 1500 rights and the actual adoption of private use. In practice, the specific timing and extent of 1000 conducting the land tenure reform13, including the changes in formal use rights and use patterns, 500 0 differed among counties in Inner Mongolia. 13 1 1979 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 2007 2009 2011 2013 4.2.2 Livestock production in Inner Mongolia Figure 4.2 Livestock population of Inner Mongolian from 1979 to 2013 Inner Mongolia is one of the main production regions for animal products in China. In 2013, it Figure 4.3 depicts Inner Mongolia's mutton and beef outputs from 1979 to 2013. Until 2002, accounted for 18% of the population of sheep14 of China, ranking first out of all Chinese the outputs of mutton and beef experienced fluctuating and slow increases. The output of provinces. The mutton output accounted for 22% of the total meat output in China. Six per cent mutton and beef increased sharply between 2002 and 2005, after which the output of mutton

13In the reminder of the paper, the land tenure reform means the grassland use rights are assigned to individual remained relatively steady, while beef continued to increase slightly between 2006 and 2013. households. In general, mutton output exceeded beef output during the whole period. Mutton output in 2013 14 The term 'sheep' in this paper includes both sheep and goats, in accordance with the China Statistical Yearbook.

65 66

households. households. conducting the conductingland tenure reform the ofextent and timing In practice, thespecific ofprivate use. adoption rights actual andthe to households.Thisreflects thegap betw however,reform is to rightsachieve fullprivateusewhen in the pastoral areas, use assigned are landtenure households.Theintentionofthe by individual areowned rights use formal when the use, assigned.use rights Private not been have istooscarcetobeassignedthe grassland toindividualhouseholds, resource orwheregrassland households, thelanduseiseithercommonorjoint.This situation mostly in areas where occurs grassland rights had not yetIf assigned. been beginningprimarily atthe before oftheim or administrative village, only ornatural type the households andcommonuseby are formal allofthevillagers. When owned userights by an by use private being grassland use,these actual collect householdsand group of a ownership by ownership offormal userights grassland havearisen, namely individualhouseholdownership, withtheimplement 4.1, inTable As presented na indicates not available managementgrassland units” anddeSource: facto from adapted “formal 14 13 accounted for 18% of the populationofsheep accounted for 18%ofthe regionsInInnerthe mainproduction foranimalproductsinChina. Mongolia isoneof 2013,it in production 4.2.2 Livestock Mongolia Inner Inner in differedamong counties Mongolia. provinces. The mutton output accounted for 22% of th accountedfor22%of output provinces. Themutton Yearbook. Collective/administrative village Collective/administrative village A groupofhouseholds Individual households by rightsFormal use owned In the reminder of the the paper, reformmeans the grassland use land tenure rights are assigned to individual The term 'sheep' termThe includes this in paper both sheep Table 4.1 Typology of land tenure reform on grasslands of Inner Mongolia Inner of Mongolia grasslands tenurereformTable 4.1Typology on ofland

13 , including the changes in formal use rights and, including use patterns, in rights formal the changes use een the household-based assignment offormal use assignment een thehousehold-based na na + Private use and goats, in accordance with Statistical in and goats, accordance China the plementation of the land tenure reform when the of plementation the formal userights owned the group by are of a 65 65 joint use and common use exist simultaneously use and common use exist joint individual households, joint use by a group of a households,jointuseby individual of grassland use is common use. This arose use.Thisarose iscommon grassland use of 14 ive ownership. There are also three types of of are also three types ownership. There ive ation of land tenure reform, three types of types of three oflandtenurereform, ation rankingof all of China, out Chinesefirst e total meat output in e total meatoutput percentChina. Six Actual grassland use grasslandActual use na + + use Joint (Banks et al., 2003) al., 2003) et (Banks + + + Common use In general, mutton output exceeded beefoutput In outputexceeded general, mutton and 2013. 2006 between increase slightly continued to while beef relativelyremained steady, sharply mutton andbeefincreased between fluc the outputsofmuttonand beefexperienced FigureInnerMongolia's 4.3depicts muttonandbeefoutputsfrom 1979to2013.Until2002, margin during the whole period. whole margin the during It 2013. decrease until isevidentthatthenumbe between2002and2010,followedbyfroma slight anincrease 1999andthenexperienced lo the reached It and2000. 1979 between unchanged sawslig The populationofcattle and 2013. between 2005 fr a decrease after from 2001to2004 increase a sharp experienced It during 1991-1993. decrease from until1999,apart a trend an increasing generally presented in1982andthen populationofsheepdecreased Figure 4.2showsthatthe population, andtheoutputsofmuttonbeefformeat output. et al., 2012b). therefore We employ population of the animals cattlearethedominant meat output.Sheepand Livestock production NianJian). Guo Tong Ji fromInner theChinaStatisticalYearbook(Zhong 1979to2013,basedondata Mongolia from productionof inChina.Figures and4.3present thelivestock in bothcasesranking second 4.2 Innerof cattlewere in Mongolia, raised and8%

10 1000 1500 2000 2500 3000 3500 4000 4500 5000 5500 6000 500 4 heads 0 1979 Figure 4.2 Livestock population of Inner Mongolian from 1979 to 2013 InnerFigureLivestock of from 4.2 population 1979to 2013 Mongolian 1981 1983 1985 1987 1989 1991 1993 1995 1997 2002 and 2005, after which the output ofmutton output and 2005,afterwhichthe 2002 om 1999 to 2001, and remained relatively steady relatively1999 to2001,andremained steady om 66 66

during the whole period. Mutton output in 2013 during thewholeperiod.Mutton 1999 of China's of totalbeefoutputwas producedthere, is interpreted by the livestock population and livestockpopulationand byis interpreted the 2001 numberswide by cattle a r of sheep exceeded tuating and slow increases. The output of tuating Theoutputof andslowincreases. west point in 2001 after a slight decrease decrease a slight after west pointin2001 2003 cattle torepresent thelivestock sheep and being raised in Inner in Mongolia beingraised (Zhang 2005 ht fluctuations but remained almost butremained almost ht fluctuations 2007 2009 2011 2013131 cattle population sheep population

4 was around 14 times higher than in 1979, while beef output was 19 times higher in 2013 than within China have been threatened by Henan and Shandong provinces, which are the dominant in 1979. crop farming areas of China where intensive animal husbandry systems have developed in recent years (Su, 2010). Furthermore, a number of ecological projects for grassland conservation and the promotion of non-farming industries have been introduced in Inner 100 90 Mongolia and have impacted upon the development of animal husbandry in the region (Squires 80 et al., 2009). As a result, the share represented by animal husbandry in the gross domestic 70 mutton output product of Inner Mongolia has decreased. Moreover, since 2001, the provincial government of 60 beef output 50 Inner Mongolia has laid emphasis on the development of livestock production in its crop 40 farming areas instead of in the pastoral areas. In 2007, 70% of livestock production in Inner 30 20 Mongolia was located in crop farming areas (Su, 2010). This geographic shift from pastoral 10 areas to the crop farming areas is discussed in the following section. 0 4.3 Data description 1979 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 2007 2009 2011 2013

4.3.1 Data collection Figure 4.3 Mutton and beef outputs of Inner Mongolia from 1979 to 2013 We aim to estimate the impact of land tenure reform on pastoral livestock production. The Figure 4.4 illustrates the development of livestock productivity in Inner Mongolia from 1979 pastoral areas are therefore distinguished from the crop farming areas for the purposes of data to 2013. The livestock productivity for both sheep and cattle has increased. The increase of collection. Given that the timing and extent of the implementation of land tenure reform differed productivity for cattle experienced obvious fluctuations, while the productivity for sheep saw a between counties, our empirical study of the effects of the reform on livestock production is slight and steady increase. It is evident that the output per cattle increased faster than the output conducted based on county-level data. The other factors that are controlled for their potential per sheep. impacts on the pastoral livestock production are grassland condition, market forces, technological development and environmental heterogeneity (Li et al., 2008; Tessema et al., ton/head 0.09 2014). We therefore collected three types of data at the county-level in Inner Mongolia: survey 0.08 data on land tenure reform of each county, observational data on grassland conditions and 0.07 0.06 statistical data on socio-economic indicators. 0.05 Output per sheep 0.04 0.03 Output per cattle First, the data on land tenure reform was collected via questionnaires in each county in Inner 0.02 Mongolia. Based on the typology shown in Table 4.1, the questionnaire focused on when formal 0.01 0 use rights were owned by collectives, groups and individual households respectively, and when grassland was actually in common use, joint use or private use. Questionnaires were sent to the 1979 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 2007 2009 2011 2013 Animal Husbandry Bureau of each county and were answered by key informants on local land Figure 4.4 Livestock productivity of Inner Mongolia from 1979 to 2013 tenure reform. Interviews were conducted by telephone to confirm the answers after receiving feedback from each county. We ultimately obtained valid feedback from 74 out of 102 counties. Although livestock population, meat output and livestock productivity have generally increased In addition, we conducted interviews with the officers who are working in the provincial in Inner Mongolia over time, the importance and competitiveness of its livestock production institutes of Animal Husbandry of Inner Mongolia as well as with local herders to verify the

67 68

in Inner Mongolia over time, the importance theimportance Inner time, Mongoliaover in Although livestockpopulation, meat outputand live per sheep. slightIt andsteady increase. isevidentthatth productivity for obvious fluc experienced cattle to 2013.Thelivestockproductivity shee for both Figuredevelopment oflive 4.4illustrates the in 1979. was 19timeshigherwas around14timeshigher in2013than thanin1979,whilebeefoutput

0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 100 ton/head 10 20 30 40 50 60 70 80 90 0 0 1979 1979 FigureInner Mongoliafrom 4.3Muttonandbeefoutputsof 1979to2013

Figure 4.4 Livestock productivity of Inner Mongolia from 1979 to 2013 InnerFigure fromLivestock of Mongolia 1979to2013 4.4 productivity 1981 1981 1983 1983 1985 1985 1987 1987 1989 1989 1991 1991 1993 1993 1995 1995 1997 1997 e output thantheoutput percattle increased faster

1999 Inner 1979 productivity Mongolia from in stock 67 67 and competitiveness of its livestock production livestock production its competitivenessof and 1999 tuations, while the productivityfor sheep saw a 2001 2001 p and cattle has increased. The increase of of Theincrease p andcattlehasincreased. stock productivitygenerally have increased 2003 2003 2005 2005 2007 2007 2009 2009 2011 2011 2013 2013 Output per cattle Output per sheep beef output mutton output

institutes of Animal Husbandry of Inner Mongoliaas of institutes ofAnimalHusbandry with th interviews Inconducted addition,we county. ultimately We feedback each obtain from telep conducted by wereInterviews tenure reform. onlocalland key by informants wereanswered and county each of Bureau Animal Husbandry grassland was actuallycommon use, joint in use or groups households respectively, and individual collectives, use rights and when by were owned formal on when focused thequestionnaire 4.1, inTable shown thetypology BasedMongolia. on reform tenure collected First, was thedataon land indicators. statistical dataonsocio-economic obs county, each of reform tenure data on land three types of data therefore collected 2014). We technological and development impacts onthepastorallivestockproducti data.Theother oncounty-level conducted based between counties, our empirical study of the timing ofth collection. Giventhatthe andextent therefore pastoraldistinguishedare areas from aimtoestimatetheimpactoflandtenur We 4.3.1Data collection 4.3 Datadescription areasfollowingareascrop farming isdiscussedinthesection. tothe Mongolia was cropfarming located in areas area in thepastoral of farming instead areas emphasisonthedeve Inner Mongoliahaslaid Innerproduct of Mongolia More has decreased. et al., 2009). As a result, the share represented thedevelopm upon impacted and have Mongolia of non-farming promotion the and conservation grassland projectsfor years ofecological (Su,2010).Furthermore,anumber recent intensive China where of areas crop farming within China have been threatened by Hena

environmental heterogeneity (Lienvironmental heterogeneity etal.,2008;Tessema n andShandong are provinces,which thedominant s. In 2007, 70% of livestock production in Inner Inner production in livestock 2007, 70%of In s. (Su, 2010).Thisgeographic(Su, shift from pastoral effects of the reform productionis onlivestock effects ofthe e reform on pastoral livestock production.The reform onpastorallivestock e 68 68 on are grasslandon market forces, condition, the crop farming areas for the purposes of data data farming for the purposes of crop areas the e officers who are working in the provincial workingofficers in the who are e e implementation of land tenure reform differed oflandtenure e implementation animal husbandry systems have developed in have husbandryin developed systemsanimal over, since 2001, the provincial government of government of provincial since2001,the over, ent of animal husbandry in the region (Squires (Squires theregion in husbandry ofanimal ent ed valid feedback from 74 out of102counties. 74 from ed validfeedback by animal husbandry in the gross domestic inthegrossdomestic byanimal husbandry lopment of livestock production initscrop oflivestockproduction lopment at the county-level in Inner Mongolia:survey in at thecounty-level ervational data on gra on data ervational factors that are controlled for their potential for theirpotential controlled thatare factors private use. Questionnaires were sent to the hone to confirm the answers after receiving receiving after hone toconfirmtheanswers via questionnaires in each county in Inner Inner in county viaquestionnairesineach industries have been introduced in Inner introduced in industries havebeen well as with local herders to verify the the with local toverify well herders as ssland conditions and ssland

4 progress of land tenure reform on local grasslands. Second, information about grassland in which formal use rights were assigned to households increased from 9% in 1985 to 91% in conditions was obtained based on remote sensing and an analysis of Geographic Information 2008; the proportion of counties that adopted actual private use rose from 7% in 1985 to 71% Systems. This relies on a database developed by the Chinese Academy of Sciences with original in 2008. A sharp increase in formal and actual use rights adoption is observed between 1995 data from Landsat Thematic Mapper / Enhanced Thematic Mapper (Plus) (TM/ETM+) images and 1998. This is consistent with the second round of land tenure reform implemented in this (Deng et al., 2011). GIS satellite images were only collected in 1985, 1995, 2000, 2005 and period. It can be seen that the adoption of actual private use lagged behind the assignment of 2008. We therefore use the average growth rate of grassland areas between the years of formal use rights; neither were fully completed until 2008. observation to estimate the grassland condition for each year. Third, the data on socio-economic 100% indicators is based on existing statistical data collected by local governments. Specifically, the 90% data about livestock production was gathered from the Statistical Yearbooks of Inner Mongolia. 80% proportion of 70% Market forces are represented by the real producer price of mutton and beef. This information counties in which 60% formal use rights was collected based on the Annual Compilation of Cost-benefit Data of Chinese Agricultural 50% have been assigned 40% products (Zhong Guo Nong Chan Pin Cheng Ben Shou Yi Hui Bian) and deflated by a producer proportion of 30% counties in which price index. Specific data on technological development is lacking, and is thus proxied by a 20% actual private use 10% has been adopted time variable. The factor of environmental heterogeneity among counties is removed by the 0% fixed effects model as a time-invariant variable. 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008

The research period covers 1985-2008, which includes the main period when land tenure reform was implemented on the grasslands of Inner Mongolia. After excluding the urbanised counties Figure 4.5 Progress of the land tenure reform in the pastoral areas of Inner Mongolia from and counties that underwent changes in administrative regions during the research period, 60 1985 to 2008 counties are retained in the research sample. They include 27 pastoral counties, 18 semi-pastoral Figures 4.6, 4.7 and 4.8 demonstrate the changes in livestock production and productivity of counties and 15 crop farming counties. In fact, the crop farming counties only have a few the pastoral areas relative to the crop farming areas from 1985 to 2008. The changes are permanent grasslands and their livestock production relies mainly on cropping rather than represented as an index where the level in 1985 is set equal to 100. Figure 4.6 indicates that grazing. Furthermore, household-based assignment of grassland use rights was barely both sheep population and mutton output increased in pastoral areas as well as in crop farming implemented because the areas of grasslands in most crop farming counties are too small to be areas from 1985 to 2008, and that the increase of sheep population was slower than that of subdivided. Therefore, the information from crop farming counties is only used in the mutton output, especially after 1995. In addition, after 2001, the growth rate of the sheep descriptive analysis to compare the relative changes in livestock production between crop population in crop farming areas significantly exceeded that in pastoral areas. And the growth farming areas (crop farming counties) and pastoral areas (pastoral and semi-pastoral counties). rate of mutton output of crop farming areas exceeded that of pastoral areas from 1999 onwards. 4.3.2 Descriptive analysis

Based on the collected data, we first illustrate the progress of the land tenure reform15 in the pastoral areas of Inner Mongolia from 1985 to 2008. In Figure 4.5, the proportion of counties

15 Considering that some counties did not complete the land tenure reform fully, we denote a county as having implemented the land tenure reform if at least 50% of the grassland area of this county has been assigned to individual households.

69 70

implemented the land tenure reform if at least 50% of the implemented land the atleast50% if tenure reform individual households. individual households. Based on the collected data, we first il weBased data, thecollected on analysis Descriptive 4.3.2 farming areas(cropfarming counties)and pastoral tocomparedescriptive analysis therelative changes betweencrop inlivestockproduction the used in only is counties farmingcrop from information the subdivided. Therefore, grasslands inmostcrop areasof farming are the implemented because counties toosmalltobe assi household-based grazing. Furthermore, rather than relies mainlyon cropping theirlivestockproduction and grasslands permanent counties and15cropfarming count counties are retained sample. in the research They and counties that underwent changes in administra Inner Mo grasslands of onthe was implemented land tenure reform when the main period includes 1985-2008, which covers period The research a time-invariant effects variable. fixed modelas time variable. of environmental The factor heterog Specificprice data index. on technological developm aproducer by Shou YiHuiBian) anddeflated products (Zhong GuoNong Ben ChanPinCheng DataChinese Agricultural of Annual Compilationof Cost-benefit was onthe collectedbased produce thereal by represented areMarket forces gathered f was livestockproduction data about indicators is based on existing observation toestimatethegrassland conditionfor growth ra theaverage therefore2008. We use (Deng et al., satellite 2011). imagesGIS were in 1985, 1995, 2000, only collected 2005 and Enhanc / Mapper Landsat Thematic data from Systems. This reliesdatabase on bya developed conditions was obtainedbasedonremote sensin progress reform oflandtenure grassl onlocal 15 pastoral areas of Inner Mongolia from 1985to Inner from Mongolia of pastoral areas Considering that some counties did did Considering counties some that statistical data collected by loca not complete tenure the land reformwefully, county denote a having as ies. Inies. fact,thecrop farming only counties afew have lustrate the progress of the land tenure reform landtenure progresslustrate of the the rom the Statistical Yearbooks of Inner Mongolia. Inner Mongolia. of rom theStatistical Yearbooks ed Thematic (TM/ETM+) (Plus) Mapper images 69 69 gnment ofgrassland rights use was barely 2008. In ofcounties Figure 4.5,theproportion 2008. the Chinese Sciences Academy of the withoriginal ngolia. After excluding the urbanised counties urbanised After the ngolia. excluding grassland areaofhas been this assigned county to include 27 pastoral counties, 18 semi-pastoral counties,18semi-pastoral 27pastoral include g and an analysis of Geographic Informationg and of an Geographicanalysis te of grassland areas between the years of the between ofgrassland areas te ands. Second, information about grasslandands. Second, information about each year. Third, the data on socio-economic on socio-economic year. Third,thedata each r price of mutton and r price of areas (pastoral and semi-pastoral counties). counties). andsemi-pastoral areas (pastoral tive regions during theresearch period,60 eneity amongeneity counties is removed by the ent islacking, andisthusproxied by a l governments. Specifically, the Specifically, the l governments. beef. This information beef. 15 in the in rate of mutton output of crop output rate of mutton far ex significantly areas population incropfarming In additi mutton output,especially 1995. after theincrease andthat 1985to2008, areas from asincropfarming muttonoutputincreased inpastoral both sheeppopulationand areasaswell in1985 represented where asanindex thelevel the pastoral areas to relative Figuresdemonstrate 4.6, 4.7 theand change 4.8 2008. formal werefully userights;completed until neither the adoption of act that Itcan period. be seen inthistenure reform implemented and 1998.Thisisconsistent ofland withthesecondround informal andactualus in 2008.Asharpincrease 2008; theproportionofcounties thatadoptedactu in whichformal userights wereto 91%in assigned tohouseholdsincreased fromin1985 9%

100% 10% 20% 30% 40% 50% 60% 70% 80% 90% Figure 4.5 Progress of the land tenure reform in the pastoral areas of InnerMongolia from of Figurereform inthepastoralareas landtenure 4.5Progress ofthe 0%

1985 1986 1987 1988 1989 1990 1991 1992 thecropfarmingareas 1985to2008. from Thechanges are

ming areas exceeded thatofpast areasexceeded ming 1993 1994 1995 1996 1985 to2008 1997 1998 ual private use lagged behind theassignmentprivate uselagged behind of ual 70 70 of sheep population was slower than that of thatof slower than populationwas sheep of on, after 2001, the growth rate of the sheep sheep growth rate ofthe the after2001, on, 1999 s inlivestockproductionandproductivity of s ceeded that in pastoral areas. And the growth And the that in pastoral areas. ceeded is setequalto100.Figure 4.6indicatesthat e rights adoption isobserved 1995 between 2000 al privateuserose from 7% in1985to 71% 2001 2002 2003 2004

oral areas from 1999onwards. oral areasfrom 2005 2006 2007 2008 have been assigned formal use rights counties in which proportion of has been adopted actual private use counties in which proportion of

4 Index (Year 1985=100) Figure 4.8 compares the development of livestock productivity between pastoral and crop 2800 farming areas from 1985 to 2008. It is evident that the increment of output per sheep in pastoral 2400 sheep population in areas proved to be larger than that of crop farming areas after 2002. The increment of output pastoral areas 2000 per cattle in pastoral areas was larger than that of crop farming areas since 1999. The sheep population in 1600 crop farming areas development of output per cattle was faster than per sheep in both areas. 1200 mutton output in pastoral areas 800 Index (Year 1985=100) mutton output in 2800 400 crop farming areas output per sheep in 2400 pastoral areas 0

8 2000 output per sheep in 0 crop farming areas 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 1600 output per cattle in 1200 pastoral areas Figure 4.6 Comparison of sheep population and mutton output between pastoral and crop 800 output per cattle in farming areas of Inner Mongolia from 1985 to 2008 400 crop farming areas 0 Figure 4.7 shows that the cattle population increased slightly compared with 1985, and that the output of beef has increased significantly in both pastoral and crop farming areas over the past 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008

24 years. The increase of beef output was more obvious than that of cattle population. Moreover, Figure 4.8 Comparison of livestock productivity between pastoral and crop farming areas of the growth rate of the cattle population in pastoral areas was exceeded by that in crop farming Inner Mongolia from 1985 to 2008 areas from 2001 onwards. Also, the growth rate of the beef output in pastoral areas was exceeded by the beef output in crop farming areas after 2000. In short, land tenure reform was implemented progressively in the pastoral areas in the period 1985-2008. Meanwhile, it appears that the increase of livestock production was faster in crop Index (Year 1985=100) farming areas than in pastoral areas, but conversely the development of livestock productivity 2800 cattle population in was slower in crop farming areas. It can be seen that the competitiveness of animal husbandry 2400 pastoral areas in pastoral areas has been threatened by crop farming areas in recent decades as the livestock cattle population in 2000 population and meat output has grown faster in crop farming areas since around 2000. This is crop farming areas 1600 beef output in consistent with the fact that in China, the main producing areas of livestock products have 1200 pastoral areas experienced an adjustment, moving from a grazing system to a crop-livestock mixed system 800 beef output in crop and from pastoral areas to crop farming areas (Li et al., 2008). Interestingly, the development farming areas 400 of livestock productivity was faster in pastoral areas than in crop farming areas, which may be 0 attributed to the transformation of traditional pastoralism in the pastoral areas of Inner Mongolia.

1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 As such, we assume that land tenure reform constrained the increase of livestock production,

but spurred the development of livestock productivity of pastoral areas. We will assess this Figure 4.7 Comparison of cattle population and beef output between pastoral and crop assumption with an empirical model based on the data from 45 sample counties of pastoral farming areas of Inner Mongolia from 1985 to 2008 areas in Inner Mongolia.

71 72

exceeded by the beef output in crop farming areas after 2000. after2000. areas cropfarming in output thebeef by exceeded areasfrom Also,thegrowth 2001onwards. rate the growth ofthecattlepopulationin rate pastora Moreover, population. thanthatofcattle moreobvious beef outputwas of years.The increase 24 inbot significantly increased output ofbeefhas 1985,andthatthe Figure slightly cattlepopulationincreased compared with 4.7showsthatthe 1200 1600 2000 2400 2800 1200 1600 2000 2400 2800 400 800 400 800 Index (Year 1985=100) Index ne (Year 1985=100) Index Figure 4.6 Comparison of sheep population and mutton output between pastoral andcropFigure sheep outputbetweenpastoral 4.6Comparison populationandmutton of 0 0 Figure cattle populationa 4.7Comparisonof 1985 1985 1986 1986 1987 1987 1988 1988 1989 1989 farming areas of Inner Mongolia from 1985 to 2008 Inner 1985to2008 Mongoliafarming from of areas 1990 Inner 1985to2008 Mongoliafarming from of areas 1990 1991 1991 1992 1992 1993 1993 1994 1994 1995 1995 1996 1996 1997 1997 1998 1998 1999 1999 71 71

2000 past the over areas farming crop pastoral and h 2000 nd beef output between pastoral and crop output between pastoral nd beef l areas was exceeded by that incropfarming by that l areas was exceeded

2001 was outputinpastoralareas ofthebeef 2001 2002 2002 2003 2003 2004 2004 2005 2005 2006 2006 2007 2007 2008 200808 farming areas beef output in crop pastoral areas beef output in crop farming areas cattle population in pastoral areas cattle population in crop farming areas mutton output in pastoral areas mutton output in crop farming areas sheep population in pastoral areas sheep population in

was slower in crop farming areas. It can beseen can It areas. farming incrop was slower livestock productivity of development converselythe but areas, farming than in pastoral areas appearsthattheincrease 1985-2008. Meanwhile,it period inthe areas pastoral inthe ssively progre was implemented tenurereform In short,land areas in Inner Mongolia. Innerareas Mongolia. in assumption with anempiricalmodelbasedon but of spurred li the development land As such,weassumethat attributed tothetransformationoftraditionalpast of livestockproductivitywas faster Interestingly, development the et al., 2008). (Li farming areas crop to areas pastoral and from adjustment, movinggr an from a experienced consistent withthefact thatinChina,th population andmeat outputhasgrown in faster fa crop by threatened been has in pastoralareas sheepdevelopment ofoutputpercattle thanper inbothareas. wasfaster than larger was areas cattle inpastoral per fa crop thanthatof tobelarger areas proved Itto 2008. farming isevidentthat from areas 1985 live developmentof the Figure 4.8 compares

1200 1600 2000 2400 2800 Figure livestockfarming productivity 4.8crop of Comparison areas and of pastoral between Index (Year 400 800 0 1985 1986 1987 1985=100) 1988 1989 1990 1991 1992 Inner Mongolia from 1985 to 2008 Inner 1985to2008 Mongolia from

tenure reform constrained the increase of livestock production, production, oflivestock theincrease reformtenure constrained 1993 1994 vestock productivity of pastoral areas. We will assess We this areas. vestock productivity pastoral of

in pastoral areas than in crop farmingin pastoral areas than in crop areas, which may be 1995 1996 1997 e main producing areas of livestock products have e main areasoflivestockproductsproducing 1998 rming areas after 2002. The increment of output of The increment 2002. rmingareas after 72 72 that of crop farming areas since 1999. The The since1999. areas farming ofcrop that azing system to a crop-livestock mixed system mixed azing system toacrop-livestock

1999 crop and pastoral between productivity stock rming areas in recent decades as the livestock livestock asthe recent decades in rming areas that the competitiveness of animal husbandry of the competitiveness that the datafrom 45sample crop farming sincearound 2000. Thisis areas oralism inthepastoral 2000 the increment of output per sheep in pastoral inpastoral sheep ofoutputper the increment

of livestock production was fasterof livestockproduction incrop 2001 2002 2003 2004 2005 2006 2007

areas of Inner of Mongolia. areas 2008 countiesofpastoral pastoral areas output per cattle in crop farming areas output persheep in pastoral areas output persheep in crop farming areas output per cattle in

4 4.4 Empirical model

A fixed effects model is widely used in economic research, primarily to study the causes of changes within entities over time (e.g., Fergusson et al., 2002; Huang et al., 2006). The model Max 23.12 18.68 68.90 64.58 25 1.94 3.10 1.11 1.00 24 1

employs within transformation to remove all time-invariant (fixed) explanatory variables, i.e. 17 the model is performed in deviations from individual means (Verbeek, 2012). As such, the fixed Min 0.20 0.12 0.60 0.14 0 0.59 0.63 0.79 0.09 1 0 effects model provides a method that takes observable as well as unobservable time-invariant explanatory variables into account, but the estimation is not dependent on the value of time- invariant (fixed) variables (Verbeek, 2012). Such an approach is appropriate in this study Std. Dev 1.74 0.89 6.47 5.93 5.69 0.49 0.71 0.05 0.20 6.93 na ne hand, these extreme hand, these be ne can values

considering that the results of land tenure reform within each county over time can be studied these government-drivenhat are factors

d due to government interventions. For instance, instance, For interventions. government to d due

effectively by controlling for the unmeasured heterogeneity among counties. On the other hand, 19

the explanatory variables of land tenure reform we are interested in show changes in value Mean 1.73 1.08 4.29 4.30 4.28 1.15 1.29 0.98 0.80 12.50 0.60 across our research period for a substantial proportion of counties in our data (see Figure 4.5), which satisfies the basic requirement of the fixed effects model (Daun-Barnett, 2011). In addition, our research sample attempts to include all counties in the pastoral areas of Inner Observations 1080 1080 1080 1080 1035 1035 1035 1035 1035 1080 1080 Mongolia, rather than random draws, which preliminarily indicates that a fixed effects model is more appropriate than a random effects model (Verbeek, 2012). 74 Based on the theoretical framework of the fixed effects model, we formulate the effects of the Unit na na na na Years Yuan/kilo Yuan/kilo na na na na land tenure reform and other potential factors on livestock production into the following equations: production may appear to be unusually small (or large). On the On (or large). small appear to be unusually may production o

௜௧ ௜ ଵ ௜௧ିଵ ଶ ௜௧ିଵ ଷ ௜௧ିଵ ସ ௜௧ିଵ ହ ௜௧ିଵ

݈݋݃ሺܵ ሻൌݏ ൅ܽ ݈݋݃ሺܴ ሻ൅ܽ ሺͳΤ ሺܴ ሻሻ ൅ ܽ ݈݋݃ሺܯ ሻ൅ܽ ݈݋݃ሺܶ ሻ൅ܽ ݈݋݃ሺܩ ሻ change sharply have structures may livestock hand, other the On ).

௧ ൅ᖡ௜௧ሺͳሻ appendix 4-A. in be found can in the model usedܻכ ൅ܽ଺ܻ௧ ൅ܽ଻ܲ௜ Table 4.2 Definition of variables and summary statistics ݈݋݃ሺܥ௜௧ሻൌ݃௜ ൅ܾଵ݈݋݃ሺܴ௜௧ିଵሻ൅ܾଶሺͳΤ ሺܴ௜௧ିଵሻሻ ൅ ܾଷ݈݋݃ሺܤ௜௧ିଵሻ൅ܾ ସ݈݋݃ሺܶ௜௧ିଵሻ൅ܾହ݈݋݃ሺܩ௜௧ିଵሻ ௧ ൅Ɋ௜௧ሺʹሻܻכ ൅ܾ଺ܻ௧ ൅ܾ଻ܲ௜ ௜௧ିଵሻܩ௜௧ିଵሻ൅ܿ ସ݈݋݃ሺܶ௜௧ିଵሻ൅ܿହ݈݋݃ሺܯ݋݃ሺܷ௜௧ሻൌݑ௜ ൅ܿଵ݈݋݃ሺܴ௜௧ିଵሻ൅ܿଶሺͳΤ ሺܴ௜௧ିଵሻሻ ൅ ܿଷ݈݋݃ሺ݈ ௧ ൅ߣ௜௧ሺ͵ሻܻכ ൅ܿ଺ܻ௧ ൅ܿ଻ܲ௜

݈݋݃ሺܨ௜௧ሻൌ݂௜ ൅݀ଵ݈݋݃ሺܴ௜௧ିଵሻ൅݀ଶሺͳΤ ሺܴ௜௧ିଵሻሻ ൅ ݀ଷ݈݋݃ሺܤ௜௧ିଵሻ൅݀ସ݈݋݃ሺܶ௜௧ିଵሻ൅݀ହ݈݋݃ሺܩ௜௧ିଵሻ ଺ ௧ ଻ ௜ ௧ ௜௧ ൅ߜand yearሺͶሻ t. Variables are defined in Table 4.2 and summary ܻכ where i and t present൅݀ theܻ ൅݀ ith countyܲ Variable definition Variable t population of county i in Ratio of sheep year over that in 1985 t population of county i in year Ratio of cattle over that in 1985 i in t over county year Ratio of mutton output that in 1985 t over i in Ratio of beef output county year that in 1985 formal that rights use of Number grassland years have been assigned to individual households in t-1 i until county year Real producer price of mutton t-1 in year Real producer t-1 price of beef in year of t area county i in Ratio of total grassland year lagover that in 1985, with one year over total quality area of good grassland Share areagrassland of county t, with one i in year year lag tYear =0 otherwise=1 if county i is a pastoral county,

statistics are provided. 1. equals variable the dummy for which observations share of to the refers ୧ ܲ

௜ ௧ ௜௧ ௜௧ ௜௧ ௜௧ ௜௧ିଵ ܻ ܲ ௜௧ିଵ ௜௧ିଵ ௜௧ିଵ ௜௧ିଵ ܵ ܨ ܥ ܷ ܶ ܩ ܴ ܤ ܯ It is noted that the (or the livestock values noted for minimum It is that maximum) 2, etc. equals 1, 1986 equals Year 1985 The summary statisticsThe variables are summary the that based on The mean of Variable the implementation of the grazing ban severely decreased the sheep population in several counties. In our analyses, we assume t in assume analyses, population we our counties. In the sheep decreased several ban severely grazing the of implementation the variable. time the by captured attributed to the high (or low) values in the base (1985 base year in the values (or low) high to the attributed 16 17 18 19

where invariant (fixed) variables (V variables invariant (fixed) esti account, but the into explanatory variables effectsmethod model that provides takes a observa from deviations indivi the modelisperformed in i.e. variables, explanatory(fixed) toremoveemploys alltime-invariant within transformation changes within entities over time (e.g., Ferg effects modeliswidelyA fixed usedineconom 4.4 Empiricalmodel of considering that theresults equations: livestockproduction into thefollowing other potentialfactorson reform and land tenure Basedframework of ef the fixed on the theoretical 2012). model(Verbeek, a randomeffects than is more appropriate effects model fixed a that indicates whichpreliminarily randomdraws, than rather Mongolia, addition, ourresearch attemptstoincl sample the requirement of basic which satisfiesthe across our research for period a substantial propo we oflandtenurereform the explanatory variables theotherhand, for On theunmeasuredeffectively controlling by heterogeneity counties. among statistics are provided. statistics are provided. ݈݋݃ሺܵ ݈݋݃ሺܨ ݈݋݃ሺܥ ݈݋݃ሺܷ i ௜௧ and ௜௧ ௜௧ ௜௧ ሻൌݏ ሻൌ݂ ሻൌ݃ ሻൌݑ t present the ௜ ௜ ௜ ௜ ൅ܽ ൅݀ ൅ܾ ൅ܿ ൅ܽ ൅݀ ൅ܾ ൅ܿ ଵ ଵ ଵ ଵ ݈݋݃ሺܴ ݈݋݃ሺܴ ݈݋݃ሺܴ ݈݋݃ሺܴ ଺ ଺ ଺ ଺ ܻ ܻ ܻ ܻ ௧ ௧ ௧ ௧ ൅ܽ ൅ܿ ൅ܾ ൅݀ i ௜௧ିଵ ௜௧ିଵ ௜௧ିଵ th ௜௧ିଵ ଻

଻ ଻ ଻ countyyear t and ܲ ܲ ܲ ܲ ሻ൅ܽ ሻ൅݀ ௜ ሻ൅ܾ ௜ ሻ൅ܿ ௜ ௜ land tenure reform within eachc land tenurereformwithin erbeek, 2012). Such anapproach 2012).Such erbeek, ܻכ ܻכ ܻכ ܻכ ௧ ௧ ௧ ௧ ଶ ଶ ଶ ଶ ൅ߣ ൅ᖡ ൅Ɋ ͳሺܴ ሺͳ ൅ߜ ͳሺܴ ሺͳ ͳሺܴ ሺͳ ͳሺܴ ሺͳ Τ Τ Τ Τ ௜௧ ௜௧ ௜௧ ௜௧ ሺ͵ሻ ሺͳሻ ሺʹሻ ሺͶሻ ௜௧ିଵ ௜௧ିଵ ௜௧ିଵ ௜௧ିଵ usson etal.,2002; Huangusson etal.,2006).Themodel ሻ൅ܽ ൅ ሻሻ ሻ൅݀ ൅ ሻሻ ሻ൅ܿ ൅ ሻሻ ሻ൅ܾ ൅ ሻሻ . Variables inTable4.2andsummary are defined 73 73 ude all counties in the pastoral areas of Inner Inner of ude all counties in the pastoral areas fixed effects model (Daun-Barnett, 2011). In 2011). (Daun-Barnett, effects model fixed mation is not dependent on dual means Assuch,thefixed (Verbeek, 2012). ic research, primarily to study the causes of tostudy causesof research, the primarily ic rtion ofcountiesin ourdata (see Figure 4.5), ଷ ଷ ଷ ଷ fects model,we formulate theeffects ofthe ݈݋݃ሺܯ ݈݋݃ሺܯ ݈݋݃ሺܤ ݈݋݃ሺܤ ble as well as unobservable time-invariant time-invariant ble aswell asunobservable are interested in show changes in value are in show changes in value interested ௜௧ିଵ ௜௧ିଵ ௜௧ିଵ ௜௧ିଵ ሻ൅݀ ሻ൅ܾ ሻ൅ܽ ሻ൅ܿ ounty can be time studied over

ସ ସ ସ ସ is appropriate inthisstudyis appropriate ݈݋݃ሺܶ ݈݋݃ሺܶ ݈݋݃ሺܶ ݈݋݃ሺܶ ௜௧ିଵ ௜௧ିଵ ௜௧ିଵ ௜௧ିଵ the value of time- thevalueof ሻ൅݀ ሻ൅ܾ ሻ൅ܿ ሻ൅ܽ ହ ହ ହ ହ ݈݋݃ሺܩ ݈݋݃ሺܩ ݈݋݃ሺܩ ݈݋݃ሺܩ ௜௧ିଵ ௜௧ିଵ ௜௧ିଵ ௜௧ିଵ ሻ ሻ ሻ ሻ

Table 4.2 Definition of variables and summary statistics16

Variable Variable definition Unit Observations Mean Std. Dev Min Max Ratio of sheep population of county i in year t na 1080 1.73 1.74 0.2017 23.12 over that in 1985 ௜௧ ܵ Ratio of cattle population of county i in year t na 1080 1.08 0.89 0.12 18.68 over that in 1985 ௜௧ ܥ Ratio of mutton output of county i in year t over na 1080 4.29 6.47 0.60 68.90 that in 1985 ௜௧ ܷ Ratio of beef output of county i in year t over na 1080 4.30 5.93 0.14 64.58 that in 1985 ௜௧ ܨ Number of years that formal grassland use rights Years 1035 4.28 5.69 0 25 have been assigned to individual households in ܴ௜௧ିଵ county i until year t-1 Real producer price of mutton in year t-1 Yuan/kilo 1035 1.15 0.49 0.59 1.94

ܯ௜௧ିଵ Real producer price of beef in year t-1 Yuan/kilo 1035 1.29 0.71 0.63 3.10 ܤ௜௧ିଵ Ratio of total grassland area of county i in year t na 1035 0.98 0.05 0.79 1.11 over that in 1985, with one year lag ௜௧ିଵ ܶ Share of good quality grassland area over total na 1035 0.80 0.20 0.09 1.00 grassland area of county i in year t, with one year ܩ௜௧ିଵ lag Year t na 1080 12.5018 6.93 1 24 19 ܻ௧ =1 if county i is a pastoral county, =0 otherwise na 1080 0.60 na 0 1 ܲ௜

16 The summary statistics based on the variables that are used in the model can be found in appendix 4-A. 17 It is noted that the minimum (or maximum) values for livestock production may appear to be unusually small (or large). On the one hand, these extreme values can be attributed to the high (or low) values in the base year (1985). On the other hand, livestock structures may have sharply changed due to government interventions. For instance, the implementation of the grazing ban severely decreased the sheep population in several counties. In our analyses, we assume that these government-driven factors are captured by the time variable. 18 Year 1985 equals 1, 1986 equals 2, etc. 19 The mean of refers to the share of observations for which the dummy variable equals 1. 74 ୧ ܲ 4 Sheep and cattle population and mutton and beef output are used to represent the livestock of grassland ( ). The grassland quality is presented by the share of the area of good quality 22 production. The ratios of these four indicators in year t compared with the base year 1985 grassland in ܶthe௜௧ total grassland area in year t ( ). ( ), expressed in logarithms, are employed as the four dependent variables in ܩ௜௧ Technological development is presented by the time variable . It should be noted that the equationsܵ௜௧ǡܥ௜௧ǡܷ ௜௧(1)-(4).ܨ௜௧ The four equations are denoted as the sheep population, cattle population, development of technology in pastoral areas is mostly government-driven,௧ mainly as a result of mutton output and beef output models, respectively. Variable indicates the land tenure ܻ the growing interest of the central government in the ecological benefits of grasslands in recent reform of county i in year t, which is presented by the number of ܴyears௜௧ in which the formal use years. For instance, a series of Ecological Construction Programs for grassland conservation rights of grassland have been owned by individual households in county i by year t. We assume have been implemented since 2000 to develop intensive animal husbandry through improving that land tenure reform has a non-linear relationship with livestock production. Plenty of the technology of animal husbandry (Li et al., 2014). These have stimulated the improvement scholars have proved the existence of a nonlinear relationship between access to land and of forage production, feeding and fattening techniques and the use of improved breeds (Liao, agricultural production (e.g. Finan et al., 2005). Ostrom (2007) claimed that the problems linked 2009). The time variable is therefore used to interpret the impact of these ecological with social-ecological systems require serious study which take account of complex, programs as well. is a dummy୲ variable that is one for pastoral counties and zero otherwise. multivariable, nonlinear, cross-scale, and changing systems. Our functions attempt to explore The interaction term௜ of and is included in the equations to recognise the differences the trend of livestock production under the land tenure reform over time based on the ܲ between pastoral counties ௧and semi-pastoral௜ counties in terms of technological development assumption of their non-linear relationship, which can be concave, convex, quasi-concave or ܻ ܲ and ecological policies. Apart from the time and dummy variables, we treat all independent quasi-convex. The logarithm and reciprocal of the variable of land tenure reform are employed variables with a one year lag because market forces, grassland condition and land tenure reform to present the non-linear relationships20. Considering that the value of the land tenure reform are all considered to affect livestock production with a time delay. Moreover, market forces and variable is zero for some counties as they had not implemented the reform by 2008, we add a 21 grassland condition are used in logarithmic form. small original value (e.g., 0.01) to when is specified in or .

ܴ௜௧ ܴ௜௧ ݈݋݃ሺܴ௜௧ሻ ͳΤ ሺܴ௜௧ሻ Finally, other factors such as elevation, terrain slope and distance to the provincial capital, Based on the existing academic research, the factors of market forces, grassland condition, which are not expected to change significantly over time, are treated as time-invariant (fixed) technology development and environmental heterogeneity are widely suggested to impact the factors. They are represented by the terms , , and in the models. The coefficients of pastoral livestock production (Li et al., 2008; Squires et al., 2009; Tessema et al., 2014). As the independent variables are , , and ௜ ௜(n=1,௜ 2, ...௜ 7) and the random error terms are ݂ ݑ ݃ ݏ such, they are considered as control variables in our functions for disentangling the impact of , , and . There are 1035௡ ௡ observations௡ ௡ used in the fixed effects model, covering 45 land tenure reform. Market forces are proxied by the local mutton price ( ) and the beef price ܽ ܾ ܿ ݀ countiesᖡ௜௧ Ɋ௜௧ ߣ and௜௧ 23 ߜyears.௜௧ ( ) which are expected to affect supply decisions of livestock producersܯ௜௧ (Komarek et al., 2012).ܤ௜௧ The grassland condition refers to the quantity and quality of grasslands. It indicates the 4.5 Model results carrying capacity for livestock, reflecting grassland degradation and climate change factors Before the estimation, we first conduct a Hausman test to choose between the fixed effects or such as temperature and water availability (Henry et al., 2012). Specifically in our model, the random effects regression as the estimation technique. The test results for the models of sheep ratio of total grassland area in year t over that in 1985 is used to present the changes in quantity population, mutton output and beef output reject the null hypothesis at p<0.01, indicating that

the fixed effects regression is more appropriate. Only the test result for the model of cattle 20We have also tested a model with a quadratic term instead of the reciprocal to specify the non-linear relationship. The results of both model specifications were comparable, which indicates the model specification is reliable and the model results are robust (see appendix 4-B for details of the specification with the quadratic form). 22 ‘Good quality grassland’ denotes grassland where the canopy cover of grass is more than 20%. Grasslands are 21 0.01 is small enough to be regarded as the original value because we tried smaller values (such as 0.001, and divided into three categories according to the canopy cover, namely dense grassland, moderate grassland and 0.0001) and obtained consistent model results. sparse grassland (Deng et al., 2011). In this paper, good quality grassland includes dense grassland and moderate grassland.

75 76

mutton output andbeefoutputmodels,respectively. Variable equa Thefour equations (1)-(4). to present thenon-linearrelationships ofth andreciprocal Thelogarithm quasi-convex. assumption oftheirnon-linearrelationship, whic the trendoflivestockproductionunderla andchan multivariable, nonlinear,cross-scale, withsystems social-ecological study of complex, serious account which require take agricultural (e.g.et al., Finan production ofanonlin theexistence scholars haveproved that land tenure relati has a non-linear reform households incounty by individual owned rightsgrassland been of have reform i county of ( ( production. The ratios ofthesefou production. Theratios Sheep and cattle population and mutton and beef out 21 20 year in grasslandratio area of total (Hen such andwater availabilityas temperature grassl reflecting livestock, for capacity carrying refersgrassland tothequanti condition 2012). The variable is zero for some counties as they ha counties some for zero variable is small original 0.01) (e.g., value 0.0001)modelconsistent and obtained results. form). form). model andthe arerobu is reliable results relationship. results The model ofwe both specifications land tenure reform. Market forces are proxied by the local mutton price ( local by muttonprice proxied the forces are Market reform. land tenure ascontrolvariables inoursuch, they considered functions are for disentangling theimpactof pastoral livestockproduction(Li etal.,2008; toimpactthe suggested technology widely and environmentalare developmentty heterogenei the research, academic existing Based onthe ܤ ܵ We have also tested a modelwith have a We alsotested aquadratic term inst 0.01 is small enough toberegarded asthe originalwevalue triedsmaller because values (such as0.001, and ௜௧ ௜௧ ǡܥ ) which are expected to affect supply deci supply to affect expected are ) which ௜௧ ǡܷ ௜௧ ܨ ௜௧ in year in t ), expressed in logarithms, are employed as the fourin dependent variables , which is presented by the number of years in which the formal use theformal yearswhich in thenumber of by , whichispresented 21 tions are denoted asthe sheeppopulation,cattle tions are denoted to st (see appendix 4-B for detailsof (see appendix 4-B st t over that in 1985 is used to present the changes in quantity changes is used to present the 1985 that in over r indicators in year t compared with the base year 1985 the base yearcompared with t r indicators in ܴ ௜௧ 20

. Considering tenure reform thatthevalueofland 2005). Ostrom (2007) claimed that the problems linked problemslinked claimed thatthe Ostrom (2007) 2005).

when ܴ d not implemented the reform by 2008, we add a add we by2008, reform implementedthe d not 75 75 ead of the reciprocal to specify the non-linear ofead the specify reciprocalto non-linear the ging systems. Our attempttoexplore functions Squires et al., 2009; Tessema etal.,2014).As Squires etal., 2009;Tessema icates the modelwhichre comparable, specification icates ind the factors of market forces, grassland condition, forces, factors ofmarket sions of (Komareklivestock et al., producers ry et al., 2012). Specifically in our model, the ourmodel, in Specifically etal.,2012). ry ௜௧ e variable of land reformtenure are employed onship with livestockproduction.onship with Plenty of ear relationship between access to land and tolandand ear relationship betweenaccess and degradation and cl and degradation and h can be concave, convex, quasi-concave or quasi-concave h canbeconcave,convex, nd tenuretime reform based over on the

ty and quality of grasslands. Itgrasslands. the indicates quality of and ty is specified in put are usedtorepresent thelivestock the specification withthe specification quadratic the ܴ ݈݋݃ሺܴ ௜௧ idcts h ln tenure land the indicates ௜௧ ܯ i by year year by imate change factors imate change ሻ ௜௧ or ) and the beef price price ) andthebeef ͳሺܴ ͳ Τ t . We assume . We ௜௧ ሻ . The interactionterm of grassland. grassland. sparse grassland this (Deng2011). In paper,good etal., divided intoaccording three categories to the canopy the fixed effectsthe fixed regression is population, mutton output andbeefoutput reject thenullhypothesis indicating atp<0.01, that regression astheestimationtechni random effects Before a Hausma estimation, first conduct we the 4.5 Modelresults years. counties and23 grassland the independent are variables 2009). The time variable 2009). Thetimevariable of forage feeding production, and fattening and the use of improved techniques breeds (Liao, et al husbandry (Li animal of the technology todeve 2000 implementedsince have been ofEcological Cons years.For aseries instance, government inth central the growing ofthe interest development oftechnology inpastora of grassland ( 22 factors. They are represented by theterms represented are factors. They tochangesignificantly which are notexpected Finally, terra as elevation, such factors other inlogarithmic form. grassland conditionareused livestock production all to affect are considered for market lag because year with aone variables allindependent wetreat dummy variables, and thetime Apart from policies. and ecological coun andsemi-pastoral counties between pastoral programs as well.programs as ᖡ Technological development is pre Technological development ௜௧

‘Good qualitygrassland’ grassland denotes where the cano , Ɋ ௜௧ , ߣ 22 ௜௧ andߜ in the total grassland area in year inthetotalgrassland( areain t ܶ ௜௧ ). The grassland quality). Thegrassland by ispresented ௜௧ ܲ There are 1035 observations used in the fixed effects model, covering 45covering effects model, fixed observations usedinthe 1035 are . There ௜ is a dummy variable that is one for pastoral counties and zero otherwise. zero and counties for pastoral that isone is adummy variable ܻ ௧ and ୲ tointerpretis therefore used ܽ more appropriate. Only the test ௡ ܲ , ܾ ௜ i icue i te equations the in included is sented by the time variable by the sented ௡ , ܿ l areasismostly mainlygovernment-driven, as aresult of ௡ and lop intensive animal husbandry through improving through lop intensiveanimalhusbandry ݏ ௜ ., 2014). These have stimulated the improvement stimulated These have ., 2014). ݀ , ݃ 76 76 in slope and distance provincial toin the capital, slope cover, namely dense grassland, moderate namelycover, grassland, grassland dense and ௡ qualitygrassland includes dense grasslandmoderate and with a time delay. Moreover, market and forces Moreover, a time delay. with ces, grassland condition and land tenure reform landtenure conditionand grassland ces, ܩ over time, are treated as time-invariant (fixed) (fixed) astime-invariant time,are treated over (n=1, 2, ... 7) and the random terms error are ௜ truction Programs conservationtruction grassland for ௜௧ ݑ , n test to choose between the fixed effects or effects the fixed n test to choose between que. The test results for the models of sheep results models ofsheep forque. Thetest the ). ties in terms of technological development development intermsoftechnological ties e ecological benefits of grasslands in recent ofgrasslands inrecent benefits e ecological pymore 20%.Grasslands is grass than cover of are ௜ and and the share of the thearea share of good quality ݂ ௜ of in themodels.The coefficients the impact of these ecological ecological theimpactofthese ܻ ௧ result for themodelofcattle . It should be noted that the Itbe noted that the . should to recognise the differences differences the to recognise

4 population does not reject the null hypothesis at p<0.01. We conducted the random effects and fixed effects regression for the model of cattle population. The results are comparable, with the

exception of the coefficient of the constant term. To compare the results of the four models ሻ ௜௧ consistently, we use the fixed effects regression for all of them. t -1.83 -1.94 na 2.95 1.09 1.21 17.11 -10.65 2.46 ݈݋݃ሺܨ Furthermore, we discuss the exogeneity of the variable of land tenure reform ( ). indicates

the number of years for which formal grassland use rights have been assigned.ܴ௜௧ In ܴpractice,௜௧ the

assignment of formal grassland use rights was implemented by the county-level government model output Beef coefficient -0.08* -0.005* na 0.19*** 0.57 0.83 0.10*** -0.05*** 0.50** 0.60 103,5 based on top-down executive orders. According to our survey, the differences in the assignment policies. This explicitly not is factor of formal grassland use rights among counties resulted mainly from the implementation ically of the variables explanatory captured by f technological and policy developments. technological policy and f We

efficiency of each county government and the county’s geographical location. For instance, a ሻ ௜௧

county that is located in a remote area was less motivated to assign formal use rights. In this t -2.63 -2.33 0.87 na -1.53 3.03 22.08 -4.76 1.44

regard, the speed of implementation of the land tenure reform by local governments was not ݈݋݃ሺܷ related to local livestock production, grassland condition, or other factors that we include in the

model. Hence, we treat as an exogenous variable. In contrast, the decision by local to be shown is cattle on population the reform land tenure effect of , the Mutton output model model output Mutton coefficient -0.09*** -0.004** 0.06 na -0.64 1.67*** 0.12*** -0.02*** 0.24 0.77 103,5 2 R households to actually adoptܴ௜௧ private use is likely to have taken into account the grassland upon livestock production in the pastoral areas pastoral the in production livestock upon condition, living customs, production pattern etc., and would therefore have created a potential

78 t -3.36 -3.52 na 3.83 4.10 -0.52 2.58 -8.51 1.00

problem of endogeneity. The exogeneity of policy implementation in China was also supported ሻ ௜௧ by Liu et al. (2010) based on their research about the implementation of the Sloping Land

Conversion Program (SLCP). ݈݋݃ሺܥ

23 The fixed effects model is used to estimate our four models based on the panel data set. The counties may counties adjustmentgovernment-driven resultedhave from Cattle population model model population Cattle coefficient -0.10*** -0.01*** na 0.18*** 1.54*** -0.25 0.01*** -0.03*** 0.15 0.12 103,5 reover, some factors that affect cattle population are not are not specif cattle population that affect factors some reover, model results about how the factors impacted upon livestock production in the pastoral areas effects potential time used the the to capture variable merely o are listed in Table 4.3. cou in crop farming prominent is thatmore cattle production much the fact ntiesned by and major changes cattle in t -5.68 -4.93 3.28 na -0.21 2.21 8.18 -0.91 2.36

ሻ our research limitation. low our Despite the of value

௜௧ ݈݋݃ሺܵ Table 4.3 Model results impact that results factors Model the 4.3 of Table Sheep population model coefficient -0.18*** -0.01*** 0.23*** na -0.08 1.17** 0.04*** -0.003 0.38** 0.30 103,5

ሻ ሻ ሻ ሻ ሻ ሻ

௧

௜௧ିଵ ௜௧ିଵ ௜௧ିଵ ௜௧ିଵ ௜௧ିଵ ଶ ௜௧ିଵ ௧

ܻכ ܻ for the cattle population model may be explai be may model population cattle the for ܴ 2 ௜ R ܲ Τ Constant Constant Variables Variables ͳ ሺܴ ݈݋݃ሺܶ ݈݋݃ሺܤ ݈݋݃ሺܩ ݈݋݃ሺܴ ݈݋݃ሺܯ Observations The low The low significant. *significant at 5%; ***significant at 1%.*significant at 10%; **significant 23 our model. For instance, changes in the cattle population in some in some cattle in population the changes For instance, our model. population overtime also occurred mainly in these counties. Mo in these counties. also occurred mainly overtime population included in our model because of data limitations, and we have and we limitations, data because of our model in included have discussionhave in the stated point as this section

Furthermore, we discuss the exogeneity of the variable of land tenure reform ( Furthermore, exogeneity of we of the land variable tenure discuss reform the effects allconsistently, regression thefixed for ofthem. we use of the of constant the coefficient term exception fixed effects regression for the model of cattle themodelof regression for effectsfixed population doesnotreject thenullhypothesis conductedtherandom atp<0.01.We effects and

The fixed effects model is used effectsThe to fixed estimate model Conversion Program (SLCP). Liu ontheirresearchby ab etal.(2010)based was supported also implementation inChina of policy The exogeneity problem of endogeneity. condition, living customs,producti households toactually privateuseislike adopt are listedinTable4.3. model results abouthow thefactors impacted based on top-down executive orders. Accordingbased ontop-downexecutive to assignmentgrasslandformal userights was of im the In practice, assigned. been have rights grassland use which formal years for the numberof model. Hence, we treat we treat model. Hence, grassland condi production, localrelated livestock to regard, ofthelandte ofimplementation thespeed county thatislocatedmotivated to assign inaremote was area less formalIn userights. this a instance, For geographical location. county’s and the government county each of efficiency of formal grassland rights mainly use resulted amongfrom theimplementation counties ܴ ௜௧ byIn local thedecision contrast, variable. as anexogenous on pattern etc., and would therefore have createdaand potential wouldtherefore on patternetc., have our four modelsbasedonth our four upon livestock production in the pastoral areas 77 77 population. Theresults comparable, arepopulation. withthe out theimplementationofout Land theSloping ly into theto have account grassland taken . To compare the results of the four models four models of the the results . To compare our survey, the differences intheassignment our survey, thedifferences plemented by the county-level government government county-level plemented by the nure reform by governments was local not tion, or other factors that we include in the weincludein that factors tion, orother e panel data set. The panel data set. e ܴ ௜௧ ).

ܴ ௜௧ indicates

Table 4.3 Model results of the factors that impact upon livestock production in the pastoral areas

Variables Sheep population model Cattle population model Mutton output model Beef output model

coefficient t coefficient t coefficient t coefficient t ݈݋݃ሺܵ௜௧ሻ ݈݋݃ሺܥ௜௧ሻ ݈݋݃ሺܷ௜௧ሻ ݈݋݃ሺܨ௜௧ሻ -0.18*** -5.68 -0.10*** -3.36 -0.09*** -2.63 -0.08* -1.83 -0.01*** -4.93 -0.01*** -3.52 -0.004** -2.33 -0.005* -1.94 ݈݋݃ሺܴ௜௧ିଵሻ 0.23*** 3.28 na na 0.06 0.87 na na ͳΤ ሺܴ௜௧ିଵሻ na na 0.18*** 3.83 na na 0.19*** 2.95 ݈݋݃ሺܯ௜௧ିଵሻ -0.08 -0.21 1.54*** 4.10 -0.64 -1.53 0.57 1.09 ݈݋݃ሺܤ௜௧ିଵሻ 1.17** 2.21 -0.25 -0.52 1.67*** 3.03 0.83 1.21 ݈݋݃ሺܶ௜௧ିଵሻ 0.04*** 8.18 0.01*** 2.58 0.12*** 22.08 0.10*** 17.11 ݈݋݃ሺܩ௜௧ିଵሻ -0.003 -0.91 -0.03*** -8.51 -0.02*** -4.76 -0.05*** -10.65 ܻ௧ Constant 0.38** 2.36 0.15 1.00 0.24 1.44 0.50** 2.46 ௧ܻכ ௜ܲ 0.30 0.1223 0.77 0.60 Observationsଶ 103,5 103,5 103,5 103,5 ܴ *significant at 10%; **significant at 5%; ***significant at 1%.

23The low R2 for the cattle population model may be explained by the fact that cattle production is much more prominent in crop farming counties and major changes in cattle population overtime also occurred mainly in these counties. Moreover, some factors that affect cattle population are not specifically captured by the explanatory variables of our model. For instance, changes in the cattle population in some counties may have resulted from government-driven adjustment policies. This factor is not explicitly included in our model because of data limitations, and we have merely used the time variable to capture the potential effects of technological and policy developments. We have stated this point in the discussion section as our research limitation. Despite the low value of R2, the effect of land tenure reform on the cattle population is shown to be significant.

4 First, we focus on the coefficients of and , indicating the effects of the land tenure the mutton price in year t-1 increases by 1%. Similarly, cattle population (beef output) in year t will reform on livestock production over ݈݋݃time. ( ܴ ௜௧ିଵAll )of these1(Τ coefficientsܴ௜௧ିଵ) are significant and negative in the increase by 0.18% (0.19%) when the beef price in year t-1 increases by 1%. The better grassland four models. Based on the coefficients of and , quasi-concave curves are obtained condition also has a significant and positive effect on livestock production. Cattle population in year t as depicted in Fig. 9. The horizontal axis݈݋݃ presents ( ܴ௜௧ିଵ the) number1(Τ ܴ ௜௧ିଵof years) that the land tenure reform has will increase by 1.54% if the total grassland area in year t-1 increases by 1%. Sheep population (mutton been implemented and the vertical axis shows the livestock production index (livestock production in output) in year t will increase by 1.17% (1.67%) if the good quality grassland area in year t-1 increases the first year of land tenure reform equals 100). Considering the longest duration for formal use rights by 1%. In an attempt to improve grassland condition, China has implemented various ecological projects. privatised until 2008 was 26 years in our research sample, we determined the maximum year of the However, in reality, 90% of the grasslands are still degraded to various degrees (Mei et al., 2013). For horizontal-axis as 26. example, between 1991 and 2002, 3.1% of the available natural grassland was lost, either through its being degraded into unused land, or being transformed into other purposes. During the same period, Index (Year 1=100) 1.68 million hectares of land was rehabilitated into natural grassland through eco-environmental 105 projects. This resulted in a net decline of 6.38million hectares (around 2.4% of the available natural 100 grassland) (Qu et al., 2011). In this regard, the deteriorating grasslands could have impeded the livestock sheep population 95 production. cattle population 90 mutton output The coefficients of the time variable are significant and positive in four models. It indicates that the 85 beef output technology development affects livestock production positively, and it can be seen that the positive 80 effects are stronger with respect to meat output than for the livestock population. The interaction term

75 Years of the time variable and the dummy variable of pastoral counties is also significant, but negative, in the 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 cattle population, mutton output and beef output models. It indicates that cattle population, mutton

Figure 4.9 Relationship between land tenure reform and livestock production output and beef output increase less in the pastoral counties than in the semi-pastoral counties with technology development. Sheep population increases significantly without a distinction between In Figure 4.9, the indexes of sheep population, cattle population, mutton output and beef output are all pastoral or semi-pastoral counties. In general, it appears that the development of livestock production decreasing with the number of years that the land tenure reform is implemented, and the reduction is is slower in the pastoral counties than in the semi-pastoral counties over time, with the exception of the fastest in the first years of the reform while it slows down in later years. Moreover, the decrease in sheep development of the sheep population which does not differ between county types. Notably, Figures 4.6 population is faster than that of mutton output, indicating that land tenure reform had a more negative and 4.7 showed that livestock production experienced a geographic shift from pastoral areas to the crop impact on the sheep population than on mutton output. Similarly, land tenure reform has affected the farming areas. Our results extended this finding to the specific comparison between pastoral counties cattle population more than beef output as the decrease in cattle population is faster than that of beef and semi-pastoral counties. That is, livestock production, with exception of the sheep population, has output. These results indicate that the implementation of land tenure reform promotes the development experienced a geographic shift from semi-pastoral counties to pastoral counties in the pastoral areas. of livestock productivity in pastoral areas although it constrains the increase in livestock population. Moreover, the mutton output (sheep population) decreases faster than beef output (cattle population). 4.6 Discussion The model results demonstrate that the implementation of land tenure reform has significantly It is interesting to compare Figure 4.9 with Figures 4.6 and 4.7 – which present the increasing trends in constrained the increase of livestock population and meat output in the pastoral areas of Inner Mongolia. the sheep population, mutton output and beef output after 1985. The seemingly contradicting trends of This finding can be explained in various ways. On the one hand, the implementation of land tenure livestock production in these figures can be explained by the effects of the other factors considered in reform may have effectively controlled the increase in livestock population, as was the aim of land our models. For instance, the producer price has a significant and positive effect on livestock production. tenure reform, which expected to internalise the social costs of overgrazing by privatisation and thereby The mutton price elasticity indicates that the sheep population in year t will increase by 0.23% when avoid the tragedy of the commons. The increase in meat output is correspondingly restricted due to the 79 80

First, we focus on the coefficients of coefficients onthe focus we First, The mutton price elasticity indicat produc the our models.For instance, explained be can livestock productioninthesefigures af outputandbeef the sheeppopulation,mutton It isinteresting tocompareFigure 4.9with Figure muttonoutput (sheeppopulation)decrea Moreover, the inpastor of livestockproductivity output. These results indicate that the implementati cattle populationmorethanbeefoutputasthedecrease impact Similarly, than onthesheep population onmuttonoutput. tenurethe land affected reform has that ofmuttonoutput,indicapopulation isfasterthan fastest inthe firstyears ofthe reform while itslows cattle population, sheep of theindexes In Figure 4.9, as26. horizontal-axis years sa privatised until2008was26 inourresearch refo year landtenure of the first shows th axis been vertical implemented and the presents axis horizontal 9.The Fig. in as depicted decreasing withthe numberyears of that the land four models. Based on the coefficients of of thecoefficients Based on four models. livestock on over production time. Allofthese coefficients are significant and negative inthe reform 100 105 Index (Year 1=100) Index 75 80 85 90 95

1 2

Figure 4.9 Relationship between landte Figure 4.9Relationshipbetween 3 4 5 6 7 8 9

10 Consideringlong equals 100). the rm al areasalthough itconstrainsth sta h he ouaini ertwl nraeb .3 when year population in sheep twillincreaseby 0.23% the es that

er price has a significant and positive effect onlivestockproduction. and positiveeffect significant a has price er 11 ݈݋݃ 12 ( ݈݋݃ 13 ܴ 14 ௜௧ିଵ ( ܴ )

15 ௜௧ିଵ and e livestock production index (livestockproductionin e livestockproductionindex . n . which present the increasing in trends s 4.6 and 4.7 –

16 has reform years thelandtenure that thenumberof 79 down inlateryears. Moreover, the decrease insheep on of land tenure reform promotes the development promotesthedevelopment on oflandtenurereform tenure reform isimplement ) nure reform andlivestockproduction reform nure 1( population,muttonoutputa 17 and ting that land tenure reform had a more negative a morenegative reform had ting thatlandtenure Τ mple, we determined th mple, wedetermined ter 1985. The seemingly contradicting trends of of trends Theseemingly contradicting ter 1985. by the effects of the other factors considered in considered otherfactors ofthe effects by the ses faster than beef thanbeef ses faster 18 ܴ in is cattle faster population than that of beef 1( ௜௧ିଵ

19 Τ

20 ܴ ) ௜௧ିଵ 21 tenure the land of , indicating effects the

22 ) , quasi-concave curves are obtained obtained are curves quasi-concave , e increase inlivestock population.e increase

23 rights formal use for est duration 24

output (cattlepopulation). 25 26 e maximum yeare maximum the of ed, and the reduction is ed, andthereduction nd beef output are all are nd beefoutput Years beef output beef mutton output cattle population sheep population will increase by 1.54% if the totalgrasslandarea will increase 1.54%if by andpositiveeffect condition alsohasasignificant in price beef the when (0.19%) 0.18% by increase yearwill t (beef output)in cattle population 1%. Similarly, year increases by t-1 the muttonpricein projects. This resulted in a net decline of 6.38mil 1.68 million landwas rehabilitated hectaresof period, thesame During intootherpurposes. transformed land,orbeing unused being into degraded theavailable of and 2002,3.1% 1991 between example, inreality,However, ofthegrasslands 90% still are Ingrasslandattempt to improvebycondition, an 1%. yearoutput) in th twillincrease1.17% (1.67%)if by grassland) (Qu et al., 2011). Ingrassland) (Qu etal.,2011). this output andbeefincrease lessinthepastoral mod output cattle population, mutton output and beef ofpastoral thedummy variable and of thetimevariable effects arerespect stronger tomeat with outputth pos affects livestockproduction technology development significan are variable thetime of The coefficients production. is slower in the pastoral counties is slowerinthepastoralcounties Ingeneral, it counties. semi-pastoral pastoral or technology Sheeppopulationincreases development. avoid the tragedy of the commons. The increase in meat output is correspondingly restricted duetothe meat restricted outputiscorrespondingly in Theincrease avoid thetragedy ofthecommons. so the tointernalise expected tenure reform, which reform controlled effectively may have the increase This findingin various waysOn theone hand, theimplementationoflandtenure can be explained . mea livestock population and constrained increase of the The model results demonstrate that the impleme 4.6 Discussion co shiftfromsemi-pastoral geographic a experienced Thatis,livestoc counties. and semi-pastoral comparison results this extended finding thespecific Our farming areas. to crop tothe pastoralareas geographic shift from a livestockproductionexperienced and 4.7showedthat di populationnot which does development the sheep of

than in the semi-pastoral counties than inthesemi-pastoral regard,grasslands thedeteriorating k production, with exception of the sheep population, has k production,withexception ofthesheeppopulation, appears that the development of livestock production of livestock development that the appears in year Sheeppopulation(mutton in t-1increasesby 1%. an for the livestock population. The interaction term an for thelivestockpopulation.Theinteraction cial costs of overgrazingby privatisation and thereby 80 80 on livestock production. Cattle population in yearin t onlivestockproduction.Cattlepopulation lion hectares (around 2.4% lion hectares(around 2.4% into natural grassland through eco-environmental into natural grassland through eco-environmental degraded tovariousdegrees (Mei etal.,2013). For t and positive in four models.It indicates that the China has implemented various ecological projects. projects. variousecological China hasimplemented ntation oflandtenure reform has significantly countiesthaninthe se e goodgrassland quality yeararea in t-1increases unties to pastoral counties in the pastoral areas. inthepastoralareas. unties topastoralcounties inlivestockpopulation,aswas theaimofland year t-1 increases by 1%. The better grassland grassland The better 1%. byyear increases t-1 ffer types. countyffer Figures between Notably, 4.6 els. It cattlepopulation, mutton that els. indicates t output in the pastoral areas of Inner Mongolia. Mongolia. Inner of areas t outputinthepastoral counties is also significant, but negative, in the in butnegative, counties isalsosignificant, significantly without a distinction between without adistinctionbetween significantly natural grassland was lost, either through its either through wasgrassland lost, natural itively, the positive beand seen it that can over time, with the exception of the ofthe time,withtheexception over could haveimpede between pastoral counties between pastoral mi-pastoral counties with mi-pastoral of the available natural d the livestock d thelivestock

4 constraints on livestock population. On the other hand, the spatial and social boundaries stemming from development and ecological policy. One challenge for future work is therefore to ascertain the specific land tenure reform have impeded livestock production because fragmentation and loss in mobility effects of technological improvements and ecological policies on livestock production in pastoral areas. increase the costs of feeding and reduce herders’ ability to withstand natural risks (e.g. Li et al., 2007). Moreover, the data on the grassland quality only represents the canopy cover, but does not indicate its These results are in line with findings from Mongolia where livestock production declined and poverty edibility by animals, which is a serious limitation related to the measurement of the grassland conditions. rose sharply after grassland privatisation in the 1990s (Griffin, 1995). The model results also illustrate Furthermore, the accuracy of Chinese livestock statistics is questionable (Waldron et al., 2007). The that the decrease in livestock numbers and meat output occurred especially in the early years of the land statistical collection of data on the livestock population is extremely difficult in China because small reform and that this effect weakened over time, which may be attributed to the gradual adaptation of rural households raise livestock on a small scale and supply chains are dominated by countless small production patterns to the privatised grasslands. Moreover, the constraining effects on the increase of traders and processors, unlike the centralised slaughter and auction systems that facilitate statistical livestock production could be a potential reason to explain why land tenure reform was so slow to be collection in developed countries (Waldron et al., 2007). In this regard, our research findings should be implemented and is still in progress after 30 years. This raises doubts about the adaptability of extending useful for detecting general trends if not specific numbers. policy measures designed for crop farming areas to the pastoral areas. Although the land tenure reform 4.7 Conclusions promoted agricultural production successfully in the crop farming areas of China, the pastoral areas may have produced different results. This paper explores the period during which the household-based assignment of grassland use rights has been implemented in order to describe the progress of land tenure reform and its impact on the Compared with the constraining effects of the land tenure reform, market forces have a strong positive grasslands of China. We focus on comparing the development of livestock production between pastoral effect on livestock production. The passing of time implies technological development, and also areas and crop farming areas under the land tenure reform and evaluating the effect of land tenure reform promotes the increase in livestock production, although more slowly in pastoral counties than in semi- on livestock production. Livestock production was presented by sheep population, cattle population, pastoral counties. This points towards potential approaches for stimulating livestock production in mutton output and beef output. The descriptive analysis shows that land tenure reform was implemented pastoral areas. In addition, improved grassland conditions also have positive effects on livestock progressively in the pastoral areas, rather than the case of the cropland areas where household-based production, which suggests that grassland conservation is necessary not only to improve the ecosystem, use rights were completed overnight (Banks et al., 2003; Ho, 2000). It appears that the increase of but also to develop sustainable livestock production. livestock production was faster in crop farming areas than in pastoral areas, while conversely the

Given the controlling effects of land tenure reform on the increase in livestock population and the development of livestock productivity was slower in crop farming areas. A fixed effects model is facilitating effects on improving livestock productivity, it is recommended that the government employed to estimate the impacts of the land tenure reform on livestock production based on a dataset supervises the implementation of assigning grassland use rights to individual households and clarifies of 45 counties in the pastoral areas of Inner Mongolia from 1985 to 2008. The model controls for factors fuzzy grassland boundaries between households. However, pastoralism that embraces common use may that obscure the relationship between livestock production and land tenure reform, such as market forces, also have benefits for the sustainable development of animal husbandry in pastoral areas due to the grassland condition, technological development and environmental heterogeneity among counties. The advantages of the mobility, flexibility and reciprocity (e.g. Fernandez-Gimenez, 2002). In this regard, model results provide quantitative evidence that land tenure reform has put a ceiling on livestock the privatisation of grassland property rights or use rights may be a precondition to ensuring cooperative production, but this constraining effect is unable to offset the impact of other factors that accelerate the use based on stable property rights or use rights in practice. In fact, recent institutional innovations in increase in livestock production, which explains the actual increase in livestock production. Moreover, pastoral areas of China has followed two distinct pathways: the strict implementation of clarifying the constraining effect of land tenure reform on the increase in livestock production decreases with the grassland use rights to individual households on the one hand, and the encouragement of co- number of years since the implementation of land tenure reform and ultimately disappears. Remarkably, management and cooperative use of grasslands based on substantive community participation on the the constraining effect is stronger on the increase in livestock population than on that of meat output, other hand (Banks, 2003). which indicates that land tenure reform stimulates the development of livestock productivity. With respect to the other factors, it appears that the cattle population tends to be affected by the quantity of As a final note, we would like to mention a number of limitations of our analysis that present challenges grassland, while the sheep population and mutton output are more affected by the quality of grassland. for future research. Our use of the time variable is a simplified way to capture the effect of technological 81 82

supervises the implementation of assign facilitating onimproving effects livestock producti Given the controlling effects of land tenure reform on the increase in livestock population and the but alsotodevelopsustainablelivestockproduction. thatgrassl whichsuggests production, pastoral Inareas. addition, improved grassland condi pastoral counties. This points to livestock production,although in promotes theincrease ti Thepassing of effect onlivestockproduction. Compared with the constraining of the land te effects may different results. produced have promoted agriculturalthe successfully production in reform the landtenure policy forfarmingthepastoral areas. crop areas Although to designed measures years. after 30 implemented andisstillinprogress to ex reason potential be a could livestock production production patterns to the privatise wh time, over reformand thatthiseffectweakened numbers that thedecreaseinlivestock privatisationinthe1990s grassland rose sharply after These results are inlinewithfindings M from al., 2007).Li et risks(e.g. ability withstandnatural herders’ to and reduce feeding costs of the increase impeded have reform land tenure constraints onlivestockpopulation.Ontheother hand, grassland userights toindividualhouseholds has followedpastoral of Chinatwo distinct areas pa rightsuse basedonstableproperty in oruserights the privatisationofgrassland property righ rights oruse In 2002). thisregard,advantages ofthemobility, flexibilitytyFernandez-Gimenez, (e.g. andreciproci sust the for also havebenefits Ho betweenhouseholds. fuzzygrassland boundaries for future research. Our use of the time variable is a si useofthetimevariable is Our for futureresearch. As afinalnote,we wouldliketo mentionanumber 2003). other hand(Banks, managementgrasslands use of and cooperative ba

ainable development of animalainable husba development wards potential approaches forwards potential stimulating approaches livestockproductionin d grasslands. Moreover, the constraining effects on the increase of grasslands. ontheincrease constrainingMoreover, effects d the livestock production because fragmentation andlossinmobility fragmentation because production livestock and meat output occurred especially in the early years the land of the early occurred in especially output and meat and conservation isnecessary not and conservation ing grasslanding userights toindi ongolia where livestockproduc and on thethe one encouragement hand, of co- This raises doubts about the adaptability of extending extending of about theadaptability This raises doubts 81 81 sed onsubstantivecommunity onthe participation of limitations of our analysis that present challenges of limitations ofouranalysisthat present challenges practice. In fact, recent institutionalInpractice. innovations recent fact, in me implies technological development, and also and me impliestechnological development, wever, pastoralism that embraces common use may pastoralismthatembracescommonuse wever, ich may be attributed to the gradual adaptation ofadaptation to the gradualattributed ich may be mplified way theeffect oftechnological tocapture nure reform, market forces have a strong positive astrong have marketforces positive nure reform, vity, itisrecommended thatthegovernment plain why land tenure reform was so slow to be slowto was so reform plain whyland tenure crop farming areas of China, the pastoral areas crop farming China,thepastoralareas areas of (Griffin, 1995). The model results also illustrate (Griffin, 1995). The thways: of clarifying thestrictimplementation the spatial and social boundaries stemming from boundariesstemming the spatialandsocial ts may apreconditiontoensuring be cooperative tions also have positive effects on livestock effects onlivestock positive tions alsohave more slowly in pastoral counties thaninsemi- counties pastoral in slowly more ndry in pastoral areas due to the duetothe pastoral areas ndry in only to improve theecosystem, toimprove only vidual householdsandclarifies tion declined and poverty grassland, while the sheep population andmutton grassland,output ar whilethesheep cat that the it appears respect other factors, to the which indicatesthatlandtenure reform stimulates the constraining isstronger effect inlivestockpopulation on theincrease than onthatofmeat output, yearsnumber of implementationof sincethe land tenure the constraining landtenurereform effect of onthe inincrease livestock the production,which explains unableto is butthis constrainingproduction, effect model results providequantitativeevidence that grassland condition,technological and development that obscure between livestockproducti therelationship of 45countiesinthepastoralarea employed to impactestimates ofthelandtenure the reform onlive in productivity slower was development oflivestock arealivestock productionwasfaster incropfarming et (Banks overnight were completed use rights thanth rather areas, progressively inthepastoral anal mutton output The and descriptive beef output. Livestock was on livestockproduction. production areas and cropfarming areas under the land refortenure grasslands focus ofChina.We the oncomparing orderhas been implemented the to describe in whichthe period during the explores This paper 4.7 Conclusions general useful for detecting numbers. trends if not specific rural householdsraise livestockon onthe statistical collectionofdata Chineselivestock of accuracyFurthermore, the collection in developed countries collection indeveloped unlike thecentralisedslaugh traders andprocessors, edibility isaseriouslimitation by animals,which grassland represe only quality dataonthe Moreover, the effects of pol improvements andecological technological policy. for fut Onedevelopment and ecological challenge

s of Inner to2008.Thes of Mongoliafrom 1985 modelcontrolsfor factors (Waldron et al., 2007).In this regard, our research findings should be livestock population isextremely livestock population a smallscaleandsupply chainsar progress of land tenure reform anditsimpactonthe reform oflandtenure progress related the grassland conditions. tothe measurementof development of livestock production ofdevelopment livestock between pastoral statistics isquestionab statistics al., 2003; Ho, 2000). It appears that the increase of that theIt increase al.,appears 2003; Ho, 2000). tle population tends to be affected by the quantity of the quantity affectedby populationtendstobe tle e case of the cropland areas where household-based where household-based areas the cropland case of e household-based assignment of grassland rights of use assignment household-based 82 82 offset the impact of other factors that accelerate the the ofother impact offsetthe factors thataccelerate ysis shows that land tenure reform was implemented was ysis reform tenure thatland shows land tenure reform put aceiling has on livestock increase in livestockproductionincrease decreases withthe actual increase in livestock production. Moreover, production.Moreover, increaseinlivestock actual environmental heterogeneity among counties. The environmental heterogeneity counties. among presented by sheep population, cattlepopulation, population, presented by sheep the developmentoflivestock productivity. With ter and auction systems thatfacilitate systems statistical andauction ter m and evaluating the effect of land tenure reform reform land tenure effect of the evaluating and m crop farming areas. A fixed effectsfarming model is A fixed crop areas. on and land tenure reform, such as market forces, s than in pastoral areas, while conversely the conversely areas, while s thaninpastoral reform andultimately disappears.Remarkably, nts the canopy cover, but does not indicate its butdoesnotindicate cover, canopy nts the icies on livestock production in pastoral areas. production inpastoralareas. icies onlivestock ure work is therefore to ascertain the specific ure work istherefore toascertainthespecific e more affected byegrassland. moreaffected of thequality stock production based on a dataset stock productionbasedonadataset difficult in China because small inChinabecause difficult e dominated bye dominated countlesssmall le (Waldron et al., 2007). The The etal.,2007). le (Waldron

4 The factor of technological development impacts on the increase in meat output more than that on the ACKNOWLEDGEMENTS: increase in livestock population, and the influence is more evident in the semi-pastoral counties than in The funding for this research was supported by a grant from the National Natural Sciences Foundation the pastoral counties. It can be seen that technological development also prompts the development of of China (71333013). The authors would like to thank Xiangzheng Deng, Fujiang Hou, Dan Wang, livestock productivity. Yangjie Wang and Jie Zhou for their useful suggestions and help. We are also grateful to Nico Heerink The development of livestock productivity that is caused by the implementation of land tenure reform and Xiaoping Shi for their helpful comments that contributed to improve the quality of this paper. is different from that due to technological development. More specifically, land tenure reform in pastoral areas has transformed the production patterns of pastoralists through the privatisation of grasslands (Ho, 2000). For instance, the indigenous people in Inner Mongolia employed nomadic management and seasonal transhumance to graze animals in areas where pastures and water were available (Wang et al., 2013). Since the spread of the land tenure reform on the grasslands, a growing number of households have built physical boundaries (such as fences) in order to prevent others from using their grasslands, which implies that mobile grazing is no longer possible and most areas of grassland are in small-scale private use (Hua and Squires, 2015). The traditional nomadic production pattern has therefore gradually been displaced by a sedentary pattern. Our research results essentially indicate that this transformation, resulting from land tenure reform, has prompted livestock productivity in the pastoral areas of Inner Mongolia. In other words, land tenure reform entailing grassland privatisation probably stimulates herders to put more effort into improving the output per animal, instead of increasing the amount of animals, which is beneficial to the development of livestock production systems in pastoral areas. However, the constraining effects of grassland privatisation on the increase in livestock production could also be a possible reason to explain why land tenure reform was implemented with difficulty on grasslands, especially in terms of the lag between the reform and the actual adoption of private use.

83 84

use. actual adoptionofprivate implemented withdifficultygrasslands, especiall on in livestock production could al systems Howeve in pastoral areas. amount of animals,whichisbene of increasing the privatisation probably herde stimulates in the pastoral Innerareas of Mongolia. In other words, land reformtenure entailing grassland resulting fromindicate thatthistransformation, land a by displaced gradually been pattern hastherefore and private use (Hua in small-scale grassland are usingwhich impliesthatmobile theirgrasslands, gr number of households built physical inorder have others from toprevent (such boundaries asfences) of Since thespread et al.,2013). available (Wang animals in areas tograze transhumance management andseasonal pastureswhere and water were indige For the instance, grasslands 2000). (Ho, th pastoral areashastransformed inreform tenure land opment. Morespecifically, devel totechnological thatdue from is different productivity that is The development of livestock livestock productivity. that technol seen can be It counties.the pastoral in counties than in the semi-pastoral inis increase livestockmore and evident the population, influence The factor technological development of impacts on th

so beapossiblereason toex r, the constraining effects of grasslr, theconstraining effects e productionpatternsofpastorali rs toputmoreeffortintoimprovi ogical development also prompts the development of prompts thedevelopment also ogical development nous people in Inner Mongolia employed nomadic employedInner nomadic Mongolia nous peoplein the land tenure reform on the grasslands, a growing a the the grasslands,land tenure reform on caused by implementation of landthe tenure reform 83 83 Squires, 2015).Thetraditionalnomadicproduction tenure reform, has prompted livestock productivity livestock productivity reform,tenure has prompted sedentary pattern. Our research results essentially results Our research pattern. essentially sedentary ficial tothedevelopmen y intermsthe ofthelag betweenthereform and azing isnolonger possibleandmostareas of e increase in meat output more than that on the morethanthat onthe e increaseinmeat output plain whyland tenurereform was and privatisation onthe and privatisation increase sts through the privatisation of of privatisation sts through the ng theoutputperanimal,instead t of livestock production t oflivestockproduction toimprovethequality helpful commentsthatcontributed Shifortheir and Xiaoping ofthispaper. Zhou for suggestions Yangjie their useful and Jie Wang FujiangHou, Dan authorswouldliketothankXiangzheng Wang, Deng, of China(71333013).The The funding for thisresearch was supportedbya grant from the National NaturalFoundation Sciences ACKNOWLEDGEMENTS:

84 84 and help. We are also grateful to Nico Heerink grateful toNicoHeerink are We also andhelp.

4 5. THE IMPACT OF ECOLOGICAL CONSTRUCTION PROGRAMS ON GRASSLAND CONSVERVATION IN THE PASTORAL AREAS OF CHINA24 Chapter 5 ABSTRACT: A series of Ecological Construction Programs have been initiated to protect the condition of grasslands in China during recent decades. However, grassland degradation is still severe and conditions have not been restored as intended. This paper aims to empirically examine the effectiveness of Ecological Construction Programs for protecting the grassland condition in the extensive pastoral areas of China. We focus on one Ecological Construction Program that has been implemented widely The impact of Ecological Construction Programs on the grasslands, the Subsidy and Incentive System for Grassland Conservation (SISGC). The Normalized Difference Vegetation Index (NDVI), measured with remote sensing technology, is used to on grassland conservation in the pastoral areas quantify the grassland condition between 2001 and 2014. With data from 54 counties in the pastoral areas of Inner Mongolia, we estimate the impact of SISGC on the grassland condition. A fixed effects of China model is employed to control for livestock production, climate, time trends and time-invariant heterogeneity between counties. The model results provide quantitative evidence that the condition of

the grasslands has improved significantly due to SISGC, but that the effectiveness of SISGC was offset to some extent by other socio-economic and climate factors, such as increased producer prices and high temperature. This may explain why the actual grassland degradation has not been prevented as effectively as was expected. In addition, the impact of SISGC was stronger in counties with a worse initial grassland condition. Furthermore, the effects of producer prices and climate changes were also more pronounced in these counties.

KEYWORDS: Ecological Construction Programs, grassland degradation, NDVI, fixed effects model, Inner Mongolia

24 Paper by Min Liu, Liesbeth Dries, Wim Heijman, Jikun Huang, Xueqin Zhu, Yuanning Hu, and Haibin Chen, has been accepted by Land Degradation and Development. 85 86

The impactofEcologicalConstructionPrograms on grasslandconservationinthepastoralareas Chapter 5 of China 85 85

accepted by CONSVERVATION IN THE PASTORAL AREAS OF CHINA CONSVERVATION INTHEPASTORALAREAS 5. THE IMPACT OF ECOLOGICAL CONSTRUCTION PROGRAMS ON GRASSLAND 24 model isemployedtocontrolforlivestockpr Inner the of impact we estimate Mongolia, of areas 2014. and between2001 condition grassland quantify the measur Index (NDVI), Vegetation Difference Normalized SubsidyIncentiveon thegrasslands, theand Sy ononeEcological Constr focus areas ofChina.We ConstructionPrograof Ecological restored conditions havenotbeen grasslandsofduring recent inChina grasslandHowever, decades. degradation isstillsevereand to some extent byto someextent othersocio-econo grasslandsthe improved has significantly that theeffectiveness toSISGC, but offset due ofSISGC was results Themodel counties. between heterogeneity Ecological of ABSTRACT: Aseries Inner Mongolia Ecological gras ConstructionPrograms, KEYWORDS: inthesecounties. more pronounced initial grassland theeffects Furthermore, condition. addition, theimpact In expected. as was effectively as prevented been hasnot degradation grassland actual why the explain may This temperature. Paper by Min Liu, Liesbeth Dries, Wim Heijman, Jikun Huang, Xueqin Zhu, Yuanning Hu, and Haibin Chen, has been been has Chen, Haibin and Hu, Yuanning Zhu, Xueqin Huang, Jikun Heijman, Wim Dries, Liesbeth Liu, byMin Paper Land Degradation and Development. as intended. paper This aims to empirically examine the effectiveness ms for protecting the grassland c grassland the protecting for ms mic and climate factors,suchas mic and Construction Programs have been Programs havebeen Construction oduction, climate,ti 86 SISGC on the grassland condition. A fixed effects effects fixed grassland condition.A onthe SISGC provide quantitative evidence that the condition of conditionof the that evidence provide quantitative uction Program thathasbe stem for Grassland Cons stem forGrassland of producer prices and climate changes were also also climatechanges were and of producerprices of SISGC was stronger in counties with a worse incountieswithaworse of SISGC wasstronger sland degradation, NDVI, fixed effectsmodel, NDVI, fixed sland degradation, ed with remote sensing technology, is used to isused technology, with remotesensing ed With data from 54 counties in the pastoral thepastoral 54counties in datafrom With 24

increased producer prices and high prices andhigh producer increased ondition pastoral in the extensive initiated to protect the condition thecondition initiated toprotect me trends and time-invariant me trendsandtime-invariant ervation (SISGC). The ervation (SISGC). en implemented widely

5 5.1 Introduction wider areas that involve most of the available natural grasslands of China and provides herders with more ecological compensations. Moreover, its monitoring system for policy implementation has been Grasslands are important as a feed source for livestock, as a habitat for wildlife, for environmental improved with the use of more targeted monitoring staffs (Guan Hu Yuan) that are employed by local protection and for the in situ conservation of plant genetic resources (FAO, 2008). However, rapid governments. increases in livestock populations have increased pressures on the world’s grasslands, leading to widespread deterioration, particularly in arid and semi-arid environments (Suttie et al., 2005). China In general, China’s Ecological Construction Programs include two categories of measures for grassland has around 392 million hectares of grasslands, accounting for 12% of the world’s grasslands and 41.7% conservation, namely sowing grass or planting trees in the eco-fragile areas to restore the degraded of the national land area (Fan et al., 2008). Nearly 80% of these grasslands are in arid and semi-arid grasslands and reducing livestock grazing to prevent grassland degradation (Ho and Azadi, 2010; Kou regions (National Bureau of Statistics of China, 2009). Approximately 17 million herders and agro- et al., 2016). This coincides with the measures for preventing desertification in China: controlling herders maintain their livelihoods on the grasslands in China (Li et al., 2014). According to Chinese desertification by plantation and combating desertification by natural recovery through isolating the governmental reports, 10% of the total area of grasslands was degraded in the 1970s, increasing to 30% degraded area from external human influences (Miao et al., 2015). In practice, these two measures cover in the 1980s and 50% in the middle of the 1990s. By the 2000s, about 90% of the grasslands were a vast array of regulations (Waldron et al., 2010). Particularly, the reduction in grazing is targeted by degraded to various extent, with significant regional variation (Unkovich and Nan, 2008; Waldron et regulations determining stocking rate and grazing bans (either permanently, temporarily or seasonally) al., 2010). The manifestation of grassland degradation includes initial lowering of grassland (Qu et al., 2011). The stocking rate is based on the theoretical maximum of the available biomass productivity, fragmentation of grass cover, reduction in soil fertility, soil compaction, an increase in production from grasslands that can be consumed by grazing herds without impairing the capacity of unpalatable grass species or a combination of all of them (Feng et al., 2009; Li et al., 2013). Grassland the pasture to regrow the following year (Fernández-Giménez et al., 2011). degradation, especially in northern China, is considered to be the cause of serious environmental and ecological problems, such as Yangtze River floods, Yellow River droughts and sand storms of Beijing Despite the fact that Ecological Construction Programs have been widely introduced in the pastoral (Harris, 2010). Grassland degradation therefore threatens not only animal husbandry on the grasslands areas of China for more than ten years, grassland degradation is still severe (Waldron et al., 2010; Li but also the livelihoods of millions of people and the ecological security of China (Huang et al., 2013). and Huntsinger, 2011; Ministry of Agriculture of China, 2012; Li et al., 2014). Although the average To address these problems, national and local governments have initiated an ambitious series of vegetation coverage of China’s grasslands increased by 0.4% between 2014 and 2015, and there has Ecological Construction Programs for grassland protection in recent decades. been an increase of 3% since 2011, more than one third of the grasslands are still suffering moderate and serious degradation (Ministry of Agriculture of China, 2014; 2016). Some academic studies have In 1985, China passed the first national Grassland Law which explicitly stipulated the protection and therefore criticized the effectiveness of the Ecological Construction Programs on grassland conservation, improvement of grasslands (Ho, 2000). The Law devolved grassland use rights from the state and and attributed the causes of this ineffectiveness to inappropriate policy measures and failed collectives to individual households through long-term contracts, and introduced maximum stocking implementation (Li and Zhang, 2009). For example, some authors argue that the determination of a rates for livestock on the grassland areas. Since the serious drought in China in 1997 and the massive maximum stocking rate and various grazing bans are inappropriate management measures for the floods of 1998, additional Ecological Construction Programs have been introduced, for example, the conservation of grasslands in arid and semi-arid areas (Higgins et al., 2007). In particular, deterministic Conversion of Cropland to Forest and Grassland Program (also known as the Grain for Green Program), stocking rates are questioned because they largely ignore the grasslands' spatial heterogeneity and the Returning Grazing to Grassland Program, and the Program to Combat Desertification around Beijing climate variability (Li and Huntsinger, 2011). Moreover, it is claimed that appropriate grazing is and Tianjin, which aimed at preventing grasslands from degradation and were actively implemented in beneficial to plant succession processes (Zhang et al., 2015). Deterministic stocking rates have also led the 2000s (Liu et al., 2010; Hua and Squires, 2015). Since 2011, another major Ecological Construction to resistance from local herders that are unwilling to limit their own livestock numbers based on a top- Program, the Subsidy and Incentive System for Grassland Conservation (SISGC), has been initiated in down government measure (Li and Zhang, 2009). High supervision costs of the implementation of the eight pastoral provinces of China, including Inner Mongolia, Xinjiang, Tibet, Qinghai, Sichuan, Gansu, policy make that deterministic stocking rates and grazing bans are difficult to sustain effectively for a Ningxia and Yunnan (State Council of the People's Republic of China, 2010). Compared to the long period, especially in extensive grassland areas with a scattered population distribution (Li and Ecological Construction Programs for grassland conservation before 2011, the SISGC includes much Zhang, 2009; Waldron et al., 2010). Considering these criticisms and the fact that the grassland

87 88

Ecological Construction Programs for grassland CouncilofthePeople' Ningxia andYunnan (State M Inner China, including of eight provinces pastoral Program, theSubsidyIncentive and System for Grassland Conservation (SISGC), in hasbeeninitiated and Squires, 2015 Hua 2010; et al., the 2000s(Liu atpreventin and Tianjin,whichaimed Grassland Program,the Returning andth Grazing to Conversion ofCroplandtoForest a floods of1998,additional Ecological Construction P rates Since grassland for th areas. livestockonthe through long- collectives toindividualhouseholds improvementLaw (Ho, ofgrasslands 2000).The devolve InLa first 1985, national China Grassland passed the Ecological Construction Programsgr for theseproblems,nationa To address but alsothelivelihoodsofmillionspeople andChina (Huang theecological of security etal.,2013). thereforethr degradation Grassland 2010). (Harris, Beijing andsandstormsof floods, Yellow Riverdroughts River asYangtze such ecological problems, consid northernChina,is degradation, especially in grass oracombinationofallth unpalatable species productivity, ofgrass cove fragmentation ofgrasslandal., 2010).Themanifestation degradat and Nan, et 2008; Waldron with significantregi variationextent, degraded to (Unkovich various onal the1990s. of themiddle 50%in in the1980sand grasslagovernmental 10% of the reports, total area of grasslands in livelihoodsonthe herders maintaintheir ofStatisticsChina, Bureau regions(National of thenationallandarea (Fan etal.,2008).Nearly80% has around 392 million hectares arid in particularly deterioration, widespread inlivestockpopulationsincreases haveincreased protection andfor thein situconservation ofpl importantasafeed Grasslands are 5.1 Introduction

of grasslands, accounting for 12% of nd GrasslandProgram Grainfor (also Green knownasthe Program), sourcefor livestock,asahabita l and local governments have initiated an ambitious series of ambitious series an governments initiated have l and local g grasslands from actively degradation implemented and were g ingrasslands assland protection in recent decades. decades. inrecent assland protection r, reduction insoilfertility,r, reduction in soilcompaction, anincrease and semi-arid environments (Suttie et al., 2005). China (Suttie etal.,2005). environments semi-arid and conservation before2011,theSISGCconservation includesmuch ant genetic resources (FAO, 2008). However, rapid 2008).However,rapid ant genetic (FAO, resources ). Since 2011, another major Ecological Construction major another ). Since 2011, 2009). Approximately 17 million herders andagro- millionherders 2009). Approximately 17 87 87 e seriousdrought inChina1997andthemassive eatens not only animal husbandry on the grasslands grasslands onthe animal husbandry eatens notonly e Program to Combat Desertification around Beijinge Program around toCombatDesertification term contracts, and introduced maximum stocking maximum andintroduced term contracts, pressures ontheworld’s grasslands, leading to ongolia, Xinjiang, Tibet, Qinghai,Gansu, Sichuan,ongolia,Xinjiang, Tibet, ered tobethecause By the 2000s, about 90% of the grasslands were ofthe about 90% 2000s, By the nds was degraded in the 1970s, increasing to 30% nds was degraded inthe1970s,increasing to30% s Republic of China, 2010). Compared to the 2010).Compared tothe s RepublicofChina, rograms the introduced, have been for example, em (Feng et al., 2009; Li em (Feng etal.,2013).Grassland etal.,2009; w which explicitlyw which stipulatedtheprotectionand China (Li et al., 2014). According to Chinese According toChinese 2014). et al., China(Li of these grasslands are in arid and semi-arid grasslands are inaridandsemi-arid ofthese ion includes initial lowering of grasslandion includesinitiallowering of d rights grassland use from the state and the world’s grasslands and 41.7% t for wildlife, for environmental of seriousenvironmentaland wider areasthatinvolve most ofth Zhang, 2009; Waldron et al., 2010). Considering these criticisms and the fact that the grassland grassland the that the fact criticisms and Consideringet al.,Zhang, these 2010). 2009; Waldron distribution (Liand with a scattered grasslandareas population extensive in especially long period, policy stocking make that deterministic rates and grazin Zhang, High and down government 2009). supervisioncosts oftheimplementation (Li measure to resistance from local herders that are unwilling proce toplantsuccession beneficial Huntsinger,More climate variabilityand (Li2011). because they largely stocking questioned ratesare conservation inarida ofgrasslands stockingmaximum grazing andvarious rate bans ForimplementationZhang, 2009). example, (Li and and attributed of thethis causes ineffectivene Ecologica thereforeof the the effectiveness criticized Agricultur(Ministry of degradation and serious th than one 2011, more since 3% increase of been an vegetation coverage of China’s gr Ch and Huntsinger, 2011; Ministry of Agriculture of Li al., 2010; years, degradation (Waldronet isstillsevere grassland than ten for more areas ofChina Construction Progra factthatEcological Despite the yearthe pasture etal.,2011). following toregrow (Fernández-Giménez the thatcanbeconsumedby production fromgrasslands (Qu etal.,2011).Thestocking isbasedon rate regulations determininggrazing stocking rate and ba a vastarray ofregulations (Waldronetal.,2010).Par (Miao humaninfluences external from area degraded and combatingdesertification by plantation desertific et al.,2016).Thiscoincideswiththemeasures for preventing inChina:controlling desertification grazing to pre grasslands and reducing livestock thedegraded torestore areas eco-fragile inthe trees planting grass or conservation, sowing namely general,In Construction Ecological Progra China’s governments. employed by local thatare Yuan) Hu (Guan monitoring staffs moretargeted use of improved withthe Moreover, hasbeen its monitoring policy implementation more ecological compensations. system for

asslands by increased 0.4%be nd semi-arid areas (Higgins etal., (Higgins areas nd semi-arid sses (Zhang et al., 2015). Determinist al., 2015). et (Zhang sses e available natural grasslands e of China, 2014; 2016). Some academic studies have studieshave academic Some of China,2014;2016). e vent grassland degradation (Ho and Azadi,2010;Kou grassland vent (Ho degradation to limit theirownlivestock numbersbasedon a top- 88 88 the theoreticalmaximum ms include two categories of measures for grassland for ofmeasures categories ms includetwo ss to inappropriate policy measuresss to inappropriate and failed are inappropriate management measures for the measures for are management inappropriate l Construction Programs on grassland conservation, l Constructionconservation, ongrassland Programs ignore the grasslands' spatial heterogeneity and heterogeneity spatial and thegrasslands' ignore ns (either permanently, temporarily or seasonally) temporarily orseasonally) ns (either permanently, et al., 2015). In practice, these two measures cover cover In thesetwomeasures practice, al., 2015). et ina, 2012; Li et al., 2014). Although the average average AlthoughLi the al., 2014). 2012; et ina, grazing herds without impairingof thecapacitygrazing herdswithout ms have been widely introduced in the pastoral pastoral inthe introduced widely been ms have ird of the grasslands aregrasslands stillsuffering moderate ird ofthe ticularly, the reduction ingrazingreduction by the istargeted ticularly, some authors argue that the somedetermination authors aargue of over, it is claimed that over, itis ation by recovery natural through isolating the g to sustain bansare effectivelyfor a difficult tween 2014and2015,theretween has of China and provides herders with of Chinaandprovidesherders with 2007). In deterministic particular, 2007). ic stocking rates have also led led also have stockingrates ic of the available biomass appropriate grazingappropriate is

5 condition is still deteriorating, despite the current policy measures, this paper will examine the grasslands of Inner Mongolia. Figure 5.1 illustrates the location of Inner Mongolia in China and the effectiveness of Ecological Construction Programs on grassland protection and discuss the potential distribution of its pastoral and semi-pastoral counties. factors that impact the policy effectiveness.

Some academic studies and governmental reports discuss the impacts of Ecological Construction Programs. A number of them focus on limited regions based on household-level or experimental field data (e.g. Gu and Li, 2013; Zhang et al., 2015). Others conduct spatial and temporal comparisons but without controlling for the possible impacts of other socio-economic and climate factors on the grassland condition. Some studies also have systematically discussed the impacts of particular aspects of eco-environmental policy interventions, including grassland laws (Nelson, 2006), grazing bans (Yeh, 2005), and herder resettlement policies (Dickinson and Webber, 2007), but they have mainly used qualitative analysis. In short, the existing literature lacks estimations on the effectiveness of Ecological Construction Programs for grassland conservation that are based on large-scale and long-term samples and control for the impacts of other potential factors through a quantitative analysis. This paper employs an econometric model based on data from 54 counties in the extensive pastoral areas of Inner Mongolia from 2001 to 2014 to disentangle the impact of one specific Ecological Construction Programs, namely the SISGC, on the grassland condition. In the following, the research region, data collection and methods are described. Next, the empirical results about the relationship between SISGC and grassland condition are presented and interpreted. Ultimately, we conclude with an assessment of the effectiveness Figure 5.1 Inner Mongolia and its pastoral and semi-pastoral counties of SISGC for grassland conservation based on the comparison between the estimated results from our model and the actual condition of the grassland. According to government reports, by the year 2000, 90% of the natural grasslands of Inner Mongolia had been degraded to some extent, reflected by the degraded composition of plant species, declining 5.2 Material and methods biodiversity, accelerated soil erosion etc. (Waldron et al., 2010; Briske et al., 2015). Moreover, the 5.2.1 Research region results of large-scale field ecological surveys highlighted that the average grassland biomass We conduct our empirical study based on Inner Mongolia, which is located in northern China and in productivity in Inner Mongolia has reduced from 1871 kg/ha in 1961 to 900 kg/ha in 2010 (Wang et al., which animal husbandry is the traditional and dominant agricultural industry. Most of the population 2013). The extensive grasslands of Inner Mongolia play an important role as ecological barriers for are indigenous Mongolians and are reliant on grazing animals on the natural grasslands in order to northern China. For instance, the colossal dust storms which rumbled through hundreds of cities and maintain a livelihood. Inner Mongolia belongs to the arid and semi-arid areas of China and contains villages of northern China and blanketed the sky of Beijing between 1998 and 2001 were said to have 21.7% of the area of China’s natural grasslands. Approximately 67% of the total land in Inner Mongolia originated from dryland areas and degraded grasslands mainly in Inner Mongolia (Wu et al., 2015). is classified as grassland, the majority of which can be sub-classified as temperate grassland (Angerer Subsequently, the pastoral areas of Inner Mongolia have become the typical regions for implementing et al., 2008). The current administrative divisions of Inner Mongolia include 102 counties, consisting various Ecological Construction Programs for grassland conservation in China since the early 2000s. of 33 pastoral counties and 21 semi-pastoral counties. The remaining 48 counties are dominated by crop Since 2011, the SISGC has been initiated in Inner Mongolia. Considering that the SISGC is currently farming or urban districts. The agricultural industry of pastoral counties is characterized by its extensive the most important large-scale Ecological Construction Program for grassland conservation and inherits grazing; natural grassland is the dominant land type. In semi-pastoral counties, both natural grassland the main management measures from earlier programs, namely the deterministic stocking rate and and cropland are the dominant land types, in which there is more intensive animal husbandry along with permanent grazing ban, we take it as a typical example to explore the effectiveness of Ecological cropping (Waldron et al., 2010). The pastoral and semi-pastoral counties include almost all of the natural 89 90

most all of the natural almostallof the natural toral countiesinclude andsemi-pas al., 2010).Thepastoral cropping (Waldronet th which in landtypes, dominant the are and cropland grazing; isthedominantland type. grassland natural farming The agricultural or urban districts. industryof pa counties. 21semi-pastoral and counties of 33pastoral administra current The et al.,2008). grassland,which can themajorityis classifiedas of 21.7% ofthearea ofChina’s Innermaintain alivelihood. Mongolia the belongs to on grazing reliant and are are Mongolians indigenous isthetraditionalanddom which animalhusbandry InnerMongol conductourempiricalstudy We on based 5.2.1 Researchregion methods5.2 Material and ofthegrassland. condition model andtheactual co grassland onthe of SISGC conservation based for Ultimately, interpreted. and we condition arepresented methods arethe empiricalresults described.Next, Inregion,collection andfollowing, condition. data theresearch the SISGC, grassland the onthe sp ofone from theimpact 2001to2014disentangle in modelbasedondatafrom 54counties an econometric and controlfor theimpactsofother potentialfactors samples and long-term Construction Programs that are on grassland large-scale based for conservation lacks estimations literature theexisting on In short, the effectiveness qualitative analysis. of Ecological 2005), andherder resettlement polic policy interventions, including of eco-environmental grassland condition.Some studies alsohavesyste without controlling for thepossibleimpactsof Li,data (e.g. Guand 2013;Zhang etal.,2015).Othe Programs. focus Anumber onlimited regi ofthem and governmentalSome di academic reports studies effectiveness. policy the factors thatimpact Programs Construction Ecological effectiveness of condition isstilldeteriorating, de

natural grasslands. Approxim spite the current policy the paper will examine measures, this tive divisions of Innertive divisions of Mongolia ies (Dickinson and Webber, 2007), and Webber, but theyies (Dickinson mainly have used matically discussedtheimpactsofparticular aspects about therelationship between SISGC andgrassland 89 89 other socio-economic and climate factorsthe other on socio-economic ons based on household-level or experimental field field orexperimental basedonhousehold-level ons inant agricultural industry. Mostofthepopulation through a quantitative analysis. This paper employs on grassland protection and discuss the potential and discuss on the grassland protection be sub-classified as temperate grassland (Angerer as temperate be sub-classified grassland laws (Nelson, 2006), grazing(Yeh, grassland (Nelson, 2006), bans laws ere is more intensive animal husbandry along with husbandry animal along is more intensive ere The remaining 48 counties are dominated by crop mparison between the estimated results from our estimatedresults from the mparison our between ecific Ecological Construction Programs, namely Construction Programs, Ecological namely ecific rs conduct spatial and temporal comparisons but In semi-pastoral counties, both natural grassland bothnatural counties, In semi-pastoral conclude with an assessment of the effectiveness effectiveness ofwith the anconclude assessment arid and semi-arid areas of China and contains arid and semi-arid scuss theimpactsofEc animals on the natural grasslands to inorder animalsonthenatural the extensive pastoral the extensive areas of Inner Mongolia storal counties is characterized by countiesischaracterized storal itsextensive ia, which is located in northern China and in and innorthernChina ia, whichislocated ately 67%ofthetotallan include 102 counties, consisting include102counties, ological Constructionological d inInner Mongolia permanent grazing ban,we takepermanent itasatypical ex and stocking rate namely the deterministic the mainmanagementmeasures programs, earlier from Constructi the mostimportantlarge-scale Ecological SinceInner 2011,theSISGC M hasbeeninitiatedin various Ecologicalgrassla ConstructionPrograms for areasSubsequently,Inner thepastoral of Mongolia grass and degraded areasoriginated dryland from of sky the blanketed Chinaand villages ofnorthern northern China.For instance,thecolossaldust pl Inner Mongolia grasslands of 2013). Theextensive productivityInner Mongolia in hasreduced from 1871kg/ha in1961to 900 kg/ha in2010(Wang etal., results ecological hi oflarge-scale surveys field biodiversity, 2010; soil Briske erosion al., etc. 2015). (Waldron accelerated Moreover, et et al., the th by reflected extent, tosome degraded had been reports,yearAccording by togovernment 2000, the distribution of its pastoral and semi-pastoral counties. counties. distribution ofitspastoralandsemi-pastoral Figure 5.1illustrate Inner Mongolia. grasslandsof

Figure 5.1 Inner Mongolia and its pastoral and semi-pastoral counties counties FigureInner anditspastoralsemi-pastoral Mongolia 5.1 storms whichrumbledthroughand hundredsofcities 90 90 lands mainlyInner in Mongolia (Wuetal.,2015). e degraded composition declining of plante degraded species, s the location of Inner Mongolia andtheInner inChina of s thelocation have becomethetypical regions for implementing ample toexplore the ofEcological effectiveness on Program for grassland conservation and inherits andinherits on Programgrassland for conservation Beijing between 1998 and 2001 were said to have tohave were said and 2001 1998 Beijing between ghlightedgrassland biomass thattheaverage 90% ofthenaturalgrasslandsInner of Mongolia ongolia. Considering thattheSISGC iscurrently nd conservation in China since the early 2000s. nd conservation inChinasincetheearly 2000s. ay an important role as ecological barriers for for barriers asecological ayan importantrole

5 Construction Programs in the extensive pastoral areas of China. It is reported by the local government Composite (MVC) method to get the NDVI of each year during 2001-2014 (Holben, 1986). Ultimately, that 67.3 million hectares of available natural grasslands are covered by SISGC, which includes almost the image layer of administrative divisions of Inner Mongolia is used to determine the NDVI for each all of the available natural grassland areas of Inner Mongolia. 30 million hectares of them belong to the county. area with the permanent grazing ban and the remaining 37.3 million hectares belong to the forage- Information about SISGC was provided by the Agriculture and Animal Husbandry Bureau of Inner livestock balance area. In practice, every county in the pastoral areas contains different areas of Mongolia. Moreover, interviews were conducted with local herders and public officers who are key grasslands which are subject to the permanent grazing ban, deterministic stocking rate or both of them. informants on the implementation of Ecological Construction Programs in Inner Mongolia to improve Almost all rural households in the pastoral areas have been covered by SISGC, and they have been our understanding of these programs and their possible impacts. The data on socio-economic indicators given subsidies (or incentives) as well as the obligation to comply with the permanent grazing ban (or was collected based on existing statistical data gathered by national and local Statistical bureaus. For the deterministic stocking rate). instance, the producer price of live sheep was collected based on the Annual Compilation of Cost- 5.2.2 Data collection benefit Data of Chinese Agricultural Products and deflated with the Producer Price Index of Agricultural Products of Inner Mongolia. The data on climate indicators were gathered from the Statistical The empirical model is conducted using county-level data to estimate the impact of the SISGC on the Yearbooks of Inner Mongolia. It should be noted that we used the city-level climate data to represent condition of grasslands in the pastoral areas of Inner Mongolia. The dataset includes 54 counties the temperature and precipitation of each county that affiliates to its corresponding city. consisting of the 33 pastoral counties and 21 semi-pastoral counties, and spans 14 years from 2001 to

2014 including the period before and after the introduction of the SISGC. The data collected include 5.2.3 Statistical analysis on grassland condition indicators of the grassland condition, the status of SISGC, and climate and socio-economic factors. In Based on the data of NDVI over the 54 counties of Inner Mongolia from 2001 to 2014, table 5.1 outlines the existing literature, the grassland condition has been indicated by vegetation coverage, height, density, the changes in grassland condition at different time intervals. biomass production and density of perennial vegetation (Gu and Li, 2013; Yu and Farrell, 2013). These indicators are commonly measured using two methods: direct sampling tests undertaken during small- Table 5.1 Grassland condition (NDVI) at different time intervals of SISGC scale fieldwork; and estimation over large areas with remote sensing technology (Gu and Li, 2013). The Std. Welch's t- Periods Status of SISGC Obs. Mean Min Max former focuses on measuring the micro-indices of grass, while the latter estimates the macro grassland Dev. test Period 1 condition on the basis of satellite images. For this study, the remote sensing technology appeared more Before SISGC 540 0.467 0.199 0.074 0.853 na (2001-2010) feasible and appropriate considering that we aim to quantify the grassland condition of the 54 counties Period 2 After SISGC 216 0.502 0.201 0.083 0.851 -0.035** in Inner Mongolia over 14 years. The Normalised Difference Vegetation Index (NDVI) is a widely- (2011-2014) The first year of 2011 54 0.482 0.208 0.084 0.851 -0.015 used indicator to quantify the grassland condition on the basis of the remote sensing technology (e.g. SISGC Period 2 The second year of Yang et al., 1998; Liu et al., 2013; Gong et al., 2015). Theoretically, the measurement of NDVI is based 2012 54 0.528 0.186 0.085 0.850 -0.061** SISGC on the differential absorption, transmittance and reflectance of energy by the vegetation in the red and The third year of 2013 54 0.520 0.203 0.084 0.840 -0.053* near-infrared portions of the electromagnetic spectrum (Senay and Elliott, 2000). NDVI is calculated SISGC The fourth year of 2014 54 0.480 0.208 0.083 0.850 -0.013 on a per-pixel basis as the ratio of the difference of the near infrared and red bands over their sum from SISGC a remotely-sensed image. It provides a clear description of land surface features and is regarded as a Sources: Statistics are derived from the data set of NASA’s Earth Science Data Systems Program Note: *significant at 10%; **significant at 5%; ***significant at 1%. proxy for terrestrial vegetation condition because it is strongly associated with percentage of vegetation coverage, leaf area index, potential photosynthesis, aboveground net primary productivity and biomass According to the Welch's t-test on NDVI between periods 1 and 2 (Lee, 1992), the overall mean of availability (Tan and Li, 2015). Based on the data set of MOD13A1 from NASA’s Earth Science Data NDVI is significantly larger in the period after SISGC than before, which implies an improvement in Systems Program, which is the Moderate Resolution Imaging Spectroradiometer (MODIS) imagery overall grassland condition. Moreover, all of the mean NDVI values are larger in the years with SISGC with 500m spatial resolution and 16 days temporal resolution, we employed the Maximum Value than the overall mean of NDVI in the period before SISGC. However, the Welch's t-test on NDVI

91 92

availability (Tan and (Tan availability coverage, leafpotentialphotosynthesis area index, proxy for terrestrial vegeta proxy terrestrial for It descri clear image. a provides a remotely-sensed of basisastheratio the difference of on aper-pixel thenear-infrared electromagnetic portions of spec reflec transmittanceabsorption, on the and differential YangLiu et al., 1998; et al., 2013; Gong et al., 2015). grassland condition quantifythe used indicatorto Inneryears.in 14 Mongoliaover TheNormalised thatweconsidering aimto appropriate feasible and condition onthebasisof satellite images.For thisstudy, technology theremote more sensing appeared of themicro-indices former onmeasuring focuses areas with estimation overlarge and fieldwork; scale measuredcommonly indicators are twomethods:directsampling undertaken during using small- tests vegetation density ofperennial biomass productionand been conditionhas grassland the literature, the existing grassland condition,thestatusof indicators ofthe int the after beforeand period 2014 including the consisting ofthe33pastoralcount in thepa condition ofgrasslands usingcounty-leve isconducted The empiricalmodel 5.2.2Data collection the deterministic stocking rate). theoblig well as as given subsidies(orincentives) Almost allrural householdsinthepastoralareasha grazgrasslands to the subject permanent which are countyIn every practice, area. livestock balance re grazing banandthe with thepermanent area all of the available natural grasslandInner areas of grass that 67.3 natural ofmillion available hectares areas Construction Programs pastoral intheextensive

with 500mspatialresolution Value and16 days resolution, temporal employed Maximum we the Systems is the Moderate Program, Resolu which Li, 2015). Based on the data set of 2015). BasedLi, onthedata tion condition because itisstronglytion condition a storal areas of Inner Mongolia. The dataset includes 54 counties Inner includes54counties Mongolia.Thedataset of storal areas ies and 21 semi-pastoral counties, ies and21semi-pastoral maining 37.3millionhectares totheforage- belong roduction of the SISGC. The data collected include include collected The data SISGC.roduction ofthe , aboveground netprimaryproductivity andbiomass tion Imagingtion (MODIS) imagery Spectroradiometer 91 91 trum (Senay and Elliott, 2000). NDVI is calculated calculated Elliott,2000).NDVI is and trum (Senay grass, while the latter estimates the macro grassland grassland estimates themacrograss, whilethelatter ation to comply with the permanent grazing ban (or (or grazing ban permanent with comply the ation to the near infrared and red bands over their sum from theirred bands sum from over and infrared the near Mongolia. 30millionhectares tothe ofthembelong ing ban,deterministicstocking orbothofthem. rate lands are covered by SISGC, which includes almost which includes almost lands are covered by SISGC, SISGC, and climate and socio-economic factors. In In factors. socio-economic SISGC, andclimate on the basis of the remote sensing technology (e.g. (e.g. technology sensing remote ofthe on thebasis quantify the grassland condition of the 54 counties quantifycondition of the 54 counties grassland the ption of land surface features and is regarded as a a regarded as and is features surface ption of land Difference Vegetation (NDVI) Index is a widely- remote sensing technology (Gu and Li, 2013). The 2013).The Li, and (Gu remote technology sensing in the pastoral areas contains different areas of of areas contains different areas in thepastoral l data to estimate the impact of the SISGC on the impactoftheSISGC onthe l datatoestimatethe ve been covered by SISGC, and they have been andthey been have SISGC, by covered ve been Theoretically, of NDVI the is based measurement indicated by vegetation coverage, height, density, density, vegetation height, coverage, by indicated tance of energy by the vegetation in the red and vegetation in and the red energy by the tance of of China. It is reported by the local governmentIt by thelocal China. isreported of (Gu andLi, 2013; Yu and Farrell, 2013). These MOD13A1 from NASA’s EarthScienc NASA’s MOD13A1 from ssociated with percentage ofvegetation percentage ssociated with and spans 14 years andspans14 from 2001to e Data e Data

the changes in grassland condition at different time intervals. intervals.condition atdifferent grasslandthe changes time in Based onthedata 5.2.3 Statistical analysis ongrassland condition ofeachcounty andprecipitation the temperature represent to climatedata city-level the that weIt used noted shouldbe Inner Mongolia. of Yearbooks onclimate InnerProducts data of Mongolia.The benefit Products of Data Chinese Agricultural and defl CompilationofCost- Annual onthe based collected was livesheep of price instance, theproducer onexistingwas statistical datagather collectedbased indicators socio-economic impacts. on The data possible and their programs these of our understanding Cons ofEcological informants ontheimplementation we Mongolia. interviews Moreover, Information aboutSISGC was providedbyAgricu the county. NDVI is significantly larger in the period after after period larger significantlyNDVI in the is between on NDVI t-test Welch's According to the at 1%. ***significant at 5%; **significantNote: *significant at10%; Sources: are se from Statistics derived thedata the image di ofadministrative layer Ultimately, 1986). (Holben, year during 2001-2014 each get of NDVI Composite to the (MVC) method than theoverall NDVI mean of intheperiodbe overall grassland condition.Moreover, allof the mean ND

(2011-2014) (2011-2014) Period 2 (2001-2010) Period 1 Periods Period 2 2 Period 2014 2013 2012 2011

Table 5.1 Grassland condition (NDVI)Table condition 5.1 Grassland at of NDVI over the SISGC year ofThe fourth SISGC yearThe third of SISGC year of The second SISGC year of The first After SISGC Before SISGC Status of SISGC

54 countiesInner of Mongolia to2014,table from 5.1outlines 2001 visions of Inner Mongoliaused to determine is for the NDVI each re conducted withlocalre herders conducted 54 54 54 54 216 540 Obs. t of NASA’s Earth Science Data NASA’s Program Systems t of Earth Science SISGC than before, whichimplies an improvement in that affiliates to its corresponding city. city. corresponding affiliatestoits that fore SISGC. However, the Welch's t-test onNDVI Welch'sfore SISGC. t-test the However, 92 92 0.480 0.520 0.528 0.482 0.502 0.467 Mean periods 1 and 2 (Lee, 1992), the overall mean of 1992), theoverall mean and 2(Lee, periods 1 ated with the Producer Price Index of Agricultural Index ofAgricultural Price ated withtheProducer tructionInner Programs Mongoliatoimprove in ed by andlocal national Statistical bureaus. For different time intervals of SISGC timeintervals different ofSISGC indicators gatheredwere from the Statistical lture and Animal Husbandry Inner lture and Animal Bureauof VI values are larger in the yearsVI valuesare inthe larger withSISGC 0.208 0.203 0.186 0.208 0.201 0.199 Dev. Std. and public officers who are key and publicofficers key whoare 0.083 0.084 0.085 0.084 0.083 0.074 Min 0.850 0.840 0.850 0.851 0.851 0.853 Max -0.013 -0.053* -0.061** -0.015 -0.035** na test Welch's t-

5 between period 1 and each year after SISGC indicates that the difference is only significant in the years statistical data on the meat output of the pastoral areas of Inner Mongolia, no distinction is made between 2012 and 2013. outputs from extensive grazing, and outputs from industrial raising and intensive animal husbandry. Since sheep is the animal which is most often raised by local herders (Zhang et al., 2012), the real 5.2.4 Fixed effects model producer price of live sheep with a two-year lag ( ) is used to indicate the situation of livestock A fixed effects model has been widely used in economic research, primarily to study the causes of production in the empirical model. In this regard, theܵ ௜௧ିଶprice is exogenously decided by the global market changes within entities over time (e.g. Fergusson et al., 2002; Huang et al., 2006). The model employs of China, and the two-year lag is considered as the timespan during which local herders adjust their within transformation to remove all time-invariant (fixed) explanatory variables, i.e. the model is sheep population to respond to the changes in sheep price. According to information from our household performed in deviations from individual means (Verbeek, 2012). As such, the fixed effects model survey in Inner Mongolia, the timespan is around two years because it covers the breeding period of provides a method that takes observable explanatory variables as well as unobservable time-invariant breeders; herders then reserve more new lambs in order to expand their sheep population during the variables into account, but the estimation does not depend on the value of time-invariant (fixed) next production stage. The climate factor is elaborated by the monthly average temperature ( ) and variables (Verbeek, 2012). Such an approach is appropriate for this study because the causes of precipitation ( ) during the growing season of grass, including the six months from ܶApril௜௧୫ to grassland condition changes within each county over time can be studied by controlling for the measured September withܴ m=4௜௧௠ to 9 (Zhao and Guo, 2009), and the variance of annual average temperature ( ) as well as unmeasured time-invariant heterogeneity between counties. In addition, our research samples and precipitation ( ). The annual variance is presented by the standard deviations of monthly averageܸܶ௜௧ include all of the counties in the pastoral areas of Inner Mongolia, rather than random draws, which temperature and precipitationܴܸ௜௧ in all twelve months of each year. In addition, time trend ( ) is included preliminarily indicates that the fixed effects model is more appropriate than the random effects model considering the impact of factors that change with time, such as the development of intensiveܻ௧ animal (Verbeek, 2012). Nevertheless, the appropriateness of the fixed effects model will be further examined husbandry and the increasing area of artificial grass for grassland restoration. The interaction term of through statistical testing. Based on the theoretical framework of the fixed effects model, we formulate time trend and the pastoral or semi-pastoral attribute of county i is also considered in order to the impacts of SISGC and other potential factors on grassland condition in the following equation: time௜ሻ change between pastoral and ܦכ explore whether there are differences in the grassland condition withሺܻ௧ semi-pastoral counties. Finally, the factors, such as elevation, slope, soil type, distance to the provincial

ଶ଴ଵସ ଶ଴ଵସ ଽ ଽ capital etc., that reflect the heterogeneity of grasslands and do not change over time are treated as time-

௜௧ ௜ ௬ ௜௬ ௬ ௜௬ ௜ ଵ ௜௧ିଶ ௠ ௜௧௠ ଵ ௜௧ ௠ ௜௧௠ invariant (fixed) factors in the model, and are denoted by in equation (1). , , , , , , , ܴ ሻ൅ܿ ܵ ൅෍݀ ܶ ൅݀ ܸܶ ൅෍݁ܦכ ൌ݄ ൅෍ܽܲ ൅෍ܾሺܲ ܩ ௬ୀଶ଴ଵଵ ௬ୀଶ଴ଵଵ ௠ୀସ ௠ୀସ and are the coefficients of the independent variables ݄and௜ is the randomܽ௬ errorܾ௬ ܿ term.ଵ ݀௠ The݀ଵ specific݁௠ ݁ଵ ௜ሻ൅ᖡ௜௧ሺͳሻܦכ ൅݁ଵܴܸ௜௧ ൅݂ଵܻ௧ ൅݃ଵሺܻ௧ where i and t indicate the ith county and year t, from 2001 to 2014. Grassland condition ( ), presented definition݂ଵ ݃ଵ of these variables is stated in table 5.2. ᖡ࢏࢚ by NDVI, is employed as the dependent variable of equation (1). is a dummy variable௜௧ to manifest ܩ whether county i implemented SISGC in the specific year y whichܲ ௜௬involves the first year with SISGC (y=2011), the second year (y=2012), the third year (y=2013) and the fourth year (y=2014). The interaction term of SISGC and the pastoral or semi-pastoral attribute of county i is considered. ሺܲ௜௬ሻ ሺܦ௜ሻ

Besides the policy variables, other potential factors are controlled. For example, the livestock production and climate factor are prevalently discussed as the drivers that impact the grassland condition in the arid and semi-arid areas (e.g. Li et al., 2012; He et al., 2014). In this study, the real producer price of live sheep ( ) is used as the proxy variable of livestock production in the pastoral areas because of the limitationsܵ of௜௧ the data. More specifically, the livestock population in the pastoral areas was under- reported by herders due to the stocking rate regulation (Brown et al., 2008). Furthermore, in the

93 94

reported by herders due to the stocking rate regulation (Brown et al., 2008). Furthermore, in the 2008). Furthermore, inthe to thestockinget al., reported due (Brown regulation rate by herders the Morespecifically, the limitationsofdata. interaction term of SISGC SISGC of interaction term y thethird year(y=2012), (y=2011), second the whether county i implemented SISGC in the specific SISGCcounty inthespecific whether iimplemented the impacts of SISGC and other potential factors on grassland condition in the following equation: on grasslandthe impacts conditioninthefollowing ofSISGC other potentialfactors and equation: fr through Basedonthetheoretical statisticaltesting. Nevertheless, ofth theappropriateness 2012). (Verbeek, preliminarily effects mode thefixed that indicates inthe include allofthecounties as well heterogeneity asunmeasured time-invariant ti over county withineach grassland conditionchanges of causes the for because thisstudy appropriate is an approach Such 2012). (Verbeek, variables but theestimation doesnot variables intoaccount, unobservable as well time-invariant explanatory asprovides observable a method that variables takes from effects thefixed performed individualmeans model 2012).Assuch, indeviations (Verbeek, is themodel i.e. variables, explanatory (fixed) all time-invariant remove transformation to within changes within entities over time (e.g. Fergusson ec in used widely been effectsmodelhas A fixed effectsmodel 5.2.4 Fixed 2012 and2013. year after indicates SISGC and each between 1 period where

of livesheep( in thearidandsemi-arid areasLi (e.g. etal.,2012; prevalently production discussed are and climate factor factors other policyBesides potential variables, the considered. by isemployed NDVI, asthe ܩ ௜௧ i and ൌ݄ t ௜ indicatethe ൅෍ܽ ܵ ௜௧ ௬ୀଶ଴ଵଵ ) is used as the proxy of livestoc variable ଶ଴ଵସ ൅݁ ଵ ܴܸ ௬ i ܲ th ௜௬ ௜௧ county and year county t, from and ൅݂ ሺܲ ൅෍ܾ dependent variab ௜௬ ௬ୀଶ଴ଵଵ ଵ ଶ଴ଵସ ሻ ܻ pastoral Inner areas of Mongolia, ௧ countyi of attribute semi-pastoral or pastoral and the ൅݃ ଵ ௬ ሺܻ ሺܲ ௧ ௜௬ ܦכ ܦכ le ofequation(1). ௜ ሻ൅ᖡ ௜ livestock population in the pastoral areas was under- was areas pastoral livestock populationinthe et al.,2002;Huang et al.,2006).Themodelemploys l is more than appropriate 93 93 ሻ൅ܿ He etal.,2014).In thisstudy, the real price producer ear (y=2013) and the fourth year the fourth (y=2013) The (y=2014). and ear onomic research, primarily to study the causes of of causes study the to primarily onomic research, between counties. In ourresearch counties. between addition, samples 2001 to 2014. Grassland condition( 2001 to2014.Grassland amework of the fixed effects model,weformulate amework ofthefixed depend on the value of time-invariant (fixed) (fixed) oftime-invariant dependonthevalue me controllingcan be studied by for the measured year y which involves the first yearyeary with SISGC the first which involves that the difference is only significant in the years only significant in is the that the difference ௜௧ as the drivers that impact the grassland condition as thedriversthatimpact ଵ are controlled. For example, the livestock the livestock For example, controlled. are ሺͳሻ e fixed effects model will be further examined examined modelwillbefurther effects e fixed ܵ ௜௧ିଶ k productioninthe of pastoral areasbecause ൅෍݀ ௠ୀସ ଽ ܲ ௜௬ rather than random draws, which than randomwhich draws, rather is adummy tomanifest variable ௠

ܶ ௜௧௠ the random effects model effects the random model ൅݀ ଵ ܸܶ ௜௧ ൅෍݁ ܩ ௠ୀସ ௜௧ ଽ ), presented ௠ ሺܦ ܴ ௜ ௜௧௠ ሻ is is temperature and precipitation in all twelve months of each year.Inaddition, timetrend( inalltwelve andprecipitation monthsof each temperature September with m=4 to 9 (Zhao and Guo, 2009), a 2009), and Guo, (Zhao to9 September withm=4 invariant (fixed) factors in the model, and are denoted by denoted are and the model, factorsinvariant in (fixed) grassla heterogeneity of capital etc.,thatreflectthe counties.Finally,semi-pastoral thefactors, suchas explore whether there are differences in the grassl whether in the there are differences explore is stated in tabledefinition 5.2. of these variables statistical dataonthemeat outputof producer price of live sheep with a two-year lag ( lag atwo-year with livesheep priceof producer whichismostoftenrais Since sheepistheanimal outputs from extensivegrazing, andoutputsfrom i next production stage.next climate factor The is elab the populationduring sheep their order toexpand lambsin new more reserve breeders; herders then surveyInner Mongolia,thetimespanisaroundtw in sheep population to respond to the changes in shee of China, and the two-year lag is considered as production intheempiricalIn model. thisregard, the

time trend and the pastoral or semi-pastoral attribute of i county attribute orsemi-pastoral time trendandthepastoral of Theinteraction term grassrestoration. grassland for artificial of area theincreasing husbandry and considering the impact of factors that change with precipitation ( ݂ and precipitation( ଵ and݃ ଵ are the coefficients of the independent variables and variables of theindependent coefficients theare ܴ ௜௧௠ ܴܸ ) during thegrowing ofgrass, season includingmonths from thesix Aprilto ௜௧ ). The annual variance is presented by th ispresented variance ). Theannual the pastoral areasInner of Mongol orated by the monthly average temperature ( temperature average byorated themonthly the timespanduring local which herders adjusttheir p price. According to information from our household and condition with timeand cha nd the variance of annual average temperature ( temperature average annual of nd thevariance 94 94 nds andas do not time change time- are over treated elevation, slope,soil type, distancetotheprovincial ܵ time, such as the development ofintensiveanimaltime, such ed by local herders (Zhang et al., 2012), the real real et al.,2012),the (Zhang herders ed by local ndustrial raising andintensiveanimalhusbandry. ௜௧ିଶ price by is exogenouslyglobal decided the market o yearso because it covers the breeding of period ) is used to indicate the situation of livestock situationoflivestock the toindicate ) isused ݄ ௜ in equation (1).in e standard deviations of monthly average of monthlyaverage e standard deviations ᖡ ࢏࢚ ሺܻ is the random error term. The specific specific The error term. randomthe is ௧ ܦכ ia, no distinction is madeia, no between ௜ ሻ isalsoconsidered inorderto ܽ nge between pastoral and pastoraland nge between ௬ , ܾ ௬ , ܿ ଵ , ݀ ܻ ௠ ௧ ) is included ) isincluded , ݀ ܶ ଵ , ݁ ௜௧୫ ௠ ) and ) and ܸܶ , ݁ ௜௧ ଵ ) ,

5 Table 5.2 Variable definitions and their expected signs in the model Furthermore, the interaction term of time variable and the attribute of county i is expected to

have a negative coefficient because the pastoral counties experienced less intensive௧ animal௜ husbandry ሻܦכ Variables Variable definition Unit Expected signs ሺܻ than the semi-pastoral counties (Waldron et al., 2010). NDVI of county i in year t na Dependent variable ௜௧ ܩ =1 if county i implemented SISGC in year y na + In addition, we expect that these explanatory variables may have different effects in different types of (y=2011, 2012, 2013 and 2014), =0 otherwise ௜௬ counties, especially for counties that have a different initial grassland condition. It is suggested that the ܲ =1 if county i is a pastoral county, =0 na na† otherwise effectiveness of large-scale ecological conservation programs has been heterogeneous in different ܦ௜ The ratio of real producer price of live sheep na - regions even under the same policy context with incentive measures (Lv et al., 2015). The 54 counties in county i between year t and year 2001 ௜௧ ܵ The average temperature of county i in month Degrees Celsius +/- in the sample are therefore divided over two groups based on the NDVI of each county in 2001. This m of year t (m=4, 5, 6, 7, 8 and 9) ௜௧௠ results in 27 counties in the low NDVI group and 27 counties in the high NDVI group. NDVI of the ܶ The temperature variation of county i in year t Degrees Celsius +/- low NDVI group in 2001 ranges from 0.074 to 0.427, and the mean is 0.254. The range of NDVI in the ܸܶ௜௧ The total precipitation of county i in month m Millimeters + of year t (m=4, 5, 6, 7, 8 and 9) high NDVI group in 2001 is between 0.453 and 0.836, and the mean is 0.614. The fixed effects model ܴ௜௧௠ The precipitation variation of county i in year Millimeters +/- (equation (1)) will be conducted separately for three groups of counties: all counties; low NDVI counties; t ௜௧ ܴܸ Year t na + high NDVI counties.

Note: †௧ indicates that is included in the interaction terms of ( ) and ( ). The text presents ܻ 5.2.5 Descriptive statistics for the variables the expected signs of ( ) and ( ) ௜ܦכ ௜ ܻ௧ܦכ ௜ ܲ௜௬ܦ propose௜ the expected signs of the explanatory variables on Table 5.3 shows the characteristics of the variables included in the fixed effects model. The overall ܦכ literature,௜ ܻ we௧ ܦכ According to the existingܲ௜௬ grassland condition in table 5.2. Given that the government reports and most of academic findings mean of NDVI ( ) is 0.477 for all 54 counties from 2001 to 2014, 0.314 for the 27 low NDVI counties, present that SISGC has improved the grassland condition of Inner Mongolia (Li et al., 2014; Ministry and 0.641 for theܩ ௜௧27 high NDVI counties. Because all counties implemented SISGC in 2011, therefore of Agriculture of China, 2016), we expect positive coefficients for in each year when SISGC was the mean value of over 14 years is 0.071. The mean value of (0.044) indicates that the percentage of samples௜ଶ଴ଵଵ from the pastoral counties in 2011 (33) is 4.4%௜ of௜ଶ଴ଵଵ the total observations (756). ܲכ ܦ ܲ implemented. The expected signs of the interaction terms ( )ܲ in௜௬ our estimations are ambiguous ௜௬ in the pastoral counties than in theܲכ௜better ܦbecause it is uncertain whether the effectiveness of SISGC is semi-pastoral counties or vice versa. Additionally, the real producer price of live sheep ( ) is expected to have negative effects on the grassland condition because higher prices stimulate ܵmore௜௧ livestock production that may result in excessive grazing pressure and further lead to grassland degradation (Zhang et al., 2014). The expected signs of monthly average temperature ( ) in the growing season are ambiguous because the high temperature facilitates plant growth as wellܶ௜௧௠ as the evaporation of the ground surfaces and the evapotranspiration from the vegetation (Deng et al., 2013), and fast evaporation and evapotranspiration have negative effects on plant growth in the arid and semi-arid areas. Given that precipitation is considered as the crucial factor in improving grassland condition in the arid and semi- arid areas (Li and Zhang, 2009), we expect positive coefficients for in the model. The expected signs of the annual variation in temperature ( ) and precipitation ( ܴ௜௧௠) are ambiguous as the uneven distribution of climate has uncertain effectsܸܶ on௜௧ grass growth. Timeܴܸ variable௜௧ ( ) is expected to have positive effects on grassland condition. For instance, the development of intensiveܻ௧ animal husbandry is supposed to replace extensive grazing and further reduce the use and damage of natural grasslands.

95 96

distribution of climate has uncertaindistribution of climate has arid areas (Li and Zhang, 2009), we expect positive for 2009),weZhang, coefficients expect arid areasand (Li im in asthecrucial factor precipitation isconsidered have and evapotranspiration ground andtheevapotranspiration from th surfaces facilitates thehigh temperature because are ambiguous signs( the expected of implemented. The expected signs of the interaction terms theinteraction ( of signs expected implemented. The signs of the annual variation in temperature ( semi-pastoral counties or vice or versa. Additiona counties semi-pastoral of effectiveness the whether isuncertain it because supposed to replaceand grazing further extensive instance, condition. For grassland on positive effects Note: † Note: † indicates that of Agriculture of China, 2016), we expect positive coefficients for positivecoefficients 2016), we expect of AgricultureChina, the present thatSISGC hasimproved 5.2.Given thatthe intable grassland condition According to the existing we propose literature,

(Zhang et al., 2014). The expected si expected (Zhang etal.,2014).The production thatmay inexcessiveresult grazing pre to have negative effects on the effects to have negative Variables Variables ܴ ܶ ܴܸ ܸܶ ܲ ܩ ܵ ܦ ܻ ௜௧௠ ௜௧௠ ௜௬ ௜௧ ௜௧ ௧ ௜ ௜௧ ௜௧

Year t Year t t yearThe precipitationvariation iin ofcounty and 9) year 5,6,7,8 of t(m=4, countyThe totalprecipitationof iinmonthm year iin variationofcounty t The temperature year 8and9) m of t(m=4,5,6,7, of county iinmonth The average temperature year yeart andin county 2001 i between priceThe ratio oflivesheep ofrealproducer otherwise =1 if county i county, is a pastoral =0 (y=2011, 2014),=0otherwise 2012,2013and yeary iimplementedSISGC=1 ifcounty in year t countyNDVI of iin Variable definition Table definitions 5.2 Variable signs and their expected in the model ܦ ܲ ௜ ௜௬ is included intheinteraction terms of( ܦכ negativeeffects onplantgrowth inthe ௜ ) and ( grassland condition because higher becausehigher grassland condition ܻ gns of monthly average temperature ( temperature gns average ofmonthly effects on grass growth. Time variable ( growth. Timevariable grass effects on ௧ grassland condition of grasslandInner condition (Liet al., 2014; Ministry Mongolia ܦכ ௜ ) ܸܶ lly, the real producer price of live sheep ( live sheep of price the realproducer lly, ௜௧ ) andprecipitation ( government reports and most of academic findings findings ofacademic most and government reports e vegetation (Deng et al., 2013), and fast evaporation e vegetation etal.,2013),andfast (Deng evaporation 95 95 the expected signs of the explanatory variables on explanatory variables signs ofthe the expected reduce ofnaturalgrasslands. theuseanddamage SISGC is better in the pastoral counties than in the the thanin counties pastoral inthe isbetter SISGC proving grassland condition inthearidand semi- grassland condition proving the development of intensive animal husbandry is animalhusbandry ofintensive development the ssure and further lead further tograssland degradation ssure and plantgrowth aswelltheevaporation ofthe ܦ ௜ ܲכ ܲ na Millimeters Millimeters Degrees Celsius Degrees Celsius na na na na Unit ௜௬ ௜௬ ܦכ arid and semi-arid areas. Given that areas.Giventhat arid andsemi-arid ) in our estimations are ambiguous are ambiguous estimations ) inour ܲ ܴܸ ௜௬ ܴ ௜ prices stimulate morelivestock stimulate prices ) and ( ௜௧௠ ௜௧ was SISGC when year each in ) are ambiguous as the uneven asthe ) are ambiguous in themodel.Theexpected ܶ ௜௧௠ ܻ ௧ ) inthegrowing) season ܻ ܦכ ௧ ) is expected to have to have expected ) is + +/- + +/- +/- - na† + variable Dependent signs Expected ௜ ). The text presents). Thetext

ܵ ௜௧ ) is expected expected ) is the mean value ofthe mean value al Because counties. NDVI the 27 high and 0.641 for percentage of samples from the pastoral counties in percentage fromthepastoralcounties ofsamples Furthermore, the interaction term time of interaction Furthermore, the

mean of NDVI( ofthevariables Table 5.3showsthecharacteristics Descriptive statisticsforthevariables5.2.5 high counties. NDVI for three separately (equation conducted (1))willbe and 0.836, 0.453 isbetween highgroup in2001 NDVI NDVI inthe of group Therange ranges 0.074to0.427,andthemeanlow NDVI from is0.254. in2001 results in27counties group the lowNDVI and grou two divided over therefore in the sample are regions policy evenunderthesame with context in conservati ecological oflarge-scale effectiveness different a thathave counties for counties, especially of types indifferent effects may havedifferent variables thattheseexplanatory Inaddition, we expect (Waldronetal.,2010). counties than thesemi-pastoral c thepastoral because coefficient negative have a ܩ ௜௧ ܲ ) is0.477 for all54countiesfrom 2001 to ௜ଶ଴ଵଵ over 14 years value of mean overis 0.071. 14 The variable and the attribute of county i of attribute and the variable ounties experienced less intensive animal husbandry animal intensive less experienced ounties ps based on the NDVI of each county in 2001. This in2001.This county each of NDVI ps basedonthe 96 96 on programs has been heterogeneous in different indifferent heterogeneous on programs hasbeen centive measures (Lvcentive measures al.,2015).The54counties et 27 countiesinthehighgroup. NDVI NDVI of the 2011 (33) is 4.4% of the total observations (756). totalobservations(756). ofthe 2011(33)is4.4% groups of counties: all c groups counties:all of l counties implemented SISGC therefore l countiesimplemented in2011, initial grassland condition. It is suggested that the It suggestedthat is condition. initial grassland included in the fixed effects model. The overall and the mean is 0.614. The fixed effects model model effects fixed The is0.614. mean and the 2014, 0.314for the 27lowNDVI counties, ܦ ௜ ܲכ ௜ଶ଴ଵଵ ounties; low NDVI counties; counties; ounties; low NDVI (0.044) indicates that the that (0.044) indicates ሺܻ ௧ ܦכ ௜ ሻ is expected to expected is

5 Table 5.3 Descriptive statistics of the variables in the fixed effects model

All counties Low NDVI counties High NDVI counties Variables Mean Std. Dev. Obs. Mean Std. Dev. Obs. Mean Std. Dev. Obs. 0.477 0.200 756 0.314 0.125 378 0.641 0.104 378 0.071 na 756 0.071 na 378 0.071 na 378 ௜௧ ܩ 0.071 na 756 0.071 na 378 0.071 na 378 ௜ଶ଴ଵଵ ܲ 0.071 na 756 0.071 na 378 0.071 na 378 ௜ଶ଴ଵଶ ܲ 0.071 na 756 0.071 na 378 0.071 na 378 ௜ଶ଴ଵଷ ܲ 0.044 na 756 0.048 na 378 0.040 na 378 ௜ଶ଴ଵସ ܲ 0.044 na 756 0.048 na 378 0.040 na 378 ௜ ௜ଶ଴ଵଵ na 756 0.048 na 378 0.040 na 378 0.044 ܲכ ܦ ௜ ௜ଶ଴ଵଶ na 756 0.048 na 378 0.040 na 378 0.044 ܲכ ܦ ௜ ௜ଶ଴ଵଷ 324 0.599 1.530 324 0.599 1.530 648 0.598 1.530 ܲכ ܦ ௜ ௜ଶ଴ଵସ 378 2.535 8.602 378 3.850 7.333 756 3.318 7.967 ܲכ ܦ ௜௧ିଶ ܵ 15.529 2.362 756 15.066 2.706 378 15.992 1.849 378 ௜௧ସ ܶ 20.813 2.055 756 20.750 2.227 378 20.877 1.868 378 ௜௧ହ ܶ 22.833 1.667 756 22.720 1.853 378 22.946 1.451 378 ௜௧଺ ܶ 21.035 1.862 756 20.905 2.032 378 21.164 1.668 378 ௜௧଻ ܶ 15.000 2.067 756 14.593 2.404 378 15.407 1.564 378 ௜௧଼ ܶ 13.076 na 756 13.683 na 378 12.470 na 378 ௜௧ଽ ܶ 14.965 15.912 756 13.287 14.345 378 16.644 17.193 378 ௜௧ ܸܶ 29.526 22.371 756 26.623 20.128 378 32.429 24.088 378 ௜௧ସ ܴ 57.486 38.445 756 50.960 36.319 378 64.013 39.439 378 ௜௧ହ ܴ 81.173 51.808 756 72.722 49.640 378 89.624 52.611 378 ௜௧଺ ܴ 57.617 39.026 756 54.424 40.651 378 60.810 37.110 378 ௜௧଻ ܴ 34.058 26.697 756 29.750 21.998 378 38.366 30.103 378 ௜௧଼ ܴ 31.864 na 756 29.134 na 378 34.594 na 378 ௜௧ଽ ܴ 7.500 4.034 756 7.500 4.036 378 7.500 4.036 378 ௜௧ ܴܸ 4.583 na 756 5.000 na 378 4.167 na 378 ௧ Note: Theܻ mean of dummy variables refers to the share of observations for which the variable equals 1. ௧ ௜ .are derived from Statistical Yearbooks and our survey ܦכSources:ܻ statistics

are significant and positive. Specifically, the significant coefficient of the significant and positive. areSpecifically, significant Incounties. groupall counties,thecoefficie the of Table 5.4 presents results the estimated of the at 1%. at 5%;***significant **significantNote: *significant at10%; estimation. for our appropriate indicating effects modelismore thatthefixed atp<0.05, thenullhypothesis rejected test model.TheHausman effects random than the The Hausman testisemployed todeterminewhet 5.3.1 Modelresults 5.3 Results

Observations Constant Variable ܦ ܦ ܦ ܦ ௜ ௜ ௜ ௜ ܻ ܲ ܲ ܲ ܲ ܵ ܲכ ܲכ ܲכ ܲכ ௧ ܴܸ ܸܶ ܴ ܴ ܴ ܴ ܴ ܴ ܶ ܶ ܶ ܶ ܶ ܶ ௜ଶ଴ଵସ ௜ଶ଴ଵଷ ௜ଶ଴ଵଶ ௜ଶ଴ଵଵ ܴ ௜௧ିଶ ܻ ܦכ ௜௧଼ ௜௧଻ ௜௧଺ ௜௧ହ ௜௧ସ ௜௧ଽ ௜௧଼ ௜௧଻ ௜௧଺ ௜௧ହ ௜௧ସ ௜௧ଽ ௧ ଶ ௜ଶ଴ଵସ ௜ଶ଴ଵଷ ௜ଶ଴ଵଶ ௜ଶ଴ଵଵ ௜௧ ௜௧

௜

648 0.3913 1.085 -0.003* 0.005*** -0.0001 0.0001 0.00005 -0.0001 -0.00001 0.0003*** -0.0002 -0.001 0.0003 -0.006*** -0.005 -0.007*** -0.008*** -0.001 -0.110*** 0.010 0.012 0.018 0.016 0.150*** 0.189*** 0.130*** 0.042*** Coefficient All counties Table 5.4 Estimated results of the fixed effects model effects ofthefixed Table 5.4Estimatedresults model 10.02 -1.69 2.61 -0.30 1.23 0.57 -0.94 -0.09 3.07 -1.21 -0.59 0.09 -2.83 -1.52 -3.66 -3.17 -1.02 -5.11 0.62 0.78 1.24 1.20 4.91 7.33 7.36 2.83 t 324 0.4483 0.985*** -0.004 0.012*** -0.001* 0.0004*** 0.0002** -0.000005 0.0001 0.001*** -0.0001 -0.004** -0.001 -0.008*** -0.011** -0.003 -0.001 0.001 -0.161*** 0.026 0.014 0.016 0.035 0.163*** 0.210*** 0.153*** 0.025 Coefficient Low NDVI counties 98 her the fixed effects modelis thefixed moreappropriate her fixed effects model for the three groups for of thethree effects model fixed nts of the variables about SISGC in each yearnts of the variables about SISGC in each 6.36 -1.10 3.96 -1.74 2.79 2.11 -0.03 0.73 3.04 -0.68 -2.69 -0.27 -3.14 -2.64 -0.72 -0.46 0.34 -3.94 1.29 0.43 0.81 1.23 3.14 3.96 5.46 0.81 t 324 0.5156 1.039*** -0.003 -0.003 0.0002 -0.00003 0.0001 -0.0001 -0.0001 0.0001 -0.0001 0.009*** -0.0003 -0.001 0.003 -0.010*** -0.015*** -0.001 -0.065* 0.014 0.016 0.022 0.008 0.141*** 0.185*** 0.110*** 0.056** Coefficient High NDVI counties ୧ଶ଴ଵଵ indicates that that indicates 5.51 -0.90 -1.24 0.34 -0.26 0.72 -0.60 -0.84 0.96 -0.88 2.85 -0.05 -0.17 0.57 -3.78 -4.91 -0.71 -2.04 0.64 0.61 0.88 0.64 3.24 4.07 3.74 2.60 t Table 5.3 Descriptive statistics of the variables in the fixed effects model

indicating that the fixed effects model is more appropriate for our estimation. estimation. for our appropriate indicating effects modelismore thatthefixed atp<0.05, thenullhypothesis rejected test model.TheHausman effects random than the The Hausman testisemployed todeterminewhet 5.3.1 Modelresults 5.3 Results are significant and positive. Specifically, the significant coefficient of the significant and positive. areSpecifically, significant Incounties. groupall counties,thecoefficie the of Table 5.4 presents results the estimated of the at 1%. at 5%;***significant **significantNote: *significant at10%;

Variable Variable Observations Constant ܦ ܦ ܦ ܦ ௜ ௜ ௜ ௜ ܻ ܲ ܲ ܲ ܲ ܵ ܲכ ܲכ ܲכ ܲכ ௧ ܴܸ ܸܶ ܴ ܴ ܴ ܴ ܴ ܴ ܶ ܶ ܶ ܶ ܶ ܶ ௜ଶ଴ଵସ ௜ଶ଴ଵଷ ௜ଶ଴ଵଶ ௜ଶ଴ଵଵ ܴ ௜௧ିଶ ܻ ܦכ ௜௧ହ ௜௧ସ ௜௧଼ ௜௧଻ ௜௧଺ ௜௧ଽ ௜௧଼ ௜௧଻ ௜௧଺ ௜௧ହ ௜௧ସ ௜௧ଽ ௧ ଶ ௜ଶ଴ଵସ ௜ଶ଴ଵଷ ௜ଶ଴ଵଶ ௜ଶ଴ଵଵ ௜௧ ௜௧

௜

0.00005 -0.0001 -0.00001 0.0003*** -0.0002 -0.001 0.0003 -0.006*** -0.005 -0.007*** -0.008*** -0.001 -0.110*** 0.010 0.012 0.018 0.016 0.150*** 0.189*** 0.130*** 0.042*** Coefficient All counties 648 0.3913 1.085 -0.003* 0.005*** -0.0001 0.0001 Table 5.4 Estimated results of the fixed effects model effects ofthefixed Table 5.4Estimatedresults model 0.57 -0.94 -0.09 3.07 -1.21 -0.59 0.09 -2.83 -1.52 -3.66 -3.17 -1.02 -5.11 0.62 0.78 1.24 1.20 4.91 7.33 7.36 2.83 t 10.02 -1.69 2.61 -0.30 1.23 324 0.4483 0.985*** -0.004 0.012*** -0.001* 0.0004*** 0.0002** -0.000005 0.0001 0.001*** -0.0001 -0.004** -0.001 -0.008*** -0.011** -0.003 -0.001 0.001 -0.161*** 0.026 0.014 0.016 0.035 0.163*** 0.210*** 0.153*** 0.025 Coefficient Low NDVI counties 98 her the fixed effects modelis thefixed moreappropriate her fixed effects model for the three groups for of thethree effects model fixed nts of the variables about SISGC in each yearnts of the variables about SISGC in each 5.46 0.81 t 6.36 -1.10 3.96 -1.74 2.79 2.11 -0.03 0.73 3.04 -0.68 -2.69 -0.27 -3.14 -2.64 -0.72 -0.46 0.34 -3.94 1.29 0.43 0.81 1.23 3.14 3.96 0.110*** 0.056** Coefficient High NDVI counties 324 0.5156 1.039*** -0.003 -0.003 0.0002 -0.00003 0.0001 -0.0001 -0.0001 0.0001 -0.0001 0.009*** -0.0003 -0.001 0.003 -0.010*** -0.015*** -0.001 -0.065* 0.014 0.016 0.022 0.008 0.141*** 0.185*** ୧ଶ଴ଵଵ indicates that that indicates 3.74 2.60 t 5.51 -0.90 -1.24 0.34 -0.26 0.72 -0.60 -0.84 0.96 -0.88 2.85 -0.05 -0.17 0.57 -3.78 -4.91 -0.71 -2.04 0.64 0.61 0.88 0.64 3.24 4.07

5 NDVI increases by 0.042 in the first year of SISGC when the effects of other potential factors grazing are conducted in the areas with more severe degradation or inferior grassland condition. on NDVI are controlled. Similarly, NDVI increases by 0.13 until the second year of SISGC, Interestingly, the significant and negative coefficient of the real producer price indicates that a 0.189 until the third year, and 0.15 until the fourth year. The effectiveness of SISGC is 1% increase causes a 0.161 decrease in NDVI in the low NDVI counties, but only a 0.065 accumulated year by year. Therefore, the effectiveness of SISGC on NDVI in the first year is decrease in NDVI in the high NDVI counties. Market forces seem to have a stronger impact in increased by 0.042, 0.088 for the second year and 0.059 for the third year. It is remarkable that the counties with worse initial grassland condition. Similarly, climate changes seem to have the effectiveness of SISGC in the fourth year is -0.039, although the accumulated effectiveness more impact in the counties with worse initial grassland condition. of SISGC on NDVI is a 0.15 increase by then. The interaction terms of SISGC in each year and 5.3.2 Comparison between the estimated results and actual grassland condition pastoral attribute of the county are not significant, which indicates that there are no significant differences in the impacts of SISGC on NDVI between pastoral counties and semi-pastoral Figure 5.2 depicts the estimated NDVI based on the significant coefficients of the variables of counties. With respect to the effects of other potential factors, the variable of the real producer SISGC in the model including all counties. The mean value of actual NDVI in the period 2001- price of live sheep presents significant and negative impacts on NDVI. It indicates that a 1% 2010 is used as the initial value. The actual NDVI is drawn from the mean value of NDVI of increase in the real producer price in year t-2 relative to the price in 2001 causes a decrease of the 54 counties in the periods 2001-2010, 2011, 2012, 2013 and 2014. 0.11 in NDVI when the factors of SISGC, climate, time trend and time-invariant heterogeneity NDVI between counties remain constant. Moreover, the temperature in May, June and August has 0.70 significant and negative effects on NDVI. This is explained through the increasing temperature 0.65 causing faster evaporation from the ground surfaces and evapotranspiration from the vegetation, 0.60 and allowing more pests and mice to breed in the arid and semi-arid areas, which are Estimated NDVI 0.55 significantly detrimental to grassland condition in the growing season (Li, 2009; Li et al., 2013). Actual NDVI 0.50 Precipitation in May, on the other hand, has significant and positive effects. Abundant precipitation is beneficial to the restoration of grassland condition (Li and Zhang, 2009). The 0.45 grassland condition is improved slowly with the factors that change with time trend, and the 0.40 2001-2010 2011 2012 2013 2014 effects of time variance on grassland condition are stronger in the semi-pastoral counties than in the pastoral counties. Figure 5.2 Comparison between the estimated NDVI and actual NDVI

Comparing the model results between the groups of low NDVI counties and high NDVI Sources: The data of actual NDVI is derived from the data set of NASA’s Earth Science Data Systems Program counties, the coefficient of the variable of SISGC in the first year is not significant in the group The estimated NDVI and actual NDVI in each year of SISGC are larger than the initial value. of low NDVI counties. This may be explained by the lagged implementation of SISGC in the Comparing the estimated NDVI and actual NDVI in figure 5.2, it shows that the actual NDVI low NDVI counties considering that these counties are not first regions for initiating ecological is clearly lower than the estimated NDVI during the years of SISGC. The differences between construction programs and are usually located in the relatively remote areas. However, the the actual and estimated grassland condition are attributed to the other potential factors that coefficients of the variables of SISGC in the second, third and fourth years are larger in the offset the effectiveness of SISGC, including the producer price and climate factors which group of low NDVI counties than in the group of high NDVI counties. This indicates that showed significant coefficients in the model results. For instance, data showed that the producer SISGC has stronger impacts on the grassland conservation in the low NDVI counties than in price continuously increased in the period 2011 to 2014 from 1.48 times to 2.54 times the the high NDVI counties. This result is consistent with the implementation planning of SISGC, producer price in 2001. Although the large increase in the producer price cannot impact upon namely, grazing bans that provide higher subsidies and implement more strict constraints for

99 100

0.189 untilthe year,third and0.15untilthef on NDVI Similarly, are controlled. NDVI in factors year potential other SISGC effects whenof of the inthefirst 0.042 by NDVI increases namely, grazing bans that providehighernamely,grazing subsid that bans iscons the highcounties. Thisresult NDVI SISGCervation hasstrongerimpactsonthegrassland inthelowNDVIcons counties thanin thaninthegroup group lowNDVI counties of inthes ofSISGC variables ofthe coefficients construction programs are and usually in located these countie low NDVI that counties considering by may explained the counties. This be of lowNDVI lagged of implementation SISGC in the in SISGC of of thevariable counties, thecoefficient Comparing the model results between the g in thepastoralcounties. on grassland conditionar oftimevariance effects grassland conditionisimprovedwith the slowly totherestorationprecipitation isbeneficial Precipitation inMay, ontheother hand,has significantly detrimentaltograssland conditionin and allowing and more micein pests to breed from thegroundsurf evaporation faster causing This NDVI. on effects and negative significant consta counties remain between and heterogeneity time-invariant time trend te, clima SISGC, factors of 0.11 the when in NDVI re year t-2 in price realproducer inthe increase andne significant presents livesheep price of producer real the of the variable factors, otherthe effectspotential tocounties. of respect With ofSISGC intheimpacts differences onNDVI notsignifican pastoral attributeofthecounty are Th by then. increase a 0.15 of SISGC NDVI is on year in the fourth of SISGC the effectiveness byyear 0.059 0.042,0.088forthesecond and increased year is first the NDVI in SISGC on of effectiveness the year. Therefore, year by accumulated

nt. Moreover, the temperature in May, June and August has and temperature August in May,nt. has Moreover, the June istent with the implementation planning of SISGC, SISGC, of implementation planningthe istent with of grasslandand Zhang, condition (Li 2009). The is -0.039, although the accumulated effectiveness effectiveness the although accumulated -0.039, is creases by 0.13 until the second year thesecond by ofSISGC,creases 0.13until 99 roups of low NDVI counties and high NDVI oflowNDVI andhighroups NDVI counties gative impacts on NDVI. It a 1% NDVI. thatgative indicates impacts on is explained through the increasing temperature temperature through theincreasing isexplained aces andevapotranspirationaces from thevegetation, econd, third and fourth years are larger in the fourth third and econd, lative to the price in 2001 causes a decrease of of adecrease causes in2001 price lative tothe t, which indicates that there are no significant whichindicatesthatthereare nosignificant t, ies andimplement morestrictconstraints fories e interaction terms of SISGC in each yearand ineach ofSISGC interaction terms e between pastoral counties and semi-pastoral and semi-pastoral counties pastoral between the growing (Li, Li season 2009; etal.,2013). ourth year. ofSISGCourth Theeffectiveness is s are notfirstregionss for ecological initiating e stronger in the semi-pastoral countiesthan e strongerinthesemi-pastoral factors that change with time trend, and the with factorschange that of high NDVI counties.This indicates that significant andpositiveeffects. Abundant the arid and semi-arid areas,which are thearidandsemi-arid therelativelyareas. remote the However, the first yeargroup isnotsignificant inthe thefirst for the third year.for thethird It that isremarkable producer price in 2001. Although the large increa continuously periodprice in the increased inthemodelre showed significant coefficients of SISGC, including effectiveness offset the the grassland actual and estimated condition are estimatedduring NDVI is clearly than the lower NDVI theactual in figure NDVIand actual 5.2,itshowsthat NDVI Comparing estimated the The estimated actual NDVI NDVI and grazing are conducted intheareasgrazing are withmorese conducted the 54 counties in the periods 2001-2010, 2011,2012,2013and2014. 2001-2010, the 54counties intheperiods ND 2010 isusedastheinitialvalue.Theactual SISGC inthemodel including allcounties.Th of theFigure of variables theestimated significant coefficients NDVI the 5.2 on depicts based 5.3.2 Comparisonbetween theestimated results and actual grassland condition condition. withworseinitialgrassland more impactinthecounties initial the countieswithworse decrease in NDVI in the high NDVI counties. Ma 0.065 butcounties, a only NDVI the low NDVI in decrease in a 0.161 causes 1% increase Interestingly, and negative the significant coef

Sources: fr actualNDVI Thedata isderived of 0.40 0.45 0.50 0.55 0.60 0.65 0.70 NDVI 0121 0121 032014 2013 2012 2011 2001-2010 Figure 5.2 Comparison between the estimated NDVI and actual NDVI Figureestimated NDVI andactualNDVI the 5.2Comparison between grassland Similarly, condition. tohave climate changes seem Systems Program Systems Program in each year lar are of SISGC in each 2011 to 2014 from 1.48 times to 2.54 times the timesto2.54the to2014from 1.48 2011 100 e mean value of actual NDVI intheperiod2001- e mean valueofactualNDVI sults. For that the producershowed instance, data ficient of the real producer priceindicatesthata ofthereal producer ficient vere degradation orinferi vere om the data set of NASA’s Earth ScienceDatathe dataEarth setofNASA’s om the producer price and climate factors which climate factorswhich price and producer the VI is drawn from the mean valueofNDVI of the VI from isdrawn the years of SISGC. The differences betweendifferences The years ofSISGC. the rket forces seem impact to have in a stronger se in the producer price cannotimpactupon se intheproducerprice attributed totheother potential factors that ger than the initial value. theinitialvalue. ger than or grassland condition.or grassland Actual NDVI Actual Estimated NDVI

5 grassland condition directly, strong market forces may stimulate herders to raise much more enforcement strength of each county, would be more meaningful for estimating the livestock. Increased grazing pressure following the large increase in the producer price induces effectiveness of policy intervention than the dummy variables. In our case, because all of the significantly negative impacts on the grassland condition that reduces or neutralises the positive counties in the sample have participated in SISGC since 2011, the dummy variables of SISGC impacts stemming from SISGC, with the result that the actual grassland condition under the do not distinguish the specific differences of implementation of SISGC between counties and SISGC is not as good as expected. between years, and only the differences between before and after SISGC are presented. Moreover, we considered SISGC as a single policy, rather than investigating the results of its 5.4 Discussion two specific regulations, deterministic stocking rates and permanent grazing bans, separately. The effectiveness of SISGC with first an increasing and then a decreasing trend is in line with Our research thus only presented the general result of SISGC. Although we showed that SISGC the results of existing research that questioned the failed implementation and ineffective has a stronger impact in low NDVI counties, the impacts of specific measures need to be management measures of Ecological Construction Programs (Bao and Chen, 1997; Waldron et explored more thoroughly. Finally, our empirical model does not allow disentangling the al., 2010). For example, it is suggested that top-down management entails high supervision impacts of other Ecological Construction Programs that were mainly implemented before costs, making the implementation of the Ecological Construction Programs difficult to sustain SISGC started, such as the Conversion of Cropland to Forest and Grassland Program, Returning for a long period effectively (Li and Zhang, 2009; Waldron et al., 2010). In our case, the local Grazing to Grassland Program, or the Program to Combat Desertification around Beijing and governments, who follow the instructions of the national government as planned at the Tianjin. beginning of SISGC, encountered increasing supervision costs when the producer price of live 5.5 Conclusions sheep soared especially in the third and fourth years of SISGC which prompted local herders to gain more economic benefits by increasing illegal grazing. It thus may incur the implementation This paper began by sketching a review about the policy interventions for grassland protection of SISGC failure during the later years of SISGC. With respect to the ineffective management in China, involving the national Grassland Law and a series of Ecological Construction measures, it has been proved that grassland is in a better condition, including plant species Programs. And then we took the latest Ecological Construction Program (SISGC) as an example diversity and herbage mass, during the early years of the permanent grazing ban, while there is to examine its effectiveness for grassland conservation in the extensive pastoral areas over the a subsequent decline or level-off during its later years (Yang et al., 2005; Zhang et al., 2015). long term. The grassland condition was presented by the NDVI, rendered by remote sensing The ecological explanation for this is that annual and biennial vegetation could invade during technology. The situation of SISGC was defined by dummy variables, which presented the the early years of permanent grazing ban, which would increase species richness. However, the situation of each year with SISGC. The fixed effects model was used to control for the factors subsequent recovery of dominant species increases their prevalence and further leads to a that obscure the relationships between grassland condition and SISGC, including livestock decrease in biodiversity in the next years (Bao and Chen, 1997). Considering that the permanent production, climate factor, time trend and time-invariant heterogeneity between counties. Based grazing ban is widely employed in SISGC, the fact that the grassland condition showed a on the data of 54 sample counties in the pastoral areas of Inner Mongolia from 2001 to 2014, gradual increase at first and then decreased under the impact of SISGC may correspond to this the empirical model results provided quantitative evidence that grassland condition has been process of ecological succession. generally improved under the auspices of SISGC. The comparison between the actual NDVI and the estimated results show that the actual grassland condition is worse than the estimated In addition, we would like to point out several limitations of our research. Firstly, the four years results when other factors are assumed constant. One explanation for this is that the of SISGC which we studied may not be long enough to estimate the changing trend of its effectiveness of SISGC on grassland condition is offset by the impact of other potential factors, effectiveness on grassland condition. Secondly, we used dummy variables to present the status such as producer price and climate changes. Moreover, the impact of SISGC on the grassland of SISGC in our model due to data limitations. Indicators that involve specific differences of condition is stronger in the low NDVI counties, while the offset from producer price and climate SISGC among counties, such as grassland areas under SISGC, amount of subsidies and factors is more pronounced in these counties as well. In conclusion, our research findings are

101 102

The effectiveness of SISGC withfi ofSISGC The effectiveness 5.4 Discussion SISGC expected. is not as goodas SISGC,impacts stemming withtheresult from significantly negative impactsonthegrassland grazingIncreased followingthe pressure livestock. directly,grassland strong condition forces market SISGCcounties, suchasgrassla among of Ind differences of SISGC duetodatalimitations. icators in ourmodel thatinvolvespecific effectiveness condition. on grassland Secondly, to present used dummy we variables the status its trend of changing estimate the enoughlong may not be to studied which of we SISGC several like to point out In would addition, we ecological succession. process of correspondtothis SISGC may theimpact of thendecreasedunder and at first gradual increase showeda condition grassland the fact that SISGC, the employedwidely in grazing is ban years the (Ba next decrease in biodiversity in further toa leads and ses theirprevalence dominant species increa subsequent of recovery grazing ban,which yearsthe early ofpermanent invadecould during and biennialvegetation annual forexplanation thisisthat The ecological during itslate a subsequent orlevel-off decline is grazing ban,whilethere yearsof thepermanent duringearly diversity mass, and herbage the measures, thatgrassland ithasbeenproved is With respectyears to the ineffective management of of SISGC.SISGC the later failure during gain by benefits moreeconomic illeg increasing the third and fourthsheepespecially soared in oflive beginning costswhentheproducerprice supervision increasing ofSISGC, encountered followtheinstructions governments, of who thelocal case, In our al., 2010). et Waldron 2009; Zhang, and (Li effectively period long for a costs, making theimplementationofEcological high supervision entails management itissuggested thattop-down Forexample, al., 2010). management Ecological Construction measures of etand Chen,1997;Waldron Programs (Bao research thatquestione existing the resultsof

rst trendisin anincreasingline with andthenadecreasing nd areas underSISGC,nd areas amountofsubsidiesand o and Chen, 1997). Considering that the permanent Chen, 1997). Consideringthatthepermanent and o 101 limitations of our research. Firstly, the four years four Firstly, the research. limitations ofour years ofSISGC local herders whichprompted to condition that reduces or neutralises thepositive thatcondition reduces orneutralises al grazing.It thusmay theimplementation incur r years (Yang et al., 2005; Zhang et al., 2015). etal.,2015). Zhang et al.,2005; years (Yang r that the actual grassland condition under grasslandthe the actual that would increase species richness. However,the richness. increasespecies would d the failed implementation and ineffective andineffective d thefailed implementation in abettercondition,including plantspecies the national government as planned at the atthe asplanned government national the may stimulateherderstoraise muchmore large increase in the producer price induces induces price large in the producer increase Construction Programs difficulttosustain counties in the sample have pa have counties inthesample thanthedum effectiveness ofpolicy intervention would ofeachcounty, strength enforcement factors is more pronounced in these counties as inthese pronounced factors ismore climate priceand theoffsetfrom producer while lowNDVIcondition isstronger counties, inthe climate changes. priceand such asproducer is SISGC grasslandeffectiveness of condition on assumed cons are results when otherfactors grresults showthattheactual and theestimated NDVI the actual comparison between The of SISGC. auspices the generally under improved the empiricalmodelresults providedquantitativ on thedataof54samplecounties Based counties. between heterogeneity time-invariant and factor, timetrend production, climate SISGC, includinggrassland livestock conditionand between relationships that obscurethe year with SISGC. effe fixed situation of The each the presented which by variables, was ofSISGC dummy technology. defined Thesituation byremote byNDVI, sensing rendered condition wasgrassland the presented long The term. in China, involving the nati grasslandpolicy abouta review protection for This paperthe beganby interventions sketching 5.5 Conclusions Tianjin. Grazing toGrasslandProgram, SISGC Conversion started, of Cropla such as the Constructionimpacts ofotherEcological Progr Finally, more thoroughly. empiriexplored our th counties, lowNDVI in impact has astronger showed thatSISGC we ofSISGC. Although generalresult the presented thusonly Our research deterministic stocking regulations, two specific SISGCMoreover, considered as a single we policy, theresults rather thaninvestigating ofits arepresented. SISGC after and before between differences years, onlythe and between do not distinguish the specific of differences to examine its effectiveness for grassland conser grassland for effectiveness its to examine Programs. Andthenwe tookthelatestEcologica

onal Grassland Law andaseries Law Grassland onal rticipated in SISGC since 2011, the dummy variables of SISGC variables of thedummy since2011, inSISGC rticipated ortheProgram CombatDesertification aroundBeijing to and in the pastoral areas of Inner Mongolia fromin thepastoral 2001to2014, areas ofInner Mongolia Moreover, the impact of SISGC onthegrassland theimpactofSISGC Moreover, 102 implementation of SISGCimplementation l Construction Programl Construction as an example (SISGC) well. In conclusion, our research findings are In are findings research well. conclusion,our nd toForest andGrassl rates and permanent grazing separately. bans, andpermanent rates offset by the impact of other potential factors, potentialfactors, offset by theimpactofother vation in the extensive pastoral areas over the the over areas pastoral intheextensive vation cal model does not allow disentangling cal model does not allow the assland condition is worse than the estimated thantheestimated conditionisworse assland tant. One for explanation thisisthatthe e evidence that grasslandcondition has been my variables. In our case, because all of the all ofthe because Incase, our my variables. ams that were mainly implemented before thatwere before ams mainly implemented cts model was used to control for the factors forfactors the control used to cts model was e impacts of specific measures need to be tobe need measures specific e impactsof be more meaningful for estimating the more meaningful be of Ecological Construction ofEcological and Program, Returning Program, and between counties and counties and between

5 that the Ecological Construction Programs such as SISGC are effective for grassland ACKNOWLEDGEMENT: conservation, but that the effectiveness on grassland protection is reduced or neutralised by Funding for this research was supported by grant from the National Natural Sciences other socio-economic and climate factors. For future policy design, an improvement in the Foundation of China (71333013) and INTASAVE-CARIBSAVE Group (ACCC-027). The Ecological Construction Programs should take into account the potential offsetting forces from authors would like to thank Jiliang Hu, Meng Yao and Guanghua Qiao for their help during the other factors related to market and climate conditions. data collection. We are also grateful to the enthusiastic herders for their any help in our survey.

103 104

climate conditions. tomarketother and factorsrelated ProgramsEcological Construction into shouldtake other socio-economic climate factors. and For conservation, but that the effectiveness on gra on effectiveness conservation, butthatthe Construction Programsthat theEcological su

103 ssland protectionisreduc ssland future policy design, an improvement inthe policyfuture animprovement design, ch as SISGC are effective for grassland for effective ch asSISGC are account the potential offsetting forces from forces offsetting from thepotential account ed or neutralised by or neutraliseded by data collection. We are also grateful to the enthusia tothe grateful also are We data collection. authors wouldliketothankJiliang Hu,Meng Ya INTASAVEFoundation ofChina(71333013)and Sciences grant Natural National from by wasFunding the supported for thisresearch ACKNOWLEDGEMENT:

104 o and Guanghua Qiao for their helpduringthe for their andGuanghua Qiao o stic herders for their any help in our survey. helpinour any for their stic herders -CARIBSAVE Group (ACCC-027). The (ACCC-027). -CARIBSAVE Group

5 6. DISCUSSION AND CONCLUSIONS

6.1 Introduction

China is faced with grassland degradation, rapid increases in demand for livestock products and Chapter 6 high levels of poverty among rural households. This dissertation outlines China’s grassland policies and studies the outcomes of the policy interventions for grassland systems in Inner Mongolia. More specifically, this thesis focuses on evaluating the effects of the land tenure reform and Ecological Construction Programs. These policy evaluations essentially show the outcomes of the government-driven transformation of traditional pastoralism and differ from other existing research that has been concerned with the internal process of pastoralism Discussion and conclusions transformation (e.g. Mwangi, 2007).

To date, the transformation of pastoralism has widely been employed by governments around the world by developing sedentarisation instead of nomadism, privatisation of grasslands instead of common use, intensification instead of extensive grazing, commercialisation of livestock systems instead of subsistence, determinative stocking rate instead of opportunistic strategies with stochastic stocking rates, to name a few (Homewood, 1995; Li and Zhang, 2009). Although plenty of studies have discussed the pros and cons of these pastoralism transformation programs from the perspective of ecology, sociology, and other fields (e.g. Suttie et al., 2005; Mwangi, 2007; Li and Zhang, 2009; Hua and Squires, 2015), there is a lack of literature based on large-scale areas and long-term observation to provide empirical evidence for scientific debate. This study employs panel data involving extensive pastoral areas and spanning several decades to quantify the ecological and economic outcomes of the pastoralism transformation in Inner Mongolia, China. Specifically, the impact of land tenure reform represents the outcomes of grassland privatisation instead of communal tenure. The effectiveness of Ecological Construction Programs presents the outcomes of current management systems on the grasslands, such as the grazing ban and deterministic stocking rate instead of stochastic stocking rates. Moreover, the preference for grasslands by stakeholders reflects the perception of grassland resource users on the pastoralism transformation. The conceptual framework for our research is based on Ostrom’s Social-Ecological Systems (SESs) framework (Ostrom, 2007; 2009). Chapters 2 through 5 address four specific research questions.

Chapter 2 starts with studying the perceptions of users on grassland use under the current policy interventions. The specific research question is, what are the individual and social optima of

105 106

Discussion and conclusions Discussion andconclusions Chapter 6 105

interventions. The specific research question is, researchinterventions. Thespecific question Chapter 2startswithstudying theperceptions questions. research fourspecific Chapters 2through 5address is based on Ostrom’s transformation.resource usersonthepastoralism grasslands by stak forMoreover, thepreference such as the grazingand deterministic ban stocking Construction of Programs theoutcomes presents inst of grassland privatisation reform representsthe outcomes landtenure of impact the Inner China.Specifically, Mongolia, economic ecological and the to quantifydecades involving data employs panel study debate. This to long-termobservation areas and on large-scale and Squire Zhang, 2009; Hua andLi Mwangi, 2007; programs fromof ecology, sociology, theperspective transformation and consofthesepastoralism Although studieshavediscussedthepros plenty of strategies withstochasticstocking rates, livestock systems ofsubsistence,determi instead instead intensification instead ofcommonuse, the world by developing sedentarisation instead of nomadism, privatisation instead developing sedentarisation the world by has ofpastoralism To date,thetransformation transformation (e.g. Mwangi, 2007). concer that hasbeen research other existing government-driven transformation outcomes ofthe Th reform Programs. and Ecological Construction tenure oftheland effects the on evaluating Mongolia. thisthesisfocuses specifically, More policies andstudiesthe outcomes ofthepolicy among rural high ofpoverty levels China is faced with grassland degradation,China isfaced rapi with 6.1 Introduction 6. DISCUSSIONANDCONCLUSIONS

Social-Ecological Systems (SESs) framework 2007;2009). (SESs) (Ostrom, Social-Ecological Systems ead ofcommunaltenure.Theef households. This dissertation outlines China’s grassland to name a few (Homewood, 1995; Li and Zhang, 2009). andZhang, Li (Homewood,to nameafew 1995; 106 of users on grassland use under thecurrentpolicy ongrassland useunder of users d increases indemandfor productsand livestock d increases widely been employed by governments around governments around employed by been widely ned with the internal process of pastoralism pastoralism of process withtheinternal ned current management systems on the grasslands, grasslands, onthe management current systems outcomes of the pastoralism transformation in transformationin pastoralism of the outcomes of extensive grazing, of commercialisation of extensive what aretheindividual and social optimaof several spanning and pastoralareas extensive eholders reflects the perception ofgrassland reflectseholders theperception interventions forinterventions grassland systemsInner in native stocking rate insteadofopportunistic native stockingrate The conceptual framework for our research for ourresearch framework The conceptual ese policy evaluations essentiallyese policy showthe evaluations provide empirical evidence for scientific rate instead ofstochasticstocking rates. rate of traditional pastoralism and differ from anddifferfrom of traditionalpastoralism s, 2015), there is a lack of literature based based literature of a lack theres, is 2015), and other fields Suttie (e.g. et al., 2005; fectiveness of Ecological

of grasslands

6 rural land allocation by stakeholders: among cultivated land, grassland, forest and other rural 6.2.2 Grassland privatisation is a significant reason to cause grassland degradation in a land? It is answered by quantifying the preference of different stakeholders over rural land long term allocation, and thereby reflects the different interests of different stakeholders over grassland Land tenure reform in the pastoral areas of China engages in privatising use rights of grasslands use. Chapters 3, 4 and 5 focus on estimating the outcomes of policy interventions targeting the (Banks et al., 2003). The impact of land tenure reform on the grassland ecosystem reflects the grasslands. Chapter 3 addresses the question, what is the impact of land tenure reform on the ecological effectiveness of grassland privatisation. The results of this research prove that grassland condition? It first outlines the process of land tenure reform in the pastoral areas of grassland privatisation has a significant and negative impact on the condition of the grasslands. China and then investigates the changes in grassland quantity and quality that are caused by the More specifically, private use causes the reduction of the total grassland area and leads to the reform. Chapter 4 answers the question, what is the impact of the land tenure reform on degradation of dense and moderate grasslands into sparse grassland in the long run. As such, livestock production? It presents the economic outcome of the land tenure reform by exploring we observe “a tragedy of privatisation” (Guelke, 2003; Li and Huntsinger, 2011), as opposed the effects of the reform on the livestock population and meat output. Chapter 5 focuses on the to the well-known “tragedy of the commons”. question, what is the impact of Ecological Construction Programs on grassland conservation?

It estimates the ecological outcome of one typical Ecological Construction Program (the 6.2.3 Grassland privatisation puts a ceiling on the livestock production of pastoral areas, Subsidy and Incentive System for Grassland Conservation). but it promotes the development of livestock productivity

This final chapter recaps the main findings of each chapter, presents the contribution to the The implementation of land tenure reform is shown to have significant and negative effects on existing scientific debates and draws lessons for policy improvement. It concludes with the the increase in livestock production. That is, land tenure reform that refers to the privatisation research limitation and recommendations for future research. of grassland use rights puts a ceiling on livestock production. This may be a possible reason for why the reform has not been implemented as successfully in the grasslands as in cropland of 6.2 Synthesis of research findings China. However, land reform in itself is unable to offset the impact of other factors that 6.2.1 Different preference on grassland use by different stakeholders stimulate the increase in livestock production in practice. Moreover, the constraining effect of land tenure reform on livestock production steadily decreases following the initial The individual optima of rural land allocation reveal the different preferences of different implementation of the reform and ultimately disappears. Furthermore, the constraining effect stakeholders for grassland use. The social optima illustrate a possible trade-off among of land tenure reform is stronger on the increase of livestock population than on that of meat stakeholders over grassland use. The comparisons between individual optima, social optima output, indicating that land tenure reform improves livestock productivity. and actual allocation reflect the conflicts over grassland use among different stakeholders. In our case study, the optima of ecological and economic authorities for desirable grassland areas 6.2.4 Ecological Construction Programs have significant impacts on preventing are within the range of the social optimum, whereas herders prefer more and farmers prefer less grassland degradation grassland than the social optimum. In addition, the actual area of grassland already exceeds the A series of Ecological Construction Programs have been initiated for grassland conservation in upper range of social optima, implying the further need to reduce the grassland area for the sake the 2000s. This research provides quantitative evidence that the condition of grasslands has of social welfare. This finding appears to contradict the prevailing policy direction of grassland improved significantly under the auspices of Ecological Construction Programs, specifically conservation that attempts to expand the grassland area. under the program of Subsidy and Incentive System for Grassland Conservation (SISGC). However, the ecological effectiveness of the Ecological Construction Programs is offset to some extent by other socio-economic and climate factors, such as market conditions and climate change. This may explain why the actual condition of the grasslands is not as good as had been

107 108

land? rural landallocation bystakeholders:amongcu question, the effects ofthereform onthelivestockpopulation andmeat output.Chapter 5focuses onthe livestock production? thequestion, reform. 4answers Chapter by the are caused quantityandgrassl quality that and in changes investigatesthe China andthen grassland condition? question, the addresses 3 grasslands. Chapter of policy onestimatinguse. Chapters 3,4and5focus theoutcomes targeting interventions the allocation, and thereby reflects the different in

conservation that attempts to expand thegrassland toexpand area. conservation thatattempts This findingof socialwelfare. appearstocontra implying thefurther socialoptima, of upper range grassland thanthesocialoptimum.In addition,th are ofthesocialoptimum,wherea within therange our case theoptimaofecological study, econo and theconflictsovergrand actualallocationreflect individualoptima,socialoptima isons between compar grassland use.The stakeholders over among grasslandstakeholders illustrate for use. apossibletrade-off The social optima The individualoptimaofrural reve land allocation useby stakeholders ongrassland different preference 6.2.1 Different findings6.2 Synthesisof research research for limitation andrecommendations futureresearch. scientificdebatesanddrawsexisting lessons This final chapter recaps the main findings of themainfindingsof recaps This finalchapter SubsidyIncentive forand System Grassland Conservation). outcomeofonetypIt estimatestheecological It isanswered byquantifying theprefere what is the impact of Ecological Cons It firstoutlines the process ofland It presents the economic outcome of the land tenure reform by exploring tenurereform by exploring oftheland economicIt outcome the presents what is the impact of the land tenurereformon what istheimpactofland 107 what is the impact of land tenure reform onthe reform landtenure istheimpactof what for policyIt improvement. withthe concludes terests of different stakeholders over grassland grassland over ofdifferentterests stakeholders dict the prevailingdict policy of grassland direction truction Programs on grassland conservation? each chapter, presents the contribution tothe the presents each chapter, ltivated land,grassland,forestltivated andotherrural nce of different stakeholders over rural land stakeholdersoverrural land ofdifferent nce e actual area of grassland already exceeds the the grasslande actualarea exceeds already of assland use among different stakeholders. In stakeholders. use among assland different need to reduce the grassland area for the sake areaforthesake grassland the need toreduce ical Ecological Construction Program (the s herders prefer more and farmers prefer less less farmers prefer moreand s herders prefer mic authorities for desirable grassland areas grasslandmic authoritiesfordesirable areas al the different preferences of different of al thedifferent preferences tenure reform inthepastoralareasof

change. Thismay why explain condition the actual and climate by some extent socio-economic other However, effectiveness the ecological of the Ec ofSubsidyIncentiveunder theprogram and improved significantly theauspicesof Ec under the 2000s.Thisresearch providesquantitativeev have Programs ofEcological Construction A series grassland degradation impacts6.2.4 EcologicalConstruction havesignificant Programs onpreventing livestockproductivity. output, indicating tenurereform improves thatland increas stronger on the reform is of land tenure implementation ofthereform andultimately disappear s livestockproduction reformland tenure on stimulate in livestoc the increase reform in itself is unable China. However, land why the reform has not as suc been implemented of grassland use rights puts a ceiling on livest the increase in livestock production. That is, land negative effects on and tohave significant reform isshown tenure The implementationofland but itpromotes thedevelopment oflivestock productivity 6.2.3 Grassland privatisationputsaceiling onth commons”. to thewell-known “tragedy ofthe Li 2003; (Guelke, privatisation” tragedy of “a we observe grasslands in moderate and dense degradation of usecausesthereduction private More specifically, grasslanda significant and nega privatisation has of grasslandecological effectiveness privatisati (Banks et al., 2003). The impact of land(Banks tenureet al., 2003). The en areas ofChina pastoral Land tenurereform inthe long term 6.2.2 Grassland privatisation is a significant

k production in practice. Moreove k productioninpractice. ock production. Thismaypossible reason for bea 108

System forSystem Grassland (SISGC). Conservation reason tocause grassland degradation ina reformgrassland on the ecosystem the reflects e of livestock population than on that of meat meat onthatof of livestockpopulationthan e tive impact on the condition of the grasslands. of the tive impact on the condition ological Construction Programs, specifically factors, such as market conditions and as climate market conditions such factors, to sparse grassland in the long run. As such, such, run. As grassland long inthe to sparse on. The results of this research prove that on. The results of this research that prove ological Construction Programs isoffset to cessfully in the grasslands asincroplandofcessfully inthegrasslands tenure reform thatrefers to theprivatisation idence thattheconditionofgrasslands has ofthetotal grassland area andleads tothe of the grasslands is not as good as had been gages in privatising rightsuse grasslandsof to offset the impact of other factors impact tothat offsetof the been initiated for grassland in for conservation been initiated e livestock production of pastoral areas, pastoral e livestock productionof teadily decreases following the initial the following teadily decreases s. Furthermore, the constraining the s. Furthermore, effect and Huntsinger, and Huntsinger, 2011), as opposed r, the constraining effect of constraining of r, effect the

6 expected. In addition, the case of SISGC shows that the impact of Ecological Construction 1990; Behnke 1994; Sneath 1998). Some new arguments have arisen in support of traditional Programs on grassland conservation is stronger in the counties with worse initial grassland pastoralism, emphasising mobility, flexibility and reciprocity of grassland use, which conforms condition, while the effects of market conditions and climate factors were also more pronounced to the attribute of grassland resources with strong temporal variability and spatial heterogeneity in these counties. Finally, the effectiveness of SISGC is shown to experience an initial (Fernandez-Gimenez and Le Febre, 2006). It is argued that traditional pastoralism could increasing but subsequently decreasing trend in the years since SISGC is first implemented. maintain ecosystem stability and biological diversity in terms of the sustainable use of grasslands (Fernandez-Gimenez and Le Febre, 2006; Fan et al., 2015). In particular, the In summary, the findings of the research presented in chapters 2 to 5 suggest that the communal tenure of traditional pastoralism buffers change and uncertainty in grasslands, transformation of traditional pastoralism through policy intervention impacts the grassland allowing pastoralists to maintain a relatively reliable flow of resources from a relatively system significantly and complexly. For instance, the outcomes of grassland privatisation (i.e. unreliable environment (Roe et al. 1998). Indeed, traditional pastoralism has developed flexible land tenure reform) have two different sides. On the one side, it causes further grassland institutions that enable the persistence of production on such patchy and unpredictable low- degradation, but on the other, it is beneficial to the development of animal husbandry in pastoral productivity environments for centuries (Fernandez-Gimenez and Le Febre, 2006; Li and areas by controlling excessive growth of the livestock population while prompting livestock Huntsinger, 2011). productivity. Moreover, Ecological Construction Programs that mainly rely on the control of livestock grazing for grassland conservation has led to an improvement in the grassland Recently, the literature has increasing interests in the common property regime of grasslands, conditions, but this effect is offset to some extent by other socio-economic and climate factors. especially considering the benefits of mobility, flexibility and reciprocity of grassland use, such as mutual aid, pooled labour as well as access to water resources, medicinal herbs and natural 6.3 Contribution to scientific debates shelters, which potentially maintain the ecological and economic resilience of the grassland

6.3.1 The tragedy of the commons versus the tragedy of privatisation systems (Cao et al., 2013). In contrast, the exclusionary land tenure and pasture enclosure are suggested to be counterproductive to improve grassland systems because they threaten the The concept of the tragedy of the commons has long been part of the conventional wisdom in integrity of ecological systems of arid and semi-arid areas, particularly impeding mobility and ecology, economics and political science (McEvoy 1987). It contends that resource degradation resource access in times of disaster (Mwangi, 2007; Galvin 2009). In this regard, the tragedy of is inevitable unless the commons are either privatised or maintained as common land with clear privatisation is put forward (Guelke 2003; Li and Huntsinger 2011), in contrast to the well- rights of entry and use (Hardin 1968). Similarly, the theorem of property rights proposes a clear known tragedy of the commons. In addition, grassland privatisation reduces the amount of land assignment of property rights as a precondition for economically efficient resource allocation available for livestock grazing, further leading to a reduction in the number of livestock that an and environmental sustainability (Coase 1960). Moreover, privatisation is expected to increase individual can potentially own, ultimately resulting in poverty Mwangi (2007). Some scholars agricultural productivity and the wise use and conservation of resources, because the suggested that the highly spatial heterogeneity of grassland resources leads to dramatically transferability of rights under private property arrangements is supposed to ensure that inequitable assignments, especially the forage quality and water access, which may cause severe resources end up with the most productive users (Mwangi, 2007). As such, some scholars and conflicts over grassland use (Yan et al., 2005). The other negative social consequences of the policy makers have repeatedly identified traditional pastoralism as an irrational, ecologically implementation of grassland privatisation include increased labour burdens for each household destructive and economically inefficient system (Homewood, 1995; Kamara et al., 2005; Ho and the loss of labour economies of scale (Yan et al., 2005; Cao et al., 2013). and Azadi, 2010). Given these arguments over traditional pastoralism, especially regarding property rights and However, Hardin’s theory of the tragedy of the commons and Coase’s theorem of property use rights, Chapters 3 and 4 of this research provide the empirical analysis on the ecological rights have been rejected by pastoral specialists, who found that these theoretical findings and economic impacts of the transformation of grassland use rights in China. Chapter 3 shows provided a poor guide toward the sustainable use and management of grasslands (Feeny et al.

109 110

provided a poor guide toward the sustainable use and management of grasslands (Feeny et al. et (Feeny grasslands and managementguide of use the sustainable provided toward a poor byrights rejected spec been pastoral have of tragedy the of Hardin’s theory However, and Azadi,2010). sy economically inefficient destructive and repeatedlypolicy traditi have identified makers resources endupwiththemostproductiveus property under private transferability rights of agricultural productivity and the wise use and environmental 1960). sustainability More (Coase assignment ofproperty rights asaprecondition for rights ofentry anduse (Hardin 1968).Similarl are eitherpr is inevitableunlessthecommons political scienceand ecology, economics (McEvo has l The conceptofthetragedy ofthecommons 6.3.1 Thetragedy privatisation thecommons of of versusthetragedy debates 6.3 Contributiontoscientific conditions, butthiseffect isoffset tosomeex conservation livestock grazinggrassland for productivity. Moreover, Ecological Construction Programs thatmainly relyon thecontrolof growth ofth excessive controllingareas by to other,degradation, itisbeneficial butonthe sides. On two different have land tenure reform) system significantly and complexly. For instan traditionalpastoralismthrou transformation of 5 suggest that the 2 to in chapters presented In research summary, thefindings ofthe isfirstimplemented. SISGC years since increasing decreasing inthe butsubsequently trend SISGC an initial is showneffectiveness of to experience the Finally, in thesecounties. condition, while the effects of market conditions a Programs ongrassland conservation isstronger SISGCIn shows addition,thecaseof expected.

stem (Homewood, 1995; Kamara et al., 2005; Ho al., 2005;Ho et 1995; Kamara stem (Homewood, ialists, who found that these theoretical findings findings theoretical ialists, whofoundthatthese e livestockpopulationwhileprompting livestock ivatised or maintained as common landwithclear common ivatised ormaintainedas tent by other socio-economic and climate factors. climatefactors. and tent byother socio-economic ers (Mwangi, 2007). As such, some scholars and some 2007). scholars(Mwangi, As such, and ers 109 y, proposesaclear theoremofproperty rights the the commons and Coase’s theorem of property of andCoase’stheorem commons the the development of animal husbandry in pastoral development ofanimalhusbandry inpastoral the ce, the outcomes of grassl and conservation of resources, because the has led to an improvement in the grassland ledtoanimprovementhas inthegrassland gh policy grassland intervention impactsthe onal pastoralism as an irrational, ecologically an irrational,ecologically onal pastoralismas nd climate factors were also more pronounced morepronounced nd climatefactors were also y 1987). Itycontends thatresource degradation 1987). ong oftheconventionalwisdomin beenpart arrangements is supposed to ensure that to ensure arrangements issupposed in the counties with worse initial grassland thecountieswithworse in initial that the impact of Ecological ConstructionEcological that the impact of economicallyresource efficient allocation the one side, it causes further grassland further the one side, it causes over, privatisation is expected to increase to increase privatisationover, is expected and privatisation (i.e.and pastoralism, emphasisingmobility, flexibility a 1990; Behnke 1998).Somenewarguments 1994;Sneath and economic the impacts transformation of of gras thisresearch 4of and 3 Chapters use rights, pastora traditional over Given thesearguments ofscale and thelossoflaboureconomies implementation ofgrassland privatisationinclude conflicts over grassland use (Yan et al., 2005). Th the forageespecially inequitable assignments, quali todramatically leads resources grassland of suggested thatthehighlyspatial heterogeneity ultimately resu own, individual can potentially grazing,available for livestock further leading Ingrass addition, commons. the known tragedy of an Li 2003; (Guelke privatisation isputforward intimes resource access 2007; of disaster(Mwangi, ofaridandsemi systems integrity ecological of suggested to be counterproductive to improve gra excl the Incontrast, al., 2013). et systems (Cao maintaintheecologica shelters, whichpotentially aid,pooledlabouras mutual asaccessto aswell consideringbenefitsofmobility,especially flex the interest increasing has literature Recently, the Huntsinger, 2011). Liand LeFebre, 2006; and (Fernandez-Gimenez for centuries productivity environments institutions of thatenablethepersistence Indeed, tr etal.1998). (Roe unreliable environment a relatively from resources of flow reliable relatively maintain a allowing to pastoralists traditionalpastoralismbuff communal tenureof Leand Febre grasslands (Fernandez-Gimenez of sustainableuse ofthe terms in biological diversity and stability maintain ecosystem pastoralismcould traditional It isargued that LeFebre, 2006). and (Fernandez-Gimenez to theattributeofgrassland resources withstrong

(Yan et al., 2005; Cao et al., 2013). al., 2005;Caoet2013). (Yan et production on such patchy and unpredictableproduction onsuchpatchy low- and 110 provide the empirical analysis on the ecological on the analysis empirical the provide s in the common property regime of grasslands, property regimes inthecommon grasslands, of to a reduction inthenumberoflivestockthatanto lting in poverty Mwangi (2007). Some scholars Some scholars (2007). Mwangi poverty lting in nd reciprocity of grassland use, which conforms nd reciprocity grassland use, of usionary land tenure and pasture enclosure are enclosure pasture and tenure usionary land -arid areas, particularly impeding mobility and mobility areas,particularly impeding -arid and lism, regardingespecially lism, property rights and , 2006; FanIn, et al., 2015). the particular, temporal variability a variability temporal increased labour burdens for each household labour burdensincreased for eachhousehold water resources, medicinwater resources, d Huntsinger 2011), in contrast to the well- to the contrast in d Huntsinger2011), land privatisation reduces the amount of land land of the amount land privatisation reduces e other negative social consequences ofthe consequences social e othernegative ers change and uncertainty in grasslands, grasslands, in and uncertainty ers change ty access,whichmayand water severe cause aditional pastoralism has developed flexible developed aditional pastoralismhas ibility and reciprocity of grassland use, such grassland use,such of ibilityreciprocity and l andeconomicresilien Galvin 2009).In thisregard, thetragedy of sland userights inChina.Chapter 3shows ssland systems because they because threatenthe ssland systems have arisen in support of havetraditional arisen in support nd spatial heterogeneity heterogeneity nd spatial al herbsandnatural cegrassland of the

6 that grassland privatisation causes grassland degradation in pastoral areas. Chapter 4 finds that unlikely (Sullivan and Rohde, 2002). As such, the non-equilibrium model embraces grassland privatisation places a ceiling on the livestock production of pastoral areas, while it opportunistic strategies with high and stochastic stocking rates in which the number of livestock stimulates the development of livestock productivity. Hence, this research supports the idea of is maximised according to the availability of forage, and grassland degradation is not caused a tragedy of privatisation, drawing from the negative ecological outcomes of privatisation. by overgrazing. Nevertheless, grassland privatisation is beneficial to the development of local animal husbandry The debates over grassland management between the equilibrium model and the non- considering the effects of controlling excessive growth of the livestock population and equilibrium model eventually are reflected in the question of whether the livestock population prompting livestock productivity, showing that grassland privatisation can stimulate herders to can be allowed to increase without the threaten of grassland degradation (Vetter, 2005). The increase efforts to improve the output per animal, instead of increasing the number of animals. Chinese policy interventions provide an outstanding example of managing grasslands based on 6.3.2 Deterministic stocking rates versus stochastic stocking rates the equilibrium model. It is believed that overgrazing is the main cause of grassland degradation, therefore the control of the livestock population has been employed for grassland conservation. The control of the livestock population has been the most widespread approach to halting Chapter 5 of this research investigates the ecological outcomes of the policy intervention of grassland degradation in China, Africa and other parts of the world since overgrazing became deterministic stocking rates and grazing bans in Inner Mongolia, China. It suggests that control a concern in the 2000s (Vetter, 2005). However, a debate around the validity of the control of of the livestock population improves the grassland condition significantly in the arid and semi- the livestock population for preventing grassland degradation in the arid and semi-arid areas arid areas of China in the short term. However, there are some doubts over the long term, as the has emerged, often referred to as a debate between the equilibrium model and the non- effectiveness of grassland conservation measures appears to decrease in later years. As such, equilibrium model (e.g. Clements, 1916; Ho, 2001; Vetter, 2005). The equilibrium model, or this research agrees that grassland management based on the equilibrium model might be Clementsian succession theory (Clements, 1916), advocates the stability and predictability of effective for grassland conservation in the arid and semi-arid pastoral areas temporarily, but grasslands, which stresses the importance of biotic feedbacks such as density-dependent that the non-equilibrium model may be more valid for long-term strategies. regulation of livestock population and the feedback of livestock density on vegetation composition, cover and productivity (Ho, 2001; Vetter, 2005). In other words, the equilibrium 6.3.3 Top-down intervention versus bottom-up intervention model suggests that the stocking rate is predictable and deterministic, and grassland degradation The debate over concrete uses and management strategies for natural resources is still ongoing: occurs if the carrying capacity25 is exceeded due to overstocking. The use and management of in particular, whether a top-down or a bottom-up approach can best achieve conservation goals grasslands based on the equilibrium model therefore focus on deterministic and low stocking (Oestreicher et al., 2009; Mauerhofer, 2011). The top-down approach assumes that legislation rates, i.e. controlling the livestock population for grassland conservation. and policies set explicit aims and objectives, providing a blueprint that is then directly translated In contrast, the non-equilibrium model advocates that the grassland condition in the arid and into action on the ground (Urwin and Jordan, 2008). Top-down intervention generally follows semi-arid areas are primarily driven by stochastic abiotic factors which result in highly variable a decision by government authorities and policy makers. Although such official acts in the and unpredictable primary production, particularly due to the factor of precipitation (Ho, 2001). public interest are considered capable of designing optimal rules to govern resource use for an The argument carried further implies that grassland degradation is part of a natural process of entire region, this prescription is not supported by empirical research (Ostrom, 1999). The top- vegetation decline and growth in response to precipitation (Ho, 2001). The grassland use and down approach has been criticised as leading to poor compliance with statutory law and limited management based on the non-equilibrium model claims that the livestock population has local community participation in monitoring and enforcing natural resource conservation negligible feedback on the vegetation as the number of animals is kept below equilibrium regulations (Hartter and Ryan, 2010). Hence, top-down resource management strategies have densities by frequent droughts, and that grassland degradation as a result of overgrazing is thus been linked to marginalising poor populations and exacerbating natural resource depletion (Oestreicher et al., 2009). 25 Carrying capacity indicates that grazing pressure is counterbalanced by the natural regeneration of plants.

111 112

occurs if the carrying capacity occurs if the carrying ispredictabl stocking rate model suggeststhatthe 2001;Ve and productivity (Ho, composition, cover regulationthe fee oflivestockpopulationand grasslands, which stressestheimportanceofbi (Clements,1916), Clementsian successiontheory Clements, 1916;Ho,200 equilibrium model(e.g. has emerged, often referred to as a debate the livestockpopulation forgrassland preventing a de However, 2005). 2000s(Vetter, inthe a concern inChina,Africaandot grassland degradation The control of the livestock population has been 6.3.2 Deterministic rates stocking ve toimprove efforts theoutputperanimal increase gr prompting showingthat productivity, livestock of controllingexce effects considering the Nevertheless, grassland privatis drawing the from a tragedyofprivatisation, stimulates thedevelopmentoflivestockproductivity. Hence, thisresearch supportstheidea of grassland privatisation places a ceiling on th grasslandthat grassland causes degrad privatisation 25

densities by frequentdroughts, andthatgrassla nu vegetation as the on the feedback negligible mo non-equilibrium management onthe based vegetationgrowth decline in response and to The argument further carried implies that grassl particularly and unpredictableproduction, primary semi-arid are primarily areas driven by stochastic In thenon-equilibrium modeladvocates contrast, livestock populationforgrasslandrates, theconservation. i.e. controlling equilibrium modelther onthe grasslands based Carrying capacityindicates that grazing pressure is 25 ation is beneficial to the deve beneficialto is ation is exceeded due to overstocking. The use and management of The use and due to overstocking. is exceeded rsus stochastic stocking rates stocking rates rsus stochastic counterbalanced by the natural regeneration of plants. counterbalanced bynatural the regenerationof plants. e livestockproductionof precipitation (Ho, 2001). The grassland use and 111 ssive growth of the livestock populationand livestock growthssive ofthe her parts of the world since overgrazing became became overgrazing parts of since the world her between theequilibriummodelbetween andthenon- nd degradationnd as a of result overgrazing is thus negative ecological outcomes of privatisation. outcomes ecological negative efore focus on deterministic and low stocking andlow ondeterministic efore focus and degradation natural process ispartof a of and vestock population has del claimsthatthelivestock , instead of increasing the number of animals. insteadnumber of of increasing animals. the , e and deterministic, and grassland degradation and and deterministic, e abiotic factors which result inhighly result factorsabiotic which variable assland privatisation can stimulate herders to herders stimulate assland privatisationcan mber of animals is kept below equilibrium below animalsiskept mber of thatthegrassland conditionin the aridand 1; Vetter, 2005). The equilibrium model, orequilibrium The 2005). 1; Vetter, dback oflivestockdensity onvegetation due to the factor of precipitation (Ho, 2001). duetothefactor ofprecipitation otic feedbacks such as density-dependent otic feedbacks suchasdensity-dependent advocates the stability and predictability thestabilityof advocates degradation in the arid and semi-arid areas areas degradation inthearidandsemi-arid tter, 2005). In other words, theequilibrium In words, other tter, 2005). ation in pastoral areas. Chapter 4finds that Chapter ation inpastoral areas. the mostwidespread approach tohalting bate around the validity of the control of control of the of validity around the bate lopment oflocal animalhusbandry pastoral areas, while it opportunistic strategies high with unlikelyand Rohde,2002).Assu (Sullivan (Oestreicher et al., 2009). et al.,2009). (Oestreicher natu exacerbating been linkedtomarginalising poor populationsand Hence, to Ryan,regulations 2010). (Hartterand local community inmonitoring participation down approach criticised has been as leading to 1999).Thetop- (Ostrom, empirical research isnotsupported by entire thisprescription region, public interestconsidered are ca a decision bygovernment authorities and policy makers. Althoughofficial acts in the such into actionon 2008).the groundand Jordan, (Urwin Top-downintervention generallyfollows aims and explicit and policiesset 2011).Th etal.,2009;Mauerhofer, (Oestreicher in particular,whether atop-downorbottom-up uses concrete andThe managementdebate over stra 6.3.3 Top-down intervention intervention versusbottom-up that thenon-equilibriummodelmay m be effectivegrassland for conservation in the arid this researchagrees that grassland manage measures conservation grassland of effectiveness arid areas of China in the short term. However, grassland improvesc the of thelivestockpopulation Ingrazing in bans and deterministic stockingrates Chapter of 5 outcomes of of this the researchpolicy investigates intervention ecological the therefore thecontroloflivestockpopulation Itthe equilibrium model. is believed that overg an outstandi provide Chinese policy interventions withoutthethrea toincrease allowed can be equilibrium model eventually reflected are in thequestionofwhether thelivestockpopulation management grassland The debatesover betw by overgrazing. of fo theavailability accordingis maximised to

pable of designing optimal rules to govern resource use for an govern an resource usefor optimalrulesto designing pable of objectives, providing a blueprint that is then directly translated translated directly thatisthen a blueprint providing objectives, and stochasticstocking rates ore valid for long-term strategies. strategies. for long-term ore valid ment based ontheequilibriummodelmightment be ten of grassland degradation (Vetter, 2005). The 2005). (Vetter, grassland degradation ten of 112 razing degradation, grassland isthemaincauseof and enforcing naturalresource conservation and there are some doubts over the long term, as the poor compliance withstatutory lawandlimited and semi-arid pastoral areas temporarily, but temporarily, pastoralareas semi-arid and has been employed grasslandconservation. for rage, and grasslandrage, anddegradationis not caused p-down resource management strategies resource management have p-down e top-down approach assumes thatlegislation approach assumes e top-down approach can best achieve conservation goals goals conservation approach bestachieve can ng example of managing grasslands based on on grasslandsbased of managing example ng ch, the non-equilibrium model embraces ner Mongolia, China. It control suggestsner Mongolia,China. that een theequilibriummodelandnon- appears to decrease in later years.As such, inlater todecrease appears tegies for isstillongoing: resources natural ondition significantly in thearidandsemi- ondition significantly in in which the number of livestock in which the number of livestock ral resourcedepletion ral

6 A bottom-up approach that aims to guide self-organisation towards desirable states (Ball and the policy intervention impacts grassland condition significantly and positively, but the Helbing, 2012), recognises the decisions and strategies of local resource users in managing effectiveness on grassland conservation is not as good as expected because of the resources (Castella et al., 2007). It has been suggested that sustainable natural resource counterbalancing effect of other factors. In this regard, the policy improvement of grassland management directives must be initiated from the grassroots level, where the problems and conservation in China could consider more bottom-up self-organisation interventions that solutions are perceived as local (Hartter and Ryan, 2010). Research in multiple disciplines has involve learning about local market forces, climate conditions, and so on. found that some top-down policy interventions accelerate resource destruction, whereas self- 6.3.4 The one-size-fits-all versus tailor-made policies for resource governance organisation by some resource users has been successful in resource conservation (Ostrom, 2009). Field studies also have found that local resource users, with or without the assistance of Scholars have tended to develop simple theoretical models to analyse aspects of resource external authorities, have created a wide diversity of institutional arrangements for coping with problems and to prescribe universal solutions for resource use (Ostrom, 2009). Governmental 26 common-pool resources (Ostrom, 1999). Moreover, a series of relatively autonomous, self- agencies often impose one-size-fits-all policy for resource governance. However, the universal organised resource governance systems might do a better job of regulating small common-pool policy solution frequently fails (Pritchett and Woolcock, 2004; Ostrom, 2009). In particular, resources than a single central authority. A decentralised, polycentric, bottom-up approach, national government agencies are often unsuccessful in their efforts to design effective and involving multiple institutions instead of a single global one, may provide better conditions uniform sets of rules to regulate important common-pool resources across a broad domain both for cooperation to thrive and for ensuring the maintenance of such institutions (Ostrom, 1999). (Vasconcelos et al., 2013). Local resource users, who share moral and ethical standards for how to behave in the groups they form, face lower transaction costs in reaching agreements and In the case of China, land tenure reform started with the Household Production Responsibility lower costs of monitoring (Ostrom, 2009). In terms of grassland management, Sneath (1998) System in cropland areas in the late 1970s and spread to the pastoral areas by the early 1980s. showed that the grasslands under central government management in China and Russia are The reform intended to reallocate communal land to peasant households. In the cropland areas, associated with more degradation than those under traditional, self-organised group-property the household-based use rights system was established rapidly, and 94.2% of rural households regimes in Mongolia. had adopted the system by the end of 1983 (Lin, 1987; Banks et al., 2003). Chinese scholars showed that this reform of cropland use rights created millions of smallholders with relative Scholars have increasingly embraced the bottom-up, self-organised approach to policy autonomy over land use decisions and crop selection (Tilt, 2008). Moreover, the household- intervention. The research cases by Schoon (2008) found that the bottom-up approach increases based farming system improved the incentive structure for agricultural production when cooperation and coordination at the operational level, but that top-down approach can generate monitoring was difficult within the commune system (Lin, 1987). This assignment of cropland successful cooperation at higher levels of governance. Policy interventions in grassland resulted in the growth of agricultural output and income in the reform period. conservation in China have generally been conducted via a top-down approach, starting with explicit aims and quotas that are distributed from the central government to the provinces, However, in the pastoral areas, the implementation of the land tenure reform is not as successful followed by subsequent distribution down through counties, townships, and finally to as was the case in the cropland areas. The assignment of grassland use rights was implemented participating villages (Bennett, 2008). This is criticised because of the lack of sufficient slowly and was still in progress after 30 years of efforts, and a second round of reform to consultation with local communities and rural households to identify their needs and learn the accelerate the implementation in the 1990s. The adoption of actual private use by individual local constraints (Bennett, 2008). Chapter 5 focuses on investigating the ecological outcome of households evidently lagged behind the assignment of formal use rights by governments. The top-down interventions into grassland conservation in China. The research results reveal that research results in Chapter 3 showed that the land tenure reform was detrimental to the

grassland ecosystem, although it is beneficial to the development of the animal husbandry 26 A common-pool resource, such as an irrigation system, a fishing ground, a forest, or open access grasslands, is system of pastoral areas according to the research results in Chapter 4. The conclusion can be a natural or man-made resource, from which it is difficult to exclude or limit users once the resource is provided, where one person’s consumption of resource units makes those units unavailable to others (Ostrom et al., 1994).

113 114

a naturalman-made or resource, fromwhich difficult it is where one person’s consumption of resource un top-down interventions into grassl top-down interventions into of ecological outcome focuses oninvestigatingthe 5 2008).Chapter local(Bennett, constraints consultation withlocal communitiesandrural hous participating(Bennett, 2008). This is villages t distributiondown followed by subsequent aimsandquotasexplicit thatare distributed conservation gene inChinahave at successful cooperation higher of gove levels atth and coordination cooperation intervention. approach foundthatthebottom-up Theresearch casesby increases (2008) Schoon topolicy approach self-organised bottom-up, embraced the increasingly Scholars have regimes inMongolia. thanthose associated withmoredegradation gove central under grasslands showed thatthe lower costsofmonitoringIn (Ostrom, 2009). to behaveinthegroupsthey face lower form, Local (Vasconcelos et al., resource 2013). users, ensu thriveandfor both for to cooperation involving multipleinstitutions insteadofasingle global one,may betterconditions provide A central authority. resources thanasingle organised systems mightgovernance resource do common-pool resources wide diversit createda authorities,have external 2009). Field studiesalso have foundthatlocal resource users, withorwithout theassistanceof been has resourceusers some organisation by found thatsometop-downpolicy interventions ac and Ry solutions are local as perceived (Hartter fr must managementbe initiated directives has It been 2007). al., et resources (Castella Helbing, the decisions and strategiesusers 2012), in managing recognises resource of local A bottom-upapproach thataimstoguide self-org 26

A common-pool resource, such ground, asansystem, aforest, irrigation afishing oropen access grasslands, is 26 autonomous, self- relatively of (Ostrom, aseries 1999).Moreover, rally been conducted via a top-down approach, starting with starting approach, viaatop-down conducted rally been e operational level, but that top-down approach can generate can approach e operational but that top-down level, and conservation in China. The research results reveal that research results that reveal inChina.The and conservation its makesunitsunavailableto others (O those its om the grassroots level, where theproblemsand 113 decentralised, polycentric, bottom-up approach, approach, bottom-up polycentric, decentralised, under traditional, self-organised group-property under traditional,self-organised from the central government the centralfrom to the provinces, terms ofgrassland management, Sneath(1998) an, 2010).Research inmultipledisciplineshasan, rnment management in China and Russia are management andRussiaare rnment inChina hrough counties, townships, and finally to to and finally hroughcounties, townships, suggested that sustainable natural resource resource natural suggested thatsustainable y of institutional forarrangements with coping criticised because of the lack of sufficient lackofsufficient ofthe criticisedbecause successful in resource conservation (Ostrom, (Ostrom, conservation resource in successful to exclude orto limit users resource once is provided, the who share moral and ethical standards for how how standards ethical share for who and moral transaction coststransaction in reaching agreements and ring the maintenance of suchinstitutions of ring themaintenance a better job of regulating small common-pool common-pool betterjobofregulatinga small anisation towards desirable states (Ball and states(Ball towardsdesirable anisation celerate resource destruction, whereascelerate resource destruction, self- rnance. Policy interventions in grasslandin Policy interventions rnance. eholds to identify the needs and their learn strom etal.,1994). as was the case in the cropland areas. Theassi areas, theimplementation inthepastoral However, In effectofotherfactors. counterbalancing the of because expected as asgood isnot conservation grassland on effectiveness condi impacts grassland the policy intervention system of pastoral areas according to the research the research according to system of pastoral areas it isbeneficial grassland although ecosystem, research detrimental to the in Chapterland tenure results 3 showed reformthat the was of rights formal Theassignment byhouseholds behindgovernments. the use evidently lagged The implementation the in the 1990s. accelerate year progress 30 after wasslowly stillin and resulted int monitoringwithin the was difficult based farming system improved theincentive crop se decisions and use land autonomy over showed thatthisreform ofcroplanduserights byhad adoptedthesystem theendof1983(Lin rightses was use system the household-based The reform intended to reallocate communal land by theearly inthelate1970sandspread 1980s. tothepastoral areas System incroplandareas reform starte tenure China,land of case In the (Ostrom, 1999). regulateuniform sets of rules to important unsuccessful arenational government agencies often policy solutionfrequentlyfailsand (Pritchett policy imposeone-size-fits-all agencies often solutions universal to prescribe problems and todevel tended Scholars have 6.3.4 tailor- versus The one-size-fits-all force marketinvolve learning aboutlocal considermore could conservation inChina

he growth inthereform period agriculturalhe output andincome of op simple theoretical models to analyse aspects of resource ofresource aspects modelstoanalyse op simpletheoretical commune made policies for resource governance resource made for policies s, climate conditions, and so on. s, climateconditions,andsoon. this regard, thepolicy ofgrassland improvement system (Lin, 1987). This assignment of cropland assignment cropland system (Lin, of 1987). This common-pool resources abroad across domain common-pool 114 tablished rapidly, and 94.2% of rural households of rural households tablished rapidly, and94.2% for resource use (Ostrom, 2009). Governmental 2009). (Ostrom, resource use for gnment of grassland use rights was implemented grasslandgnment of rights use for resource governance. However, the universal universal the However, governance. for resource d with the Household Production Responsibility theHouseholdProduction with d Woolcock, 2004;Ostrom,2009).In particular, bottom-up self-organisation interventions that self-organisation bottom-up lection (Tilt, 2008). Moreover, the household- lection s of efforts, and a second round of reform toreform roundof a second and of efforts, s to the development of the animal husbandry animalhusbandry the thedevelopmentof to created millions of smallholders withrelative millionsofsmallholders created , 1987; Banks, et al., 2003). Chinese scholars adoption of actual private use by individual use actualadoption private of to peasant households. In the cropland areas, Into peasant households. the cropland areas, structure for agricultural production when of the land tenure reform is notassuccessful reform of thelandtenure results in Chapter 4. The conclusion can be The canbe conclusion results inChapter4. tion significantly and positively, but the and positively, but the tion significantly in their efforts to design effective and .

6 made that the land tenure reform that seemed suitable for managing cropland resources may not privatisation, tearing down wire fences between the patches, repealing the deterministic fit the management of grassland resources. This result is in line with the research finding that stocking rates and further resuming nomadism (Wu and Du, 2008). However, there has also individual property may not be suitable for most pastoral settings in which resource productivity been criticism of traditional pastoralism as being incapable of coping with the rapidly increasing is variable due to environmental variability (Mwangi, 2007). population pressure and economic demands relying on finite grassland resources, because traditional pastoralism is primarily subsistence-oriented. In other words, traditional pastoralism With respect to resource conservation, the Chinese government initiated a series of ecological may only operate well during the phases of low economic demand and minimal environmental programs for ecological restoration and resource development in its forest sector in the late degradation. As such, it is important to transform the traditional pastoralism into a model that 1990s, including the Natural Forest Protection Program, the Sloping Land Conversion Program, can adapt to contemporary socio-economic conditions. the Desertification Combating Program around Beijing and Tianjin, the Shelterbelt Development Program, and the Wildlife Conservation and Nature Reserve Development The case of Inner Mongolia presents lessons for the improvement of the government-driven Program (Liu et al., 2010). Plenty of scholars have shown that these ecological programs for transformation of traditional pastoralism in arid and semi-arid pastoral areas. In particular, it forest conservation have significant and positive impacts on improving resource conditions as shows that the assignment of grassland use rights to individual households has negative impacts well as increasing rural households’ income (e.g. Liu et al., 2010; Mullan et al., 2010). on improving the grassland condition. On the other hand, the privatised use rights have put a Following these ecological programs for forest conservation, a series of programs have been ceiling on the livestock production of pastoral areas, while it promotes the development of initiated to protect the grasslands of China, especially during the 2000s. In general, sowing livestock productivity. Some scholars have mentioned that the household is not an efficient unit grass and reducing grazing (such as with a grazing ban) are imposed as key components for for meeting the demands of large-scale animal husbandry, and household-based privatisation grassland conservation, which coincide with the main measures of planting trees and restricting has provided a rapid increase in livestock production costs in the pastoral areas of China (Li et logging (such as with a logging ban) for forest conservation (Mullan et al., 2010). Chapter 5 of al., 2014). Admittedly, the cooperative use of grasslands on a large scale offers the opportunity this research shows that the effectiveness of the large-scale ecological programs differs across to benefit from the maximum heterogeneity in the landscape. Given the status quo of different regions with the same policy context. In addition, Chapter 2 of this research shows privatisation implemented in pastoral societies, a possible solution is to stimulate the local that different stakeholders have divergent preferences in different types of rural land. Given community toward cooperative use and management of grassland resources based on the these research results, it is evident that there is no one-size-fits-all policy for the governance of existing privatised property rights or use rights of grasslands. In this regard, the privatisation of cropland, grassland and forest land, as well as for the interests of different stakeholders. In this grassland property rights or use rights could be viewed as a precondition for guaranteeing stable case, tailor-made policies that take into account the specific circumstances, such as the resource cooperation. The cooperative households who reach an agreement are encouraged to remove types and stakeholders, are needed for resource governance. their fences and pool their grassland resource in order to provide a better match for forage and water availability. Besides the grassland resource, other agricultural inputs like labour and 6.4 Policy implications capital could be pooled for efficient distribution as well. Economies of scale in production and With severe grassland degradation, high population pressure and increased production the exclusion of outsiders are also important (Mwangi, 2007). The proposal of cooperative use requirements, the sustainable use and management of grasslands, which enables the is in line with the research findings of Mwangi (2007) that showed how parcel re-aggregation maintenance of productive capacity without compromising the ecological interests of human can offer promise for ecosystem sustainability as well as for the economic sustainability of communities and the ecosystem itself, are essential. An increasing number of scholars support African grasslands. the model of traditional pastoralism as one approach for sustainable grassland use, due to its In addition, a wide diversity of institutional arrangements for grassland use are needed, advantages of mobility, flexibility and reciprocity that fit grassland use for heterogeneously considering the heterogeneity in grassland resources per se and the different socio-economic distributed resources. In this respect, some have suggested abolishing the existing grassland

115 116

With severeWith grasslanddegradation, high 6.4 Policy implications governance. resource for needed are stakeholders, types and sp thattake the intoaccount case, tailor-made policies cropland, grassland and forest land, as well as for results, itisevidentthatthere is these research divergent have pref stakeholders that different policycontext. with thesame regionsdifferent this research programs shows ecological that the effectiveness differs of across the large-scale 2010).Chapter5of et al., (Mullan forest conservation for ban) with alogging logging as (such grassland conservation, which coincide with the keycomponents forareas withagrazinggrazing imposedas ban) (such grass andreducing China, grasslands of initiated toprotectthe Following programs ecological these forest for co households’income(e.g.well rural Liu asincreasing etal.,2010;Mullan2010). posit and significant have conservation forest Program Plenty (Liu etal.,2010). ofscholars and theWildlifeDevelopment C Program, theShelterbelt and Tianjin, Beijing around Program Combating the Desertification 1990s, including theNatural Forest Protection de programs restoration for andresource ecological Chines conservation, respect the With toresource variability (Mwangi,2007). is variableduetoenvironmental individual property besuitablefor most may not resu This resources. grassland of fit themanagement not may resources for cropland suitable managing reform seemed tenure made that thattheland distributed resources. In this respect, some have suggested abolishing the existing grassland existing suggestedabolishingIn the this respect, distributed resources. somehave advantages of mobility, flexibility and reciproc the modeloftraditional pastoralism asoneappro essen communities andtheecosystem itself, are human interests of ecological the without compromising capacity productive maintenance of requirements, the sustainable use and manag ement ofgrasslands, whichenablesthe

populationpressure andincreased production especially during the 2000s. In general,In sowing the 2000s. especially during 115 ive impacts on improving resource conditions as conditionsas resource impactsonimproving ive onservation and Nature Reserve Development Development onservation Reserve andNature Program, theSlopingLand Program, Conversion no one-size-fits-all policy for the governance of governance of the policy for one-size-fits-all no have shown that these ecological programs for programs for shownthattheseecological have erences in different types of rural land. Given of ruralland.Given types indifferent erences In addition, Chapter 2 of this research shows shows this research of 2 In addition,Chapter pastoral settingspastoral in productivity which resource tial. An increasing number of scholars support Anincreasing ofscholarstial. support number main measures ofplanting treesandrestricting ity that fit grassland use for heterogeneously heterogeneously that fitgrasslandity use for the interests of different stakeholders. In this stakeholders. ofdifferent the interests e government initiated a series of ecological ecological of series initiated a government e ach for sustainable grassland use,duetoitsach for sustainable nservation, a series of programs have been havebeen ofprograms nservation, aseries velopment initsforest sectorinthelate lt is in line with the research finding that research with the lt isinline ecific circumstances, such as the resource resource asthe circumstances, such ecific considering the heterogeneity in grassland res In addition,awidediversity ofinstitutional African grasslands. can offer promisefor ecosystem sustainabilityas is in line with theresearch findingsMwang of ofoutsidersarethe exclusion alsoimportant efficient distribut for capital couldbepooled water Besides availability. thegrassland resourc their fences andpooltheirgrassland resource in households who reach cooperative cooperation. The grassland rights orproperty use rights could be privatisedpropertyexisting rights oruserights community useandmanag cooperative toward to benefit from the maximum heterogeneity al., 2014).Admittedly,gra cooperative useof the in livestockproduc a rapid increase has provided privatisation implementedinpa for meeting animal of large-scale the demands mentione have Somescholars livestock productivity. ceiling onthelivestockproductionofpastoral on improvinggrassland condition.Ontheother the shows thattheassignment ofgrassland userights pastoralisminarid transformation oftraditional Inner presents of Mongolia The lessonscase the improvement forgovernment-driven of the conditions. socio-economic tocontemporary can adapt important to transform degradation. As such, it is may only operate wellduring thephases of traditional pastoralismis primarily subsistence- population pressure economic demands rely and pastoralismasbeing been criticismoftraditional stocking rates and further resuming nomadis privatisation, tearing down wire fences betw wirefences down privatisation, tearing

storal a possible societies, soluti low economic demandand minimal environmental ion aswell.Economies (Mwangi, 2007). The proposal of cooperative use ofcooperative (Mwangi,The proposal 2007). m (Wu and Du, 2008). However, there has also has However,there andDu,2008). m (Wu 116 i (2007) that showed how parceli (2007) re-aggregation ources per se and the different socio-economic socio-economic perseandthedifferent ources viewed as a precondition for guaranteeing stable guaranteeing preconditionfor stable a as viewed of grasslands. In this regard, the privatisationofIngrasslands. of thisregard, the oriented. In other words, traditional pastoralismIn otherwords,traditional oriented. in the landscape. Given the statusquoof thelandscape.Given in husbandry, and househol een the patches, repealing thedeterministic repealing thepatches, een order to provide a better match for forage and matchforand forage toprovideabetter order incapable of coping withtherapidly coping increasing of incapable arrangements for grassland use are needed, useare needed, arrangementsgrassland for areas, whileitpromotestheareas, developmentof tion costs in the pastoral areas of China (Li of et China areas costs in the pastoral tion sslands on a large scale offerssslands onalarge theopportunity scale to individualhouseholdshas negativeimpacts and semi-arid pastoral areas. In particular, it In particular, areas. pastoral and semi-arid ement ofgrasslandement basedonthe resources e, other agricultural inputs likelabourande, otheragricultural inputs the traditional pastoralism intoamodelthat the traditionalpastoralism ing on finite grassland resources, because resources, onfinitegrassland because ing well asfor theeconomic of sustainability d that the household is not an efficient unit not anefficient unit householdis d thatthe hand, the privatised use rights have put a have put rights privatised use hand, the an agreement are encouraged to remove are encouraged toremove anagreement on is to stimulate the local on istostimulatethelocal of scale inproductionand of scale d-based privatisation d-based

6 and cultural backgrounds of pastoral societies, instead of extending the same policy measures 6.5 Limitations and recommendations for future research designed for crop-farming areas to pastoral areas. This study has found that the same land tenure To conclude, I would like to point out a number of limitations of this study and challenges for reform had different policy effectiveness on grassland and cropland. Especially some future research. First, I follow Ostrom (2007)’s recommendation that the problems linked to unintended effects were observed on grassland systems, which suggest the need to clarify the SESs require serious study of complex, multivariable, nonlinear, cross-scale and changing different attributes between grassland systems and cropland systems before transforming the systems, and that more knowledge about specific variables of SESs, the interactions between property rights or use rights of grasslands. Clearly, the top-down decision by central variables and dynamics of SESs need to be explored more deeply. More specifically, in Chapter governments for resource management makes it difficult to consider all of the specific attributes 2, we assume four categories of rural land, four stakeholder types and three scenarios to simplify of each resource at local levels, but it is advocated that resource users who understand the local the reality of land allocation in the SES. For further research, the more components and the Social-Ecological Systems (SESs) well are able to use the resource more efficiently through more complicated interactions between the components of rural land allocation linked SESs self-organisation and self-crafted institutions (e.g. Ostrom, 1999; Baerlein et al., 2015). Users need to be considered. Moreover, chapters 3, 4 and 5 present general outcomes of the policy that have full autonomy at the collective-choice level to craft and enforce some of their own interventions in grassland systems, whereas the interaction between grassland condition and rules, also confront lower transaction costs (Ostrom, 2009). As such, it is recommended to livestock production is a significant aspect to be studied more concretely. create specific institutional agreements for grassland use by local communities. This bottom- up self-organisation management would have the advantage of relatively complete information With regard to the complex SESs of grasslands, our research provides preliminary outcomes of for developing efficient use of grasslands, which specifically takes into account the highly governance system to grassland system at county-level. An interesting avenue for future variable topography of grasslands, strongly spatial and temporal heterogeneity of grassland research is to understand the specific impacts of policy interventions at the household-level, resources, traditional knowledge about grassland use and production, and other important such as the costs and benefits that herders faced during the policy interventions. Further factors (Yan et al., 2005; Wu and Du, 2008; Fernandez-Gimenez et al., 2012). research would be to explore this in more detail, for example: what changes herders adopted in order to adapt to the government-driven transformation of traditional pastoralism; how herders Theoretically, cooperative use of grasslands among neighbouring households could conform to adjusted their grassland use and livestock production patterns coping with the increasing the theory of “common pool resources” which involves self-organisation by the local population, the growing demand for livestock products and the grassland degradation; what is community in efforts to achieve a sustainable Social-Ecological System (Ostrom, 2009). the resilience of pastoral societies in the changing natural and social environment. Moreover, Nevertheless, cooperative use of grasslands and bottom-up self-organisation management our research focuses on the pastoral areas of Inner Mongolia, which is a typical region for should undergo further research concerning internal organisation, transaction costs, etc. studying the SESs of grasslands of China. However, the crop-farming areas that also include Moreover, although the self-organisation by local herders is predominantly suggested as a some permanent grassland have more complicated SESs and more serious grassland localised management model, the assistance of external authorities plays an important role in degradation than the pastoral areas, and thus even more problems to be addressed. The same balancing the various needs of the whole society. The long-term sustainability of rules devised research questions that are addressed in this study for the pastoral areas would be also at a SES level depends on monitoring and enforcement, without being overruled by government interesting for those crop farming areas, such as the dynamics of their grassland ecosystem, policies (Ostrom, 2009). As such, a combination of top-down and bottom-up approaches is livestock production and household livelihoods under the changing social, economic and perhaps the most effective approach to be considered. The study by Fernandez-Gimenez et al. political settings. Moreover, the accuracy of Chinese livestock statistics is questionable (2012) also demonstrated that the strengthening of adaptive capacity and resilience of grassland (Waldron et al., 2007). The standardisation of data collection on livestock population and meat systems requires joint efforts and improved communication between the government and local output must be improved, especially regarding the production of rural households. Given more communities. reliable data on livestock production, the scepticism over the existence and extent of

117 118

designed for crop-farming areas to pastoral areas. areas topastoral forcrop-farming designed of instead societies, of pastoral and cultural backgrounds communities. government andlocal systems betweenthe andimproved jointefforts requires communication of thatthestrengthening (2012) alsodemonstrated cons tobe effective approach perhaps themost Assuch,acombinatio policies (Ostrom, 2009). enfor monitoringand SESlevel on at a depends sustainability devised The long-term ofthewholesociety.balancing ofrules thevarious needs localised model,theassistanceof management lo by self-organisation althoughMoreover, the in concerning research should undergo further grasslands useof cooperative Nevertheless, community ineffortstoachieveasustaina involvesself which poolresources” “common the theory of Theoretically, useofgrasslands cooperative etal.,2012). and Du,2008;Fernandez-Gimenez al., 2005; Wu factors et (Yan gra about resources, traditionalknowledge grasslands, strongly variable of topography grasslands, which use of efficient for developing management th wouldhave up self-organisation create specific institutional agreements for rules, also (O confront costs transaction lower that have full autonomyat the collective-choice institutions andself-crafted self-organisation (SESs)wellable areSocial-Ecological Systems levels, atlocal resource of each governments management for makes resource itdiffic central by decision property rights top-down rights the or grasslandsuse . Clearly, of grasslandattributes between different systems grasslandunintendedobserved sy effects on were reform on differenteffectiveness had policy

but it is advocated that resource users who understand the local the understand who that resource users but it is advocated grassland use by This bottom- local communities. among neighbouring households could conformamong neighbouring households to ssland use and production, and other important use other and production, important and ssland (e.g. Ostrom, 1999; Baerlein et al., 2015). Users Baerlein al.,2015). et Ostrom, 1999; (e.g. spatial and temporal heterogeneity of grasslandof heterogeneity and temporal spatial 117 ble Social-Ecological System (Ostrom, 2009). Social-Ecological System ble and bottom-up self-organisation management self-organisation bottom-up and cement, without being overruled by government government by overruled withoutbeing cement, idered. The study by Fernandez-Gimenez et al. etal. Fernandez-Gimenez by Thestudy idered. and cropland systems before transforming the the transforming before croplandsystems and e advantage of relatively complete information strom, 2009).Assuch,it is recommended to external authorities plays an importantrolein an authoritiesplays external grassland and cropland. Especially some some Especially cropland. grassland and level tocraft someoftheirown andenforce This study has found that the same land tenure tenure the sameland foundthat This study has cal herders is predominantly suggested as a a as cal herders is predominantly suggested n of top-down and bottom-up approachesis andbottom-up oftop-down n to use the resource more efficiently through through efficiently more to usetheresource ternal organisation, transaction costs, etc. transactioncosts,etc. ternal organisation, stems, need which to the suggest clarify the adaptive capacity and resiliencegrasslandadaptive and of capacity specifically takes into account the highly highly account the into takes specifically ult toconsiderallofthespecificattributes extending the same policy measures policyextending measures thesame -organisation by the local local by the -organisation reliable thescep dataonlivestockproduction, regarding especially output mustbeimproved, (Waldron standardisation et al., 2007). of da The of Ch political settings.accuracy Moreover,the livestock productionandhouseholdlivelihoods areas, such interesting farming forthosecrop research questions that are would be addressed inalso areas this study for the pastoral and thus areas, thepastoral degradation than grassland more compl have some permanent includeareas thatalso studying the crop-farming ofChina.However, theSESsofgrasslands for typical region a which is Inner Mongolia, of areas the pastoral on focuses our research Moreover, the resiliencesocietiesinchanging andsocialenvironment. natural ofpastoral demand forlivestockproducts population, thegrowing adjusted their use and grassland livestock produ transfor government-driven tothe adapt order to research this in more to detail, explore would be such as the costs and benefits that herders policy the specific impactsof research interventions is to understand atthehousehold-level, governance atcoun system tograssland system SESs of grasslands,regardWith complex to the livestock production is a significantaspe interventions systems, ingrassland whereas need more complicated interactions between the reality intheSES. landallocation For of four of rural 2, land, wefour categories assume stak SESsneedtobeexplored moredeeply of variables anddynamics systems, thatmore and mult complex, of SESs require seriousstudy I First, follow future research. a number topointout Iwould like To conclude, future6.5 Limitationsresearch for andrecommendations

to be considered. considered. to be Moreover, chapters 3, 4 and 5 present general of the policyand 5 present outcomes 3, 4 chapters Moreover, knowledge about specific variables of variables of specific about knowledge Ostrom (2007)’sthat recommendation the components of rural land allocation thecomponentsofrural landallocation ct to be studied more concretely. more concretely. studied ct to be the interaction between grassland condition and between the interaction 118 even more problems to be addressed. The same same The addressed. moreproblemstobe even the production of rural households. Given more rural Givenmore production of households. the faced during thepolicyfaced interventions. Further as the dynamics of their grassland ecosystem, oftheir thedynamics as further research, the more components andthe themorecomponents research, further ivariable, nonlinear, cross-scale and changing and changing cross-scale ivariable, nonlinear, our research preliminary of our provides outcomes for example: what changes herders adopted herders in what changes example: for mation of traditional pastoralism; how herders howherders pastoralism; mation oftraditional ta collectiononlivestockpopulationand meat ta of limitations of this study and challenges for challenges and limitationsofthisstudy of icated SESs and more serious grassland grassland moreserious icated SESsand eholder types and three scenarios to simplify scenarios three and eholder types ty-level. An interestty-level. under the changing social, economic and social,economicand under thechanging ticism over the existence of and extent ction patterns copingwith the increasing inese livestock statistics is questionable livestockstatisticsisquestionable inese and the grassland degradation; what is grassland is degradation; andthe what SESs . More specifically, , the interactions between , the interactionsbetween the problems linked to ing avenue for future for avenue ing linked SESs in Chapter

6 overgrazing would be dissected, which would significantly contribute to the debates between References the equilibrium model and non-equilibrium model for arid and semi-arid areas. Angelsen, A. (2001). Playing games in the forest: State-local conflicts of land appropriation. Last but not least, cooperative use of grasslands on a large scale and bottom-up self-organisation Land Economics 77(2): 285-299. management are considered possible solutions for policy improvement in this research, but Angerer, J., Han, G., Fujisaki, I., and Havstad, K. (2008). Climate change and ecosystems of more empirical analysis is required to testify to their validity and investigate the possibility of Asia with emphasis on Inner Mongolia and Mongolia. Rangelands 30: 46-51. Baerlein, T., Kasymov, U., and Zikos, D. (2015). Self-governance and sustainable common implementation. There are no simple solutions to complex governance problems (Ostrom, pool resource management in Kyrgyzstan. Sustainability (Switzerland) 7(1): 496-521. 2007). As such, we call for more research to learn about the SESs of grasslands based on the doi: 10.3390/su7010496 same framework introduced by Ostrom, in order to develop broader and more integrated Ball, P., and Helbing, D. (Eds.) (2012). Why society is a complex matter: Meeting twenty- first century challenges with a new kind of science. Springer: Berlin Heidelberg insights into SESs. A large number of government-driven transformations of traditional Bao, Y., Chen, M. (1997). The studies of changes in plant diversity on degenerated steppe in pastoralism are under way in Africa, Asia and other regions. The quest to better understand the enclosed process. Journal of Inner Mongolia Universit (Acta Scientiarum Naturalium approaches and paths of these transformations is an important part of ensuring the sustainability Universitatis NeiMongol) 28: 87-91. Banks, T. (2003). Property rights reform in rangeland China: Dilemmas on the road to the of grassland SESs around the world. household ranch. World Development 31(12): 2129-2142. Banks, T., Richard, C., Li, P., and Yan, Z. (2003). Community-based grassland management in Western China: rationale, pilot project experience and policy implications. Mountain Research and Development 23: 132-140. Barker, A. J., and Selman, P. H. (1990). Managing the rural environment: An emerging role for planning authorities. Journal of Environmental Management 31(2): 185-196. Behnke, R. (1994). Natural resource management in pastoral Africa. Development Policy Review 12: 5-27. Bennett, M. T. (2008). China's sloping land conversion program: Institutional innovation or business as usual? Ecological Economics 65(4): 699-711. doi: 10.1016/j.ecolecon.2007.09.017 Boone, R. B., and Hobbs, N. T. (2004). Lines around fragments: Effects of fencing on large herbivores. African Journal of Range and Forage Science 21(3): 147-158. Boone, R. B., BurnSilver, S. B., Thornton, P. K., Worden, J. S., and Calvin, K. A. (2005). Quantifying declines in livestock due to land subdivision. Rangeland Ecology and Management 58(5): 523-532. doi: 10.2111/1551-5028(2005)58[523:qdildt]2.0.co;2 Boone, R. B., and Hobbs, N. T. (2004). Lines Around Fragments: Effects of Fencing on Large Herbivores. African Journal of Range and Forage Science 21: 147-158. Briske, D.D., Zhao, M., Han, G., Xiu, C., Kemp, D.R., Willms, W., Havstad, K., Kang, L., Wang, Z., Wu, J., Han, X., and Bai, Y. (2015). Strategies to alleviate poverty and grassland degradation in Inner Mongolia: Intensification vs production efficiency of livestock systems. Journal of Environmental Management 152: 177-182. doi: 10.1016/j.jenvman.2014.07.036 Brown, C. G., Waldron, S. A., and Longworth, J. W. (Eds.) (2008). Sustainable development in Western China: Managing people, livestock and grasslands in pastoral areas. Edward Elgar Pub: Cheltenham, United Kingdom Bureau of Animal husbandry of Inner Mongolia. (2000). History of Animal husbandry of Inner Mongolia. People's publishing house of Inner Mongolia: Hohhot. [in Chinese]

119 120 overgrazing would be dissected, which would significantly contribute to the debates between References the equilibrium model and non-equilibrium model for arid and semi-arid areas. Angelsen, A. (2001). Playing games in the forest: State-local conflicts of land appropriation. Last but not least, cooperative use of grasslands on a large scale and bottom-up self-organisation Land Economics 77(2): 285-299. management are considered possible solutions for policy improvement in this research, but Angerer, J., Han, G., Fujisaki, I., and Havstad, K. (2008). Climate change and ecosystems of more empirical analysis is required to testify to their validity and investigate the possibility of Asia with emphasis on Inner Mongolia and Mongolia. Rangelands 30: 46-51. Baerlein, T., Kasymov, U., and Zikos, D. (2015). Self-governance and sustainable common implementation. There are no simple solutions to complex governance problems (Ostrom, pool resource management in Kyrgyzstan. Sustainability (Switzerland) 7(1): 496-521. 2007). As such, we call for more research to learn about the SESs of grasslands based on the doi: 10.3390/su7010496 same framework introduced by Ostrom, in order to develop broader and more integrated Ball, P., and Helbing, D. (Eds.) (2012). Why society is a complex matter: Meeting twenty- first century challenges with a new kind of science. Springer: Berlin Heidelberg insights into SESs. A large number of government-driven transformations of traditional Bao, Y., Chen, M. (1997). The studies of changes in plant diversity on degenerated steppe in pastoralism are under way in Africa, Asia and other regions. The quest to better understand the enclosed process. Journal of Inner Mongolia Universit (Acta Scientiarum Naturalium approaches and paths of these transformations is an important part of ensuring the sustainability Universitatis NeiMongol) 28: 87-91. Banks, T. (2003). Property rights reform in rangeland China: Dilemmas on the road to the of grassland SESs around the world. household ranch. World Development 31(12): 2129-2142. Banks, T., Richard, C., Li, P., and Yan, Z. (2003). Community-based grassland management in Western China: rationale, pilot project experience and policy implications. Mountain Research and Development 23: 132-140. Barker, A. J., and Selman, P. H. (1990). Managing the rural environment: An emerging role for planning authorities. Journal of Environmental Management 31(2): 185-196. Behnke, R. (1994). Natural resource management in pastoral Africa. Development Policy Review 12: 5-27. Bennett, M. T. (2008). China's sloping land conversion program: Institutional innovation or business as usual? Ecological Economics 65(4): 699-711. doi: 10.1016/j.ecolecon.2007.09.017 Boone, R. B., and Hobbs, N. T. (2004). Lines around fragments: Effects of fencing on large herbivores. African Journal of Range and Forage Science 21(3): 147-158. Boone, R. B., BurnSilver, S. B., Thornton, P. K., Worden, J. S., and Calvin, K. A. (2005). Quantifying declines in livestock due to land subdivision. Rangeland Ecology and Management 58(5): 523-532. doi: 10.2111/1551-5028(2005)58[523:qdildt]2.0.co;2 Boone, R. B., and Hobbs, N. T. (2004). Lines Around Fragments: Effects of Fencing on Large Herbivores. African Journal of Range and Forage Science 21: 147-158. Briske, D.D., Zhao, M., Han, G., Xiu, C., Kemp, D.R., Willms, W., Havstad, K., Kang, L., Wang, Z., Wu, J., Han, X., and Bai, Y. (2015). Strategies to alleviate poverty and grassland degradation in Inner Mongolia: Intensification vs production efficiency of livestock systems. Journal of Environmental Management 152: 177-182. doi: 10.1016/j.jenvman.2014.07.036 Brown, C. G., Waldron, S. A., and Longworth, J. W. (Eds.) (2008). Sustainable development in Western China: Managing people, livestock and grasslands in pastoral areas. Edward Elgar Pub: Cheltenham, United Kingdom Bureau of Animal husbandry of Inner Mongolia. (2000). History of Animal husbandry of Inner Mongolia. People's publishing house of Inner Mongolia: Hohhot. [in Chinese]

119 120

Bureau of Animal husbandry of Inner Mongolia. (2015). Report of the progress on privatising Fan, J., Zhong, H., Harris, W., Yu, G., Wang, S., Hu, Z., and Yue, Y. (2008). Carbon storage grassland use rights of Inner Mongolia. Bureau of Animal husbandry of Inner in the grasslands of China based on field measurements of above- and below-ground Mongolia: Hohhot. [in Chinese] biomass. Climatic Change 86(3-4): 375-396. doi: 10.1007/s10584-007-9316-6 Cao, J., Yeh, E.T., Holden, N.M., Yang, Y., and Du, G. (2013a). The effects of enclosures Fan, M., Li, Y., and Li, W. (2015). Solving One Problem by Creating a Bigger one: The and land-use contracts on rangeland degradation on the Qinghai-Tibetan plateau. Consequences of Ecological Resettlement for Grassland Restoration and Poverty Journal of Arid Environments 97: 3-8. DOI: 10.1016/j.jaridenv.2013.05.002 Alleviation in Northwestern China. Land Use Policy 42: 124-130. doi: Cao, J., Yeh, E. T., Holden, N. M., Qin, Y., and Ren, Z. (2013b). The roles of overgrazing, 10.1016/j.landusepol.2014.07.011 climate change and policy as drivers of degradation of china's grasslands. Nomadic FAO. (2008). Are grasslands under threat? Brief analysis of FAO statistical data on pasture Peoples 17(2): 82-101. doi: 10.3167/np.2013.170207 and fodder crops (27-1-2016). http://www.fao.org/ag/agp/agpc/doc/grass_stats/grass- Castella, J. C., Pheng Kam, S., Dinh Quang, D., Verburg, P. H., and Thai Hoanh, C. (2007). stats.htm Combining top-down and bottom-up modelling approaches of land use/cover change FAO. (2015). Livestock and the environment. From http://www.fao.org/livestock- to support public policies: Application to sustainable management of natural resources environment/en/ in northern Vietnam. Land Use Policy 24(3): 531-545. doi: FAO. (2016). Livestock environment and development. From 10.1016/j.landusepol.2005.09.009 http://www.fao.org/agriculture/lead/lead/networks/china1/en/ China Animal Industry Yearbook. (2014). Editorial Office of China Animal Industry Feeny, D., Berkes, F., McCay, B. J., and Acheson, J. M. (1990). The Tragedy of the Yearbook. Agricultural publishing house of China: Beijing (in Chinese). Commons: Twenty-Two Years Later. Human Ecology 18: 1-19. Clements, F. E. (Eds.) (1916). Plant Succession: An Analysis of the Development of Feng, Y., Lu, Q., Tokola, T., Liu, H., and Wang, X. (2009). Assessment of Grassland Vegetation. Carnegie Institute Publications: Washington, DC Degradation in Guinan County Qinghai Province China In the Past 30 Years. Land Coase, R. H. (1960). The problem of social cost. Journal of Law and Economics. 56(4): 837- Degradation and Development 20: 55-68. doi: 10.1002/ldr.877 877. Fergusson, D. M., Swain-Campbell, N. R., and Horwood, L. J. (2002). Deviant peer Conte, T. J. (2015). The effects of China's grassland contract policy on Mongolian herders' affiliations, crime and substance use: A fixed effects regression analysis. Journal of attitudes towards grassland management in northeastern Inner Mongolia. Journal of Abnormal Child Psychology 30(4): 419-430. doi: 10.1023/a:1015774125952 Political Ecology 22(1): 79-97. Fernandez-Gimenez, M. E. (2000). The role of Mongolian nomadic pastoralists' ecological Daun-Barnett, N. (2011). Community college baccalaureate: A fixed effects, multi-year study knowledge in rangeland management. Ecological Applications 10(5): 1318-1326. doi: of the influence of state policy on nursing degree production. Higher Education Policy 10.2307/2641287 24(3): 377-398. doi: 10.1057/hep.2011.10 Fernandez-Gimenez, M. E. (2002). Spatial and social boundaries and the paradox of pastoral De Haan, C., Steinfeld, H., and Blackburn, H. (Eds.), (1996). Livestock and the Environment: land tenure: A case study from postsocialist Mongolia. Human Ecology 30(1): 49-78. Finding a Balance. Commission of the European Communities, World Bank. doi: 10.1023/a:1014562913014 http://agrienvarchive.ca/bioenergy/download/livestock_env_fao.pdf. Fernandez-Gimenez, M. E., and Le Febre, S. (2006). Mobility in pastoral systems: Dynamic DeFries, R., Hansen, A., Turner, B. L., Reid, R., and Liu, J. G. (2007). Land use change flux or downward trend? International Journal of Sustainable Development and World around protected areas: Management to balance human needs and ecological function. Ecology 13(5): 341-362. Ecological Applications 17(4): 1031-1038. doi: 10.1890/05-1111 Fernandez-Gimenez, M. E., Wang, X., Batkhishig, B., Klein, JA., and Reid, RS. (Eds.) Demsetz, H. (1967). Toward a theory of property rights. American Economic Review 62, (2012). Restoring community connections to the land: Building resilience through 347-359. community-based rangeland management in China and Mongolia. CABI Publishing: Deng, S.F., Yang, T.B., Zeng, B., Zhu, X.F., and Xu, H.J. (2013). Vegetation cover variation Wallingford, UK in the Qilian Mountains and its response to climate change in 2000-2011. Journal of Finan, F., Sadoulet, E., and De Janvry, A. (2005). Measuring the poverty reduction potential Mountain Science 10: 1050-1062. doi: 10.1007/s11629-013-2558-z of land in rural Mexico. Journal of Development Economics 77(1): 27-51. doi: Deng, X., Huang, J., Huang, Q., Rozelle, S., and Gibson, J. (2011). Do roads lead to grassland 10.1016/j.jdeveco.2004.02.004 degradation or restoration? A case study in Inner Mongolia, China. Environment and Galvin, K. A. (2009). Transitions: Pastoralists Living with Change. Annual Review of Development Economics 16(6): 751-773. doi: 10.1017/s1355770x11000180 Anthropology 38: 185-198. Dickinson, D., and Webber, M. (2007). Environmental resettlement and development, on the Gao, Q.Z., Wan, Y.F., Xu, H.M., Li, Y., Jiangcun, W.Z., and Borjigidai, A. (2010). Alpine Steppes of inner Mongolia, PRC. Journal of Development Studies 43: 537-561. doi: Grassland Degradation Index and Its Response toRecent Climate Variability in 10.1080/00220380701204513 Northern Tibet, China. Quaternary International 226: 143-150. DOI: 10.1016/j.quaint.2009.10.035

121 122

Gerlagha, R., and Keyzer, M. A. (2003). Efficiency of conservationist measures: An optimist Higgins, S.I., Kantelhardt, J., Scheiter, S., and Boerner, J. 2007. Sustainable management of viewpoint. Journal of Environmental Economics and Management 46(2): 310-333. extensively managed savanna rangelands. Ecological Economics 62: 102-114. doi: doi: 10.1016/s0095-0696(02)00037-2 10.1016/j.ecolecon.2006.05.019 Gerlagh, R., and Keyzer, M. A. (2004). Path-dependence in a Ramsey model with resource Hobbs, N. T., Galvin, K. A., Stokes, C. J., Lackett, J. M., Ash, A. J., Boone, R. B., Reid, R.S., amenities and limited regeneration. Journal of Economic Dynamics and Control and Thornton, P. K. (2008). Fragmentation of rangelands: Implications for humans, 28(6):1159-1184. doi: 10.1016/s0165-1889(03)00078-2 animals, and landscapes. Global Environmental Change 18(4): 776-785. doi: Gollier, C. (2010). Ecological discounting. Journal of Economic Theory 145(2): 812-829. 10.1016/j.gloenvcha.2008.07.011 Gong, Z., Kawamura, K., Ishikawa, N., Goto, M., Wulan, T., Alateng, D., Yin, T., and Ito, Y. Holben, B.N. (1986). Characteristics of maximum-value composite images from temporal (2015). MODIS normalized difference vegetation index (NDVI) and vegetation AVHRR data. International Journal of Remote Sensing 7: 1417-1434. doi: phenology dynamics in the Inner Mongolia grassland. Solid Earth 6: 1185-1194. doi: 10.1080/01431168608948945 10.5194/se-6-1185-2015 Homewood, K. (1995). Development, demarcation and ecological outcomes in Maasailand. Gu, Y., and Li, W. (2013). Research on the “Grazing Ban” Policy’s Effect to Grassland Africa 65(3): 331-350. doi: 10.2307/1161050 Quality: Analysis on the Scale of Herdsman Households. Acta Scientiarum Ho, P. (1996). Ownership and control in Chinese grassland management since Mao: The case Naturalium Universitatis Pekinensis 49: 288-296 (in Chinese). of free-riding in Ningxia. Pastoral Development Network Series 39. Overseas Guelke, L. (2003). The Tragedy of Privatisation: Some Environmental Consequences of the Development Institute: London Dutch Invasion of Khoikhoi South Africa. South African Geographical Journal 85: 90- Ho, P. (2000). The Clash Over State and Collective Property: The Making of The Rangeland 98. law. China Quarterly 161: 240-263. Griffin, K. (Eds.) (1995). Poverty and the Transition to a Market Economy in Mongolia. Ho, P. (2001). Rangeland degradation in North China revisited? A preliminary statistical Macmillan: St. Martin's Press. analysis to validate non-equilibrium range ecology. Journal of Development Studies Han, J.G., Zhang, Y.J., Wang, C.J., Bai, W.M., Wang, Y.R., Han, G.D., and Li, L.H. (2008). 37(3): 99-133. doi: 10.1080/00220380412331321991 Rangeland Degradation and Restoration Management in China. Rangeland Journal 30: Ho, P., and Azadi, H. (2010). Rangeland degradation in North China: Perceptions of 233-239. DOI: 10.1071/rj08009 pastoralists. Environmental Research 110(3): 302-307. Hardin, G. (1968). The Tragedy of Commons. Science 162: 1243-1248. Hobbs, N.T., Galvin, K.A., Stokes, C.J., Lackett, J.M., Ash, A.J., Boone, R.B., Reid, R.S., Harris, R. B. (2010). Rangeland degradation on the Qinghai-Tibetan plateau: A review of the and Thornton, P.K. (2008). Fragmentation of Rangelands: Implications for Humans, evidence of its magnitude and causes. Journal of Arid Environments 74(1): 1-12. doi: Animals and Landscapes. Global Environmental Change 18: 776-785. DOI: 10.1016/j.jaridenv.2009.06.014 10.1016/j.gloenvcha.2008.07.011 Hartter, J., and Ryan, S. J. (2010). Top-down or bottom-up?. Decentralization, natural Hu, W. (1997). Household land tenure reform in China: Its impact on farming land use and resource management, and usufruct rights in the forests and wetlands of western agro-environment. Land Use Policy 14(3): 175-186. Uganda. Land Use Policy 27(3): 815-826. doi: 10.1016/j.landusepol.2009.11.001 Hua, L., and Squires, V. R. (2015). Managing China's Pastoral Lands: Current Problems and He, C., Tian, J., Gao, B., and Zhao, Y. (2014). Differentiating Climate- and Human-induced Future Prospects. Land Use Policy 43: 129-137. doi: Drivers of Grassland Degradation in the Liao River Basin, China. Environmental 10.1016/j.landusepol.2014.11.004 Monitoring and Assessment 187. DOI: 10.1007/s10661-014-4199-2 Huang, L., Xiao, T., Zhao, Z., Sun, C., Liu, J., Shao, Q., Fan, J., and Wang, J. (2013). Effects He, S., and Richards, K. (2015). Impact of Meadow Degradation on Soil Water Status and of grassland restoration programs on ecosystems in arid and semiarid China. Journal Pasture Management-A Case Study in Tibet. Land Degradation and Development of Environmental Management 117: 268-275. doi: 10.1016/j.jenvman.2012.12.040 26(5): 468-479. doi: 10.1002/ldr.2358 Huang, Q., Rozelle, S., Lohmar, B., Huang, J., and Wang, J. (2006). Irrigation, Agricultural Heijman, W. J., and Mouche, P. (2013). A procedure for determining an optimal landscape Performance and Poverty Reduction in China. Food Policy 31: 30-52. and its monetary value. In CMVD Heide and WJM Heijman (Ed.), The Economic DOI:10.1016/j.foodpol.2005.06.004 Value of Landscapes: 123-129. Routledge: Oxon Hua, L., and Squires, V.R. (2015). Managing China's pastoral lands: Current problems and Henry, B., Charmley, E., Eckard, R., Gaughan, J. B., and Hegarty, R. (2012). Livestock future prospects. Land Use Policy 43: 129-137. doi: 10.1016/j.landusepol.2014.11.004 production in a changing climate: Adaptation and mitigation research in Australia. IFAD. (2009). Livestock and Pastoralists. Livestock Thematic Papers Tools for Project Crop and Pasture Science 63(3): 191-202. doi: 10.1071/cp11169 Design http://www.ifad.org/lrkm/factsheet/Pastoralists.pdf. Hinton, W. (Eds.) (1990). The Privatisation of China, the Great Reversal. Earthscan Kamara, A., Kirk, M., and Swallow, B. (2005). Property rights and land use change: Publications: London Implications for sustainable resource management in Borana, southern Ethiopia. Journal of Sustainable Agriculture 25(2): 45-61. doi: 10.1300/J064v25n02_05

123 124

Kang, L., Han, X., Zhang, Z., and Sun, O. J. (2007). Grassland ecosystems in China: Review Li, W., Ali, S.H., and Zhang, Q. (2007). Property rights and grassland degradation: a study of of current knowledge and research advancement. Philosophical Transactions of the the Xilingol pasture, Inner Mongolia, China. Journal of Environmental Management Royal Society B: Biological Sciences 362(1482): 997-1008. doi: 85: 461-470. DOI: 10.1016/j.jenvman.2006.10.010 10.1098/rstb.2007.2029 Li, W., and Huntsinger, L. (2011). China's grassland contract policy and its impacts on herder Komarek, A. M., Waldron, S. A., and Brown, C. G. (2012). An exploration of livestock- ability to benefit in Inner Mongolia: Tragic feedbacks. Ecology and Society 16(2). development policies in western china. Food Policy 37(1): 12-20. doi: Li, W., Ali, S. H., and Zhang, Q. (2007). Property rights and grassland degradation: A study 10.1016/j.foodpol.2011.10.001 of the Xilingol Pasture, Inner Mongolia, China. Journal of Environmental Koo, J. L., Koo, B. L., and Shin, Y. H. (2013). An optimal investment, consumption, leisure, Management 85(2): 461-470. doi: 10.1016/j.jenvman.2006.10.010 and voluntary retirement problem with Cobb-Douglas utility: Dynamic programming Li, W., and Zhang, Q. (Eds.) (2009). Interpretation the predicament on pastural area of China: approaches. Applied Mathematics Letters 26 (4): 481-486. doi: fosucing on the use and management issues for arid and semi-arid grassland. Beijing: 10.1016/j.aml.2012.11.012 Economic Science Press. [in Chinese] Kou, M., Jiao, J., Yin, Q., Wang, N., Wang, Z., Li, Y., Yu, W., Wei, Y., Yan, F., and Cao, B. Li, X., Yuan, Q., Wan, L., and He, F. (2008). Perspectives on livestock production systems in (2016). Successional Trajectory Over 10 Years of Vegetation Restoration of China. Rangeland Journal 30(2): 211-220. doi: 10.1071/rj08011 Abandoned Slope Croplands in the Hill-Gully Region of the Loess Plateau. Land Li, X.B., Li, R.H., Li, G.Q., Wang, H., Li, Z.F., Li, X., and Hou, X.Y. (2016). Human- Degradation and Development 27: 919-932. doi: 10.1002/ldr.2356 induced vegetation degradation and response of soil nitrogen storage in typical steppes Krusekopf, C. C. (2002). Diversity in land-tenure arrangements under the household in Inner Mongolia, China. Journal of Arid Environments 124: 80-90. DOI: responsibility system in China. China Economic Review 13(2-3): 297-312. 10.1016/j.jaridenv.2015.07.013 Lee, A.F.S. (1992). Optimal sample sizes determined by two-sample welch's t test. Li, X. L., Gao, J., Brierley, G., Qiao, Y. M., Zhang, J., and Yang, Y. W. (2013). Rangeland Communications in Statistics - Simulation and Computation 21: 689-696. doi: degradation on the Qinghai-Tibet plateau: implications for rehabilitation. Land 10.1080/03610919208813043 Degradation and Development 24(1): 72-80. doi: 10.1002/ldr.1108 Lesorogol, C.K. (2003). Transforming Institutions among Pastoralists: Inequality and Land Li, Y., Gong, B., and Li, W. (2014). A review of China’s rangeland management policies Privatization. American Anthropologist 105: 531-542. IIED Country Report. London: IIED. Lesorogol, C.K. (2008). Land Privatization and Pastoralist Well-being in Kenya. Liao, Z. (2009). The changes in price of production factors, technology development and Development and Change 39: 309-331. grassland tenure. Issues in Agricultural Economy 4: 64-69. doi: Levin, S., Xepapadeas, T., Crepin, A.-S., Norberg, J., De Zeeuw, A., Folke, C., Hughes, T., 10.13246/j.cnki.iae.2009.04.012. [in Chinese] Arrow, K., Barrett, S., Daily, G., Ehrlich, P., Kautsky, N., Maler, K.-G., Polasky, S., Lin, J. Y. (1987). The Household Responsibility System Reform in China: A peasant's Troell, M., Vincent, J. R., and Walker, B. (2013). Social-ecological systems as institutional choice. American Journal of Agricultural Economics 69(2): 410-415. doi: complex adaptive systems: modeling and policy implications. Environment and 10.2307/1242295 Development Economics 18: 111-132. doi:10.1017/s1355770x12000460 Lin, L., Li, Y.K., Xu, X.L., Zhang, F.W., Du, Y.G., Liu, S.L., Guo, X.W., and Cao, G.M. Li, J. (2012). Land tenure change and sustainable management of alpine grasslands on the (2015). Predicting parameters of degradation succession processes of Tibetan tibetan plateau: A case from hongyuan county, Sichuan province, China. Nomadic Kobresia grasslands. Solid Earth 6: 1237-1246. doi: 10.5194/se-6-1237-2015 Peoples 16(1): 36-49. Little, P. (Eds.) (1992). The elusive granary: herder, farmer and state in Northern Kenya. Li, M., Wu, J., and Deng, X. (2016). Land use change and soil carbon sequestration in China: Cambridge University Press: Cambridge, UK. Where does it pay to conserve? Regional Environmental Change 1-13. doi: Liu, C., Lu, J., and Yin, R. 2010. An estimation of the effects of china's priority forestry 10.1007/s10113-016-0948-9 programs on farmers' income. Environmental Management 45: 526-540. Li, P. J. (2009). Exponential growth, animal welfare, environmental and food safety impact: Liu, J., Zhang, Z., Xu, X., Kuang, W., Zhou, W., Zhang, S., Li, R., Yan, C., Yu, D., Wu, S., The case of China's livestock production. Journal of Agricultural and Environmental and Jiang, N. (2010). Spatial patterns and driving forces of land use change in China Ethics 22(3): 217-240. doi: 10.1007/s10806-008-9140-7 during the early 21st century. Journal of Geographical Sciences 20 (4): 483-494. doi: Li, Q.L. (2009). The grey correlative degree analysis between forage yield with temperature 10.1007/s11442-010-0483-4 and precipitation in natural grasslands. Qinghai Prataculture 18: 12-14. [in Chinese] Liu, J., and Diamond, J. (2005). China's environment in a globalizing world. Nature Li, S., Verburg, P.H., Lv, S., Wu, J., and Li, X. (2012). Spatial snalysis of the driving factors 435(7046): 1179-1186. of grassland degradation under conditions of climate change and intensive use in Inner Liu, M., Dries, L., Heijman, W., Huang, J., Zhu, X., and Deng, X. (2015). Tragedy of the Mongolia, China. Regional Environmental Change 12: 461-474. commons or tragedy of privatisation? The impact of land tenure reform on grassland condition in Inner Mongolia, China. (unpublished)

125 126

Liu, M., Dries, L., Heijman, W., Huang, J., Zhu, X., Hu, Y., and Chen, H. (2016). The impact Ministry of Agriculture of China. (2015). China national grassland monitoring report of 2014. of eco-environmental policies on the condition of grasslands in China’s pastoral area. Beijing: Retrieved from (unpublished) http://www.moa.gov.cn/zwllm/jcyj/zh/201504/t20150414_4526567.htm. access: 19 Liu, S., Yan, D.H., Shi, X.L., Wang, G., Yuan, Z., and Yin, J. (2013). Grassland NDVI May 2016 response to climate factors in different vegetation regionalizations in China. Ministry of Agriculture of China. (2016). China national grassland monitoring report of 2015. Communications in Computer and Information Science 399: 370-380. Beijing: Retrieved from http://www.dxddcx.com/news/a/201602/10027.html. access: Lopez, R. A., Shah, F. A., and Altobello, M. A. (1994). Amenity benefits and the optimal 19 May 2016 allocation of land. Land Economics 70(1): 53-62. Mullan, K., Kontoleon, A., Swanson, T. M., and Zhang, S. (2010). Evaluation of the impact Lu, X., Yan, Y., Sun, J., Zhang, X., Chen, Y., Wang, X., and Cheng, G. (2015). Short-term of the natural forest protection program on rural household livelihoods. Environmental grazing exclusion has no impact on soil properties and nutrients of degraded alpine Management 45(3): 513-525. doi: 10.1007/s00267-009-9288-6 grassland in Tibet, China. Solid Earth 6: 1195-1205. doi: 10.5194/se-6-1195-2015 Mwangi, E. (Eds.) (2007). Socioeconomic change and land use in Africa : the transformation Lv, Y., Zhang, L., Feng, X., Zeng, Y., Fu, B., Yao, X., Li, J., and Wu, B. (2015). Recent of property rights in Maasailand. Palgrave Macmillan: New York, US ecological transitions in China: Greening, browning, and influential factors. Scientific National Bureau of Statistics of China. (2009). China Statistical Yearbook 2008. Statistics Reports 5. doi: 10.1038/srep08732 Press: Beijing, China Manne, A., and Rutherford, T. F. (1994). International trade, capital flows and sectoral National People's Congress of the People's Republic of China. (2002). The Grassland Law of analysis: formulation and solution of intertemporal equilibrium models. New The People's Republic of China. Beijing (in Chinese). Directions in Computational Economics. Kluwer Academic Publishers: Dordrecht National Research Council. (1992). Grasslands and grassland sciences in northern China. Mauerhofer, V. (2011). A bottom-up 'Convention-Check' to improve top-down global National Academy Press: Washington, D.C. protected area governance. Land Use Policy 28(4): 877-886. doi: Negishi, T. (Eds.) (1972). General equilibrium theory and international trade. North-Holland 10.1016/j.landusepol.2011.03.004 Publishing Company: Amsterdam McEvoy, A. F. (1987). Toward an interactive theory of nature and culture: ecology, Niamir-Fuller, M., and Turner, M. D. (1999). A review of recent literature on pastoralism and production, and cognition in the California fishing industry. Environmental Review transhumance in Africa. In M. Niamir-Fuller (Ed.), Managing Mobility in African 11(4): 289-305. Rangelands, the Legitimization of Transhumance. Intermediate Technology Mei, H., Zhang, G., Gan, X., Ranlong, L., and Han, G. (2013). Carbon and nitrogen storage Publications: London and loss as affected by grassland degradation in Inner Mongolia, China. Journal of Nelson, R. (2006). Regulating grassland degradation in China: shallow-rooted laws? Asian- Food, Agriculture and Environment 11(3-4): 2071-2076. Pacific Law and Policy Journal 7: 385-416. Meng, L., and Gao, H. (2002). The status of grassland degradation of China and the strategies Oestreicher, J. S., Benessaiah, K., Ruiz-Jaen, M. C., Sloan, S., Turner, K., Pelletier, J., Guay, for grassland conservation. Paper presented at the 6th Chinese International Rangeland B., Clark, K. E., Roche, D. G., Meiners, M., and Potvin, C. (2009). Avoiding Congress: Beijing. deforestation in Panamanian protected areas: An analysis of protection effectiveness Meyer, N. (2006). Desertification and restoration of grasslands in Inner Mongolia. Journal of and implications for reducing emissions from deforestation and forest degradation. Forestry 104: 328-331. Global Environmental Change 19(2): 279-291. doi: 10.1016/j.gloenvcha.2009.01.003 Miao, L., Moore, J. C., Zeng, F., Lei, J., Ding, J., He, B., and Cui, X. (2015). Footprint of Ostrom, E. (Eds.) (1990). Governing the commons: the evolution of institutions for collective research in desertification management in China. Land Degradation and Development action. Cambridge University Press: Cambridge. 26: 450-457. doi: 10.1002/ldr.2399 Ostrom, E., Gardner, R., and Walker, J. (Eds.) (1994). Rules, games, and common-pool Millennium Ecosystem Assessment. (2003). Ecosystems and human well-being: a framework resources. The University of Michigan Press: Ann Arbor for assessment. Island Press: Washington, DC Ostrom, E. (1999). Coping with tragedies of the commons. Annual Review of Political Ministry of Agriculture of China. (2012). China national grassland monitoring report of 2011. Science 2: 493-535. Beijing: Retrieved from Ostrom, E. (2007). A diagnostic approach for going beyond panaceas. PNAS 104(39): 15181- http://www.agri.gov.cn/V20/SC/jjps/201204/t20120409_2598547.htm. access: 19 15187. doi: 10.1073/pnas.0702288104 May 2016 Ostrom, E. (2009). A general framework for analysing sustainability of social-ecological Ministry of Agriculture of China. (2014). China national grassland monitoring report of 2013. systems. Science 325 (5939): 419-422. doi: 10.1126/science.1172133 Beijing: Retrieved from http://www.sdcy.gov.cn/art/2014/7/31/art_5561_82075.html. Ostrom, E., and Cox, M. (2010). Moving beyond panaceas: A multi-tiered diagnostic access: 19 May 2016 approach for social-ecological analysis. Environmental Conservation 37(4): 451-463. doi: 10.1017/s0376892910000834

127 128

Ouyang, Z.Y. (2013). Ecosystem service assessment and applications for conservation Schoon, M. L. (2008). Building robustness to disturbance: Governance in southern African policies in China. Paper presented at the Sustainable Natural Resource Management in peace parks. (Ph.D Dissertation), Indiana University (3344618) Rural China –Governing markets? Nanjing, P.R. China. Sheehy, D. P., Miller, D., and Johnson, D. A. (2006). Transformation of traditional pastoral Pacione, M. (2013). Private profit, public interest and land use planning-A conflict livestock systems on the Tibetan steppe. Science et changements interpretation of residential development pressure in Glasgow's rural-urban fringe. planétaires/Sécheresse 17(1): 142-151. Land Use Policy 32: 61-77. Sneath, D. (1998). State Policy and Pasture Degradation in Inner Asia. Science 281(5380): Parras-Alcántara, L., Díaz-Jaimes, L., and Lozano-García, B. (2015). Management effects on 1147-1148. soil organic carbon stock in mediterranean open rangelands-treeless grasslands. Land Squires, V. R., Lu, X.S., Lu, Q., Wang, T., and Yang, Y.L. (Eds.) (2009). Rangeland Degradation and Development 26: 22-34. doi: 10.1002/ldr.2269 degradation and recovery in China's pastoral lands. CABI: Wallingford, UK Perman, R., Ma, Y., Common, M. S., Maddison, D., and McGilvray, J. (Eds.) (2011). Natural State Council of the People's Republic of China. (2010). The decision of establishing the resource and environmental economics (4 ed.). Pearson Education Limited: England subsidy and incentive system for grassland conservation by the State Council of the Petit, S. (2009). The dimensions of land use change in rural landscapes: Lessons learnt from People's Republic of China. The Central People's Government of the People's the GB Countryside Surveys. Journal of Environmental Management 90(9): 2851- Republic of China: Beijing. Retrieved from http://www.gov.cn/ldhd/2010- 2856. doi: 10.1016/j.jenvman.2008.05.023 10/12/content_1720555.htm. access: 19 May 2016 Piao, S., Mohammat, A., Fang, J., Cai, Q., and Feng, J. (2006). NDVI-based increase in Stanton, E. A. 2011. Negishi welfare weights in integrated assessment models: The growth of temperate grasslands and its responses to climate changes in China. Global mathematics of global inequality. Climatic Change 107(3): 417-432. Environmental Change 16(4): 340-348. doi: 10.1016/j.gloenvcha.2006.02.002 Su, Y. (2010). Research on the competitiveness of grassland animal husbandry in Inner Pollak, R. A. (1979). Bergson-samuelson social welfare functions and the theory of social Mongolia. Inner Mongolia University Press: Hohhot [in Chinese] choice. Quarterly Journal of Economics 93(1): 73-90. Sullivan, S., and Rohde, R. (2002). On non-equilibrium in arid and semi-arid grazing systems. Pritchett, L., and Woolcock, M. (2004). Solutions when the solution is the problem: Arraying Journal of Biogeography 29(12): 1595-1618. doi: 10.1046/j.1365-2699.2002.00799.x the disarray in development. World Development 32(2): 191-212. doi: Suttie, J. M., Reynolds, S. G., and Batello, C. (Eds.) (2005). Grasslands of the World. Food 10.1016/j.worlddev.2003.08.009 and Agricultural Organization of the United Nations: Rome Qasim, M., Hubacek, K., and Termansen, M. (2013). Underlying and proximate driving Tan M, and Li X. (2015). Does the green great wall effectively decrease dust storm intensity causes of land use change in district Swat, Pakistan. Land Use Policy 34: 146-157. in China? A study based on NOAA NDVI and weather station data. Land Use Policy Qu, F., Kuyvenhoven, A., Shi, X., and Heerink, N. (2011). Sustainable natural resource use in 43: 42-47. doi: 10.1016/j.landusepol.2014.10.017 rural China: recent trends and policies. China Economic Review 22(4): 444-460. doi: Taylor, J. L. (2006). Negotiating the grassland: The policy of pasture enclosures and 10.1016/j.chieco.2010.08.005 contested resource use in Inner Mongolia. Human Organization 65(4): 374-386. Rambonilaza, M., and Dachary-Bernard, J. (2007). Land-use planning and public preferences: Tessema, W. K., Ingenbleek, P. T. M., and Van Trijp, H. C. M. (2014). Pastoralism, What can we learn from choice experiment method?” Landscape and Urban Planning sustainability, and marketing. A review. Agronomy for Sustainable Development 83(4): 318-326. doi: 10.1016/j.landurbplan.2007.05.013 34(1): 75-92. doi: 10.1007/s13593-013-0167-4 Ren, H., Shen, W. J., Lu, H. F., Wen, X. Y., and Jian, S. G. (2007). Degraded ecosystems in Tilt, B. (2008). Smallholders and the 'household responsibility system': Adapting to China: Status, causes, and restoration efforts. Landscape and Ecological Engineering institutional change in Chinese agriculture. Human Ecology 36(2): 189-199. doi: 3(1): 1-13. doi: 10.1007/s11355-006-0018-4 10.1007/s10745-007-9127-4 Richard, C., Yan, Z., and Du, G. (2006). The Paradox of the individual household Train, K., and McFadden, D. (1978). The goods/leisure trade-off and disaggregate work trip responsibility system in the grasslands of the Tibetan plateau, China. Paper presented mode choice models. Transportation Research 12(5): 349 -353. at the Rangelands of Central Asia: Proceedings of the Conference on Transformations, Turner Ii, B. L., Moss, R. H., and Skole, D. L. (1993). Relating land use and global land- Issues, and Future Challenges: Salt Lake City, UT. cover change. A proposal for an IGBP-HDP core project. Saaty, T. (Eds.) (1980). The analytic hierarchy process. McGraw-Hill: New York Turner Ii, B. L., Meyer, W. B., and Skole, D. L. (1994). Global land-use/land-cover change: Saaty, T. (1990). How to make a decision: The analytic hierarchy process. European Journal towards an integrated study. AMBIO 23(1): 91-95. of Operational Research 48(1): 9-26. Turner, M. D. (1999). The role of social networks, indefinite boundaries and political Senay, G.B., and Elliott, R.L. (2000). Combining AVHRR-NDVI and landuse data to bargaining in maintaining the ecological and economic resilience of the transhumance describe temporal and spatial dynamics of vegetation. Forest Ecology and systems of sudano-sahelian west africa. In M. Niamir-Fuller (Ed.), Managing mobility Management 128: 83-91. doi: 10.1016/s0378-1127(99)00275-3 in african rangelands: the legitimization of transhumance. Intermediate Technology Publications Ltd, London.

129 130

Unkovich, M., and Nan, Z. (2008). Problems and prospects of grassland agroecosystems in Wu, J., Zhang, Q., Li, A., and Liang, C. (2015). Historical landscape dynamics of Inner western China. Agriculture, Ecosystems and Environment 124(1-2): 1-2. doi: Mongolia: patterns, drivers, and impacts. Landscape Ecology 30(9): 1579-1598. doi: 10.1016/j.agee.2007.08.012 10.1007/s10980-015-0209-1 Urwin, K., and Jordan, A. (2008). Does public policy support or undermine climate change Wu, Z., and Du, W. (2008). Pastoral nomad rights in Inner Mongolia. Nomadic Peoples adaptation? Exploring policy interplay across different scales of governance. Global 12(2): 13-33. doi: 10.3167/np.2008.120202 Environmental Change 18(1): 180-191. doi: 10.1016/j.gloenvcha.2007.08.002 Wu, Z.Y. (Eds.) (1980). China's vegetation. Science press: Beijing, China Vaidya, O. S., and Kumar, S. (2006). Analytic hierarchy process: An overview of Yan, J., Wu, Y., and Zhang, Y. (2011). Adaptation strategies to pasture degradation: Gap applications. European Journal of Operational Research 169(1): 1-29. between government and local nomads in the eastern Tibetan Plateau. Journal of Vargas, L. G. (1990). An overview of the analytic hierarchy process and its applications. Geographical Sciences 21(6): 1112-1122. doi: 10.1007/s11442-011-0904-z European Journal of Operational Research 48(1): 2-8. doi: 10.1016/0377- Yan, Z., Wu, N., Dorji, Y., and Ru, J. (2005). A review of rangeland privatisation and its 2217(90)90056-h implications in the Tibetan Plateau, China. Nomadic Peoples 9(1-2): 31-51. Vasconcelos, V. V., Santos, F. C., and Pacheco, J. M. (2013). A bottom-up institutional Yang, D., Zhang, Y., Wu, Y., Wang, D., Bao, C., (2004). The problems and sustainable approach to cooperative governance of risky commons. Nature Climate Change 3(9): development of pratacultural industry in Hulunbeier. Grassland and Turf (Quarterly) 797-801. doi: 10.1038/nclimate1927 1, 14-16. Verbeek, M. (Eds.) (2012). Models Based on Panel Data: A guide to modern econometrics (4 Yang, Y. F., Niu, P., and Zhu, L. Q. (2012). Study on the evaluation of rural land-use conflict ed.). John Wiley and Sons Ltd: United Kingdom. intensity. Journal of Food, Agriculture and Environment 10(3-4): 1479-1482. Verburg, P. H., Eickhout, B., and Van Meijl, H. (2008). A multi-scale, multi-model approach Yang, L. (2007). Grassland management: how to improve the implementation of Household for analyzing the future dynamics of European land use. Annals of Regional Science Production Responsibility System in the pastoral areas of China. Chinese Rural 42(1): 57-77. doi: 10.1007/s00168-007-0136-4 Economy: 62-67. [in Chinese] Vetter, S. (2005). Rangelands at Equilibrium and Non-equilibrium: Recent developments in Yang, L., Wylie, B.K., Tieszen, L.L., and Reed, B.C. (1998). An analysis of relationships the debate. Journal of Arid Environments 62: 321-341. DOI: among climate forcing and time-integrated NDVI of grasslands over the U.S. northern 10.1016/j.jaridenv.2004.11.015 and central Great Plains. Remote Sensing of Environment 65: 25-37. doi: Waldron, S., Brown, C., Longworth, J., and Cungen, Z. (Eds.) (2007). China's livestock 10.1016/s0034-4257(98)00012-1 revolution: Agribusiness and policy developments in the sheep meat industry. Yang, X.H., Zhang, K.B., and Hou, R.P. (2005). Impacts of exclusion on vegetative features Athenaeum Press: Gateshead, UK and aboveground biomass in semi-arid degraded rangeland. Ecology and Environment Waldron, S., Brown, C., and Longworth, J. (2010). Grassland degradation and livelihoods in 14: 730-734. China's western pastoral region: A framework for understanding and refining China's Yang, Z., and Nordhaus, W. D. (2006). Magnitude and direction of technological transfers for recent policy responses. China Agricultural Economic Review 2(3): 298-320. doi: mitigating GHG emissions. Energy Economics 28(5-6): 730-741. doi: 10.1108/17561371011078435 10.1016/j.eneco.2006.05.015 Wang, J., Brown, D. G., and Agrawal, A. (2013). Climate adaptation, local institutions, and Ybarra, M. (2009). Violent visions of an ownership society: The land administration project rural livelihoods: A comparative study of herder communities in Mongolia and Inner in Petén, Guatemala. Land Use Policy 26(1): 44-54. Mongolia, China. Global Environmental Change 23(6): 1673-1683. doi: Yeh, ET. (2005). Green governmentality and pastoralism in Western China: 'Converting 10.1016/j.gloenvcha.2013.08.014 pastures to grasslands'. Nomadic Peoples 9: 9-29. Wang, X. (2009). From ‘rangeland leasing’ to ‘recentralization in rangeland conservation’— Yuan, J., Ouyang, Z., Zheng, H., and Xu, W. (2011). The research on the ecosystem —policies of rangeland conservation in North China. China Rural Survey 27(1): 36- management in the eco-fragile regions of China. China Population, Resources and 46. Environment 21(3): 97-99. Williams, K. J. H., and Schirmer, J. (2012). Understanding the relationship between social Yu, L., and Farrell, K. N. (2013). Individualized pastureland use: responses of herders to change and its impacts: The experience of rural land use change in south-eastern institutional arrangements in pastoral China. Human Ecology 41(5): 759-771. doi: Australia. Journal of Rural Studies 28(4): 538-548. doi: 10.1007/s10745-013-9580-1 10.1016/j.jrurstud.2012.05.002 Zhan, J. Y., Deng, X. Z., Jiang, Q., and Shi, N. (2007). The application of system dynamics Wu, D., Gao, P., and Dong, J. (2012). Impact of subsidy on low-rent housing lessees’ welfare and CLUE-S model in land use change dynamic simulation: a case study in Taips in China. International Journal of Information Technology and Decision Making County, inner Mongolia of China. Proceedings Paper. Conference on Systems 11(3): 643-660. Science, Management Science and System Dynamics.

131 132

ecosystem change and induce desertification in Otindag sandy land, China over the ௖௨௟ሺ௜ሻǡ݈௥௔௡ሺ௜ሻǡ݈௙௢௥ሺ௜ሻǡ݈௢௧௛ሺ௜ሻ ൒ͲǢͲ൑ܽ௜ǡܾ௜ǡܿ௜ǡ݀௜ ൑ ͳǢܽ௜ ൅ܾ௜ ൅ܿ௜ ൅݀௜ ൌ ͳ݈݁ݎpast 40 years?” Journal of Arid Environments 64(3): 523-541. ݓ݄݁ Zhou, G., Wang, Y., and Wang, S. (2002). Responses of grassland ecosystems to precipitation The logarithmic transformation of the utility function gives: and land use along the Northeast China Transect. Journal of Vegetation Science 13: 361-368. Zhu, X. (2004). Environmental-economic modelling of novel protein foods: A general Ǥ ͵ሻܣʹݑ௜ ൌ ݈ܷ݊௜ ൌܽ௜݈݈݊௖௨௟ሺ௜ሻ ൅ܾ௜݈݈݊௥௔௡ሺ௜ሻ ൅ܿ௜݈݈݊௙௢௥ሺ௜ሻ ൅݀௜݈݈݊௢௧௛ሺ௜ሻǤሺ equilibrium approach. Ph.D Thesis. Social Sciences, Wageningen University, The Lagrange function is then: Wageningen. Zhu, X., and van Ierland, E. (2006). The enlargement of the European Union: Effects on trade and emissions of greenhouse gases. Ecological Economics 57(1): 1-14. doi: ௜ ௜ ௖௨௟ሺ௜ሻ ௜ ௥௔௡ሺ௜ሻ ௜ ௙௢௥ሺ௜ሻ ௜ ௢௧௛ሺ௜ሻ ݑ ൌ݈݈ܽ݊ ൅ܾ݈݈݊ ൅݈݈ܿ݊ ൅݈݈݀݊ܮ j.ecolecon.2005.03.030/10.1016 ൅ߣ൫ܮെ݈௖௥௢ሺ௜ሻ െ݈௥௔௡ሺ௜ሻ െ݈௙௢௥ሺ௜ሻ െ݈௢௧௛ሺ௜ሻ൯ሺʹܣǤ Ͷሻ Taking the partial derivations of with respect to each variable, and setting the first order

condition gives: ୧

ݑ௜ ܽ௜ܮ߲ ൌ െ ߣ ൌ ͲǡሺʹܣǤ ͷሻ ߲݈௖௨௟ሺ௜ሻ ݈௙௔௥ሺ௜ሻ

ݑ௜ ܾ௜ܮ߲ ൌ െ ߣ ൌ ͲǡሺʹܣǤ ͸ሻ ߲݈௥௔௡ሺ௜ሻ ݈௥௔௡ሺ௜ሻ

ݑ௜ ܿ௜ܮ߲ ൌ െ ߣ ൌ ͲǡሺʹܣǤ ͹ሻ ߲݈௙௢௥ሺ௜ሻ ݈௙௢௥ሺ௜ሻ

ݑ௜ ݀௜ܮ߲ ൌ െ ߣ ൌ ͲǤሺʹܣǤ ͺሻ ߲݈௢௧௛ሺ௜ሻ ݈௢௧௛ሺ௜ሻ 133 134

Zhang, H., Buongiorno, J., and Zhu, S. (2012a). Domestic and foreign consequences of Appendixes China's land tenure reform on collective forests. International Forestry Review 14(3): 349-362. doi: 10.1505/146554812802646648 Appendix 2-A: Derivation of the individually optimal allocation of land Zhang, K., Yu, Z., Li, X., Zhou ,W., and Zhang, D. (2007). Land use change and land degradation in China from 1991 to 2001. Land Degradation and Development 18: For the individual optimal allocation of land, each stakeholder type i 209-219. DOI: 10.1002/ldr.757 representing four stakeholder types, including the herders, the farmers, the ሺecological ൌ ͳǡ ʹ͵Ͷ public Zhang, L., Pan, J., and Chen, J. (2012). The optimal scale of livestock production in differnt authorities and economic public authorities) maximizes its utility function by choosing the types of rangelands. Issues in Agricultural Economy 4: 90-97. [in Chinese] Zhang, J., Zhang, L., Liu, W., Qi, Y., and Wo, X. (2014). Livestock-carrying capacity and amounts of land for different purposes. That is: overgrazing status of alpine grassland in the Three-River Headwaters region, China. Journal of Geographical Sciences 24: 303-312. doi: 10.1007/s11442-014-1089-z Zhang, W.N., Ganjurjav, H., Liang, Y., Gao, Q.Z., Wan, Y.F., Li, Y., Baima, Y.Z., and Xirao, ௔೔ ௕೔ ௖೔ ௗ೔ ܯܣܷܺ௜ ൌ݈௖௨௟ሺ௜ሻ݈௥௔௡ሺ௜ሻ݈௙௢௥ሺ௜ሻ݈௢௧௛ሺ௜ሻǡሺʹܣǤ ͳሻ Z.M. (2015). Effect of a grazing ban on restoring the degraded alpine meadows of Subject to: Northern Tibet, China. Rangeland Journal 37: 89-95. doi: 10.1071/rj14092 Zhao, J., and Guo, J. (2009). Study on statistic models of drought prediction for grassland in Inner Mongolia. Pratacultural Science 26: 14-19. Zheng, Y. R., Xie, Z. X., Robert, C., Jiang, L. H., and Shimizu, H. (2006). Did climate drive ݈௖௨௟ሺ௜ሻ ൅݈௥௔௡ሺ௜ሻ ൅݈௙௢௥ሺ௜ሻ ൅݈௢௧௛ሺ௜ሻ ൌ ܮǡሺʹܣǤ ʹሻ ecosystem change and induce desertification in Otindag sandy land, China over the ௖௨௟ሺ௜ሻǡ݈௥௔௡ሺ௜ሻǡ݈௙௢௥ሺ௜ሻǡ݈௢௧௛ሺ௜ሻ ൒ͲǢͲ൑ܽ௜ǡܾ௜ǡܿ௜ǡ݀௜ ൑ ͳǢܽ௜ ൅ܾ௜ ൅ܿ௜ ൅݀௜ ൌ ͳ݈݁ݎpast 40 years?” Journal of Arid Environments 64(3): 523-541. ݓ݄݁ Zhou, G., Wang, Y., and Wang, S. (2002). Responses of grassland ecosystems to precipitation The logarithmic transformation of the utility function gives: and land use along the Northeast China Transect. Journal of Vegetation Science 13: 361-368. Zhu, X. (2004). Environmental-economic modelling of novel protein foods: A general Ǥ ͵ሻܣʹݑ௜ ൌ ݈ܷ݊௜ ൌܽ௜݈݈݊௖௨௟ሺ௜ሻ ൅ܾ௜݈݈݊௥௔௡ሺ௜ሻ ൅ܿ௜݈݈݊௙௢௥ሺ௜ሻ ൅݀௜݈݈݊௢௧௛ሺ௜ሻǤሺ equilibrium approach. Ph.D Thesis. Social Sciences, Wageningen University, The Lagrange function is then: Wageningen. Zhu, X., and van Ierland, E. (2006). The enlargement of the European Union: Effects on trade and emissions of greenhouse gases. Ecological Economics 57(1): 1-14. doi: ௜ ௜ ௖௨௟ሺ௜ሻ ௜ ௥௔௡ሺ௜ሻ ௜ ௙௢௥ሺ௜ሻ ௜ ௢௧௛ሺ௜ሻ ݑ ൌ݈݈ܽ݊ ൅ܾ݈݈݊ ൅݈݈ܿ݊ ൅݈݈݀݊ܮ j.ecolecon.2005.03.030/10.1016 ൅ߣ൫ܮെ݈௖௥௢ሺ௜ሻ െ݈௥௔௡ሺ௜ሻ െ݈௙௢௥ሺ௜ሻ െ݈௢௧௛ሺ௜ሻ൯ሺʹܣǤ Ͷሻ Taking the partial derivations of with respect to each variable, and setting the first order

condition gives: ୧

ݑ௜ ܽ௜ܮ߲ ൌ െ ߣ ൌ ͲǡሺʹܣǤ ͷሻ ߲݈௖௨௟ሺ௜ሻ ݈௙௔௥ሺ௜ሻ

ݑ௜ ܾ௜ܮ߲ ൌ െ ߣ ൌ ͲǡሺʹܣǤ ͸ሻ ߲݈௥௔௡ሺ௜ሻ ݈௥௔௡ሺ௜ሻ

ݑ௜ ܿ௜ܮ߲ ൌ െ ߣ ൌ ͲǡሺʹܣǤ ͹ሻ ߲݈௙௢௥ሺ௜ሻ ݈௙௢௥ሺ௜ሻ

ݑ௜ ݀௜ܮ߲ ൌ െ ߣ ൌ ͲǤሺʹܣǤ ͺሻ ߲݈௢௧௛ሺ௜ሻ ݈௢௧௛ሺ௜ሻ 133 134

Combining (2A.5) to (2A.8) gives: Appendix 2-B: Derivation of the socially optimal allocation of land

For the social optimal allocation of land, maximizing the social welfare is:

ܽ௜ ܾ௜ ܿ௜ ݀௜ ൌ ൌ ൌ ǡሺʹܣǤ ͻሻ MAX ݈௖௨௟ሺ௜ሻ ݈௥௔௡ሺ௜ሻ ݈௙௢௥ሺ௜ሻ ݈௢௧௛ሺ௜ሻ Or : ସ ఉ೔ ܹൌς௜ୀଵ ܷ௜ ǡሺʹܤǤ ͳሻ Subject to:

ܽ௜݈௥௔௡ሺ௜ሻ ൌܾ௜݈௖௨௟ሺ௜ሻǡܽ௜݈௙௢௥ሺ௜ሻ ൌܿ௜݈௖௨௟ሺ௜ሻǡܽ௜݈௢௧௛ሺ௜ሻ ൌ݀௜݈௖௨௟ሺ௜ሻǤሺʹܣǤ ͳͲሻ Combining (A2) with (A9) ݈௖௨௟ሺ௦ሻ ൅݈௥௔௡ሺ௦ሻ ൅݈௙௢௥ሺ௦ሻ ൅݈௢௧௛ሺ௦ሻ ൌ ܮǡሺʹܤǤ ʹሻ ǡ ݓ݄݁ܽݒ݁ǣ ସ ସ ܽ௜ ܽ௜ ௜ ௜ ௜ ௜ ௜ ௜ ௖௨௟௦ ௥௔௡௦ ௙௢௥௦ ௢௧௛௦ ௜ ෍ ߚ ൌͳǡǤߚ ൒ Ͳǡ Ͳ ൑ ܽ ǡܾǡܿǡ݀ ൑ ͳǢ݈ ǡ݈ ǡ݈ ǡ݈ ൒ ͲǢ෍ ߚ ݁ݎǤ ͳͳሻ ݓ݄݁ܣʹൌ ǡሺ ݈௖௨௟ሺ௜ሻ ܮെ݈௥௔௡ሺ௜ሻ െ݈௙௢௥ሺ௜ሻ െ݈௢௧௛ሺ௜ሻ ௜ୀଵ ௜ୀଵ Rearrange (A10) and (A11) combined with ൌ ͳܽ݊݀ߚ௜ ൒Ͳ Plugging equation (2A.1) into (2B.1), we obtain: ௜ ൅ܾ௜ ൅ܿ௜ ൅݀௜ ൌ ͳǡ ݓ݁݋ܾݐܽ݅݊ǣܽ

଴ ݈௖௨௟ሺ௜ሻ ൌܽ௜ ൈ ܮǡሺʹܣǤ ͳʹሻ ఉభ ఉమ ௔భ ௕భ ௖భ ௗభ ௔మ ௕మ ௖మ ௗమ ቀ݈௖௨௟ሺ௦ሻ݈௥௔௡ሺ௦ሻ݈௙௢௥ሺ௦ሻ݈௢௧௛ሺ௦ሻቁ כ ଴ ܹ ൌ ቀ݈௖௨௟ሺ௦ሻ݈௥௔௡ሺ௦ሻ݈௙௢௥ሺ௦ሻ݈௢௧௛ሺ௦ሻቁ ௥௔௡ሺ௜ሻ ௜ ݈ ൌܾ ൈ ܮǡሺʹܣǤ ͳ͵ሻ ఉయ ௔య ௕య ௖య ௗయ ቀ݈௖௨௟ሺ௦ሻ݈௥௔௡ሺ௦ሻ݈௙௢௥ሺ௦ሻ݈௢௧௛ሺ௦ሻቁ כ ଴ ௙௢௥ሺ௜ሻ ௜ ఉర ݈ ൌܿ ൈ ܮǡሺʹܣǤ ͳͶሻ ௔ర ௕ర ௖ర ௗర

Ǥ ͵ሻܤʹቀ݈௖௨௟ሺ௦ሻ݈௥௔௡ሺ௦ሻ݈௙௢௥ሺ௦ሻ݈௢௧௛ሺ௦ሻቁ ሺכ ଴ The logarithmic transformation of the social welfare function is: ݈௢௧௛ሺ௜ሻ ൌ ݀௜ ൈ ܮǡሺʹܣǤ ͳͷሻ This completes the proof of equation (3) to (6) in the main text of Chapter 2.

௔భ ௕భ ௖భ ௗభ ௔మ ௕మ ௖మ ݓ ൌ ݈ܹ݊ ൌ ߚଵ݈݊ ቀ݈௖௨௟ሺ௦ሻ݈௥௔௡ሺ௦ሻ݈௙௢௥ሺ௦ሻ݈௢௧௛ሺ௦ሻቁ ൅ ߚଶ݈݊ ቀ݈௖௨௟ሺ௦ሻ݈௥௔௡ሺ௦ሻ݈௙௢௥ሺ௦ሻሻቁ ௔య ௕య ௖య ௗయ ଷ ൅ ߚ ݈݊ ቀ݈௖௨௟ሺ௦ሻ݈௥௔௡ሺ௦ሻ݈௙௢௥ሺ௦ሻ݈௢௧௛ሺ௦ሻቁ ௔ర ௕ర ௖ర ௗర ൅ ߚସ݈݊ ቀ݈௖௨௟ሺ௦ሻ݈௥௔௡ሺ௦ሻ݈௙௢௥ሺ௦ሻ݈௢௧௛ሺ௦ሻቁ ሺʹܤǤ Ͷሻ The Lagrange function is then:

௔భ ௕భ ௖భ ௗభ ௔మ ௕మ ௖మ ௗమ ݓ ൌ ߚଵ݈݊ ቀ݈௖௨௟ሺ௦ሻ݈௚௥௔ሺ௦ሻ݈௙௢௥ሺ௦ሻ݈௢௧௛ሺ௦ሻቁ ൅ ߚଶ݈݊ ቀ݈௖௨௟ሺ௦ሻ݈௥௔௡ሺ௦ሻ݈௙௢௥ሺ௦ሻ݈௢௧௛ሺ௦ሻቁܮ ௔య ௕య ௖య ௗయ ௔ర ௕ర ௖ర ௗర ൅ ߚଷ݈݊ ቀ݈௖௨௟ሺ௦ሻ݈௥௔௡ሺ௦ሻ݈௙௢௥ሺ௦ሻ݈௢௧௛ሺ௦ሻቁ൅ ߚସ݈݊ ቀ݈௖௨௟ሺ௦ሻ݈௥௔௡ሺ௦ሻ ݈௙௢௥ሺ௦ሻ݈௢௧௛ሺ௦ሻቁ

൅ߣ൫ܮെ݈௖௨௟ሺ௦ሻ െ݈௥௔௡ሺ௦ሻ െ݈௙௢௥ሺ௦ሻ െ݈௢௧௛ሺ௦ሻ൯ሺʹܤǤ ͷሻ Taking the partial derivations of Lw with respect to each variable, and setting the first order condition gives:

135 136

Combining (2A.5) to (2A.8) gives: Appendix 2-B: Derivation of the socially optimal allocation of land

For the social optimal allocation of land, maximizing the social welfare is:

135 136

ସ ସ σ௜ୀଵ ߚ௜ܾ௜ כ ݓ σ௜ୀଵ ߚ௜ܽ௜ܮ߲ ൌ െ ߣ ൌ ͲǡሺʹܤǤ ͸ሻ ݈௥௔௡ሺ௦ሻ ൌ ସ ସ ସ ସ ܮǡሺʹܤǤ ͳ͸ሻ ߲݈௖௨௟ሺ௦ሻ ݈௖௨௟ሺ௦ሻ σ௜ୀଵ ߚ௜ܽ௜ ൅ σ௜ୀଵ ߚ௜ܾ௜ ൅ σ௜ୀଵ ߚ௜ܿ௜ ൅ σ௜ୀଵ ߚ௜݀௜

ସ ସ σ௜ୀଵ ߚ௜ܿ௜ כ ݓ σ௜ୀଵ ߚ௜ܾ௜ܮ߲ ൌ െ ߣ ൌ ͲǡሺʹܤǤ ͹ሻ ݈௙௢௥ሺ௦ሻ ൌ ସ ସ ସ ସ ܮǡሺʹܤǤ ͳ͹ሻ ߲݈௥௔௡ሺ௦ሻ ݈௥௔௡ሺ௦ሻ σ௜ୀଵ ߚ௜ܽ௜ ൅ σ௜ୀଵ ߚ௜ܾ௜ ൅ σ௜ୀଵ ߚ௜ܿ௜ ൅ σ௜ୀଵ ߚ௜݀௜

ସ ସ ௜ ௜ ݀ σ௜ୀଵ ߚ כ ݓ σ௜ୀଵ ߚ௜ܿ௜ܮ߲ ൌ െ ߣ ൌ ͲǡሺʹܤǤ ͺሻ ݈௢௧௛ሺ௦ሻ ൌ ସ ସ ସ ସ ܮǡሺʹܤǤ ͳͺሻ ߲݈௙௢௥ሺ௦ሻ ݈௙௢௥ሺ௦ሻ σ௜ୀଵ ߚ௜ܽ௜ ൅ σ௜ୀଵ ߚ௜ܾ௜ ൅ σ௜ୀଵ ߚ௜ܿ௜ ൅ σ௜ୀଵ ߚ௜݀௜ Since ସ ସ ସ ସ ସ ௜ୀଵ ௜ ௜ ௜ୀଵ ௜ ௜ ௜ୀଵ ௜ ௜ ௜ୀଵ ௜ ௜ ݓ σ௜ୀଵ ߚ௜݀௜ We have:σ ߚ ܽ ൅ σ ߚ ܾ ൅ σ ߚ ܿ ൅ σ ߚ ݀ ൌͳܮ߲ ൌ െ ߣ ൌ ͲǡሺʹܤǤ ͻሻ ߲݈௢௧௛ሺ௦ሻ ݈௢௧௛ሺ௦ሻ Combining (B6) to (B9) gives:

ସ כ ௖௨௟ሺ௦ሻ ௜ ௜ ସ ସ ସ ସ ݈ ൌܮൈ෍ߚܽ ሺʹܤǤ ͳͻሻ σ௜ୀଵ ߚ௜ܽ௜ σ௜ୀଵ ߚ௜ܾ௜ σ௜ୀଵ ߚ௜ܿ௜ σ௜ୀଵ ߚ௜݀௜ ௜ୀଵ ൌ ൌ ൌ ǡሺʹܤǤ ͳͲሻ ݈௖௨௟ሺ௦ሻ ݈௥௔௡ሺ௦ሻ ݈௙௢௥ሺ௦ሻ ݈௢௧௛ሺ௦ሻ Or: ସ כ ݈௥௔௡ሺ௦ሻ ൌܮൈ෍ߚ௜ܾ௜ ሺʹܤǤ ʹͲሻ ௜ୀଵ

ସ ସ ସ כ ௜ ௜ ௥௔௡ሺ௦ሻ ௜ ௜ ௙௔௥ሺ௦ሻ ෍ߚܽ ݈ ൌ෍ߚܾ ݈ ǡሺʹܤǤ ͳͳሻ ௙௢௥ሺ௦ሻ ௜ ௜ ௜ୀଵ ௜ୀଵ ݈ ൌܮൈ෍ߚܿ ሺʹܤǤ ʹͳሻ ௜ୀଵ

ସ ସ ସ כ ௜ ௜ ௙௢௥ሺ௦ሻ ௜ ௜ ௙௔௥ሺ௦ሻ ෍ߚܽ ݈ ൌ෍ߚܿ ݈ ǡሺʹܤǤ ͳʹሻ ௢௧௛ሺ௦ሻ ௜ ௜ ௜ୀଵ ௜ୀଵ ݈ ൌܮൈ෍ߚ݀ ሺʹܤǤ ʹʹሻ ௜ୀଵ This completes the proof of equation (9) to (12) in the main text of Chapter 2. ସ ସ

෍ ߚ௜ܽ௜ ݈௢௧௛ሺ௦ሻ ൌ෍ߚ௜݀௜ ݈௙௔௥ሺ௦ሻǤሺʹܤǤ ͳ͵ሻ ௜ୀଵ ௜ୀଵ Combining (B2) with (B10)

ǡ ݓ݄݁ܽݒ݁ǣ

ସ ସ σ௜ୀଵ ߚ௜ܽ௜ σ௜ୀଵ ߚ௜ܽ௜ ൌ ǡሺʹܤǤ ͳͶሻ ݈௖௨௟ሺ௦ሻ ܮെ݈௥௔௡ሺ௦ሻ െ݈௙௢௥ሺ௦ሻ െ݈௢௧௛ሺ௦ሻ Plug

ͳͶሻǡ ݓ݁݋ܾݐܽ݅݊ǣܤͳ͵ሻ݅݊ݐ݋ሺܤͳʹሻܽ݊݀ሺܤͳͳሻሺܤሺ

ସ σ௜ୀଵ ߚ௜ܽ௜ כ ݈௖௨௟ሺ௦ሻ ൌ ସ ସ ସ ସ ܮǡሺʹܤǤ ͳͷሻ σ௜ୀଵ ߚ௜ܽ௜ ൅ σ௜ୀଵ ߚ௜ܾ௜ ൅ σ௜ୀଵ ߚ௜ܿ௜ ൅ σ௜ୀଵ ߚ௜݀௜

137 138

෍ ߚ௜ܽ௜ ݈௢௧௛ሺ௦ሻ ൌ෍ߚ௜݀௜ ݈௙௔௥ሺ௦ሻǤሺʹܤǤ ͳ͵ሻ ௜ୀଵ ௜ୀଵ Combining (B2) with (B10)

ǡ ݓ݄݁ܽݒ݁ǣ

ସ ସ σ௜ୀଵ ߚ௜ܽ௜ σ௜ୀଵ ߚ௜ܽ௜ ൌ ǡሺʹܤǤ ͳͶሻ ݈௖௨௟ሺ௦ሻ ܮെ݈௥௔௡ሺ௦ሻ െ݈௙௢௥ሺ௦ሻ െ݈௢௧௛ሺ௦ሻ Plug

ͳͶሻǡ ݓ݁݋ܾݐܽ݅݊ǣܤͳ͵ሻ݅݊ݐ݋ሺܤͳʹሻܽ݊݀ሺܤͳͳሻሺܤሺ

ସ σ௜ୀଵ ߚ௜ܽ௜ כ ݈௖௨௟ሺ௦ሻ ൌ ସ ସ ସ ସ ܮǡሺʹܤǤ ͳͷሻ σ௜ୀଵ ߚ௜ܽ௜ ൅ σ௜ୀଵ ߚ௜ܾ௜ ൅ σ௜ୀଵ ߚ௜ܿ௜ ൅ σ௜ୀଵ ߚ௜݀௜

137 138

Appendix 2-C: Questionnaires and results of Analytic Hierarchy Process 8. Comparing the forest land and other rural land, which one concerning the economic benefit is more beneficial to you, and how many benefits? Appendix 2-C1: Questionnaires of Analytic Hierarchy Process Forest land Other rural land 9 8 7 6 5 4 3 2 1 2 3 4 5 6 7 8 9 1. Which is your stakeholder type in rural land allocation? a. Herder 9. Comparing the cultivated land and grassland, which one concerning the ecological b. Farmer benefit is more beneficial to you, and how many benefits? c. Economic authorities Cultivated land Grassland d. Ecological authorities 9 8 7 6 5 4 3 2 1 2 3 4 5 6 7 8 9

2. Comparing the economic benefit and ecological benefit of rural land, which one 10. Comparing the cultivated land and forest land, which one concerning the ecological concerning the efficient allocation of rural land is more important for you, and how benefit is more beneficial to you, and how many benefits? much important? Cultivated land Forest land Economic benefit Ecological benefit 9 8 7 6 5 4 3 2 1 2 3 4 5 6 7 8 9 9 8 7 6 5 4 3 2 1 2 3 4 5 6 7 8 9 11. Comparing the cultivated land and other rural land, which one concerning the 3. Comparing the cultivated land and grassland, which one concerning the economic ecological benefit is more beneficial to you, and how many benefits? benefit is more beneficial to you, and how many benefits? Cultivated land Other rural land Cultivated land Grassland 9 8 7 6 5 4 3 2 1 2 3 4 5 6 7 8 9 9 8 7 6 5 4 3 2 1 2 3 4 5 6 7 8 9 12. Comparing the grassland and forest land, which one concerning the ecological benefit 4. Comparing the cultivated land and forest land, which one concerning the economic is more beneficial to you, and how many benefits? benefit is more beneficial to you, and how many benefits? Grassland Forest land Cultivated land Forest land 9 8 7 6 5 4 3 2 1 2 3 4 5 6 7 8 9 9 8 7 6 5 4 3 2 1 2 3 4 5 6 7 8 9 13. Comparing the grassland and other rural land, which one concerning the ecological 5. Comparing the cultivated land and other rural land, which one concerning the benefit is more beneficial to you, and how many benefits? economic benefit is more beneficial to you, and how many benefits? Grassland Other rural land Cultivated land Other rural land 9 8 7 6 5 4 3 2 1 2 3 4 5 6 7 8 9 9 8 7 6 5 4 3 2 1 2 3 4 5 6 7 8 9 14. Comparing the forest land and other rural land, which one concerning the 6. Comparing the grassland and forest land, which one concerning the economic benefit ecological benefit is more beneficial to you, and how many benefits? is more beneficial to you, and how many benefits? Forest land Other rural land Grassland Forest land 9 8 7 6 5 4 3 2 1 2 3 4 5 6 7 8 9 9 8 7 6 5 4 3 2 1 2 3 4 5 6 7 8 9 Note: other rural land indicates land used for raising animals, agricultural facilities, agricultural roads, pit-ponds, fishponds, irrigation, drying grains and forming ridges among croplands. 7. Comparing the grassland and other rural land, which one concerning the economic For above scoring in paired comparisons, the following Fundamental Scale is used to make benefit is more beneficial to you, and how many benefits? judgments:1= Equal; 2=Between Equal and Moderate; 3=Moderate; 4=Between Moderate and Grassland Other rural land Strong;5=Strong; 6=Between Strong and Very Strong; 7=Very Strong; 8=Between Very Strong 9 8 7 6 5 4 3 2 1 2 3 4 5 6 7 8 9 and Extreme; 9=Extreme

139 140

Appendix 2-C2: Results of Analytic Hierarchy Process Table 2-C2 Values of (farmers)

ଶ ଶ ଶ ଶ Based on the data from above questionnaires, we employ Super Decisions software to calculate ܽ ǡܾ ǡܿ ݀ Weights for interviewees in the farmer type every stakeholder’s preference on four types of rural land. And then we take the average value Cultivated land Grassland Forest land Other rural land Alternatives of all representatives of each stakeholder type to represent each stakeholder type’s decision in ( ) ( ) ( ) ( ) 1 0.4873 0.2432 0.2003 0.0692 allocation of rural land. ܽଶ ܾଶ ܿଶ ݀ଶ 2 0.3031 0.4334 0.1906 0.0729 Table 2-C1 Values of (herders) 3 0.4909 0.1336 0.2824 0.0932 4 0.4564 0.2182 0.2531 0.0723 ଵ ଵ ଵ ଵ Weights for interviewees in the ܽherderǡܾ ǡܿ type݀ 5 0.5433 0.1180 0.2527 0.0859 Alternatives Cultivated land Grassland Forest land Other rural land 6 0.4681 0.2129 0.2324 0.0866 ( ) ( ) ( ) ( ) 7 0.3800 0.2071 0.2929 0.1200 1 0.1377 0.6311 0.1032 0.1279 8 0.4695 0.2701 0.1813 0.0790 ܽଵ ܾଵ ܿଵ ݀ଵ 2 0.1207 0.6529 0.1120 0.1145 9 0.4768 0.3130 0.1466 0.0636 3 0.1232 0.5577 0.1877 0.1314 10 0.5239 0.1900 0.1831 0.1030 4 0.2252 0.5715 0.1381 0.0651 11 0.3352 0.4396 0.1739 0.0513 5 0.1730 0.5388 0.1653 0.1230 12 0.4853 0.1285 0.2944 0.0918 6 0.2014 0.5049 0.2174 0.0763 13 0.3750 0.3639 0.1942 0.0669 7 0.2781 0.4993 0.1336 0.0890 14 0.3661 0.3595 0.2215 0.0529 8 0.1836 0.3494 0.2909 0.1760 15 0.3146 0.4114 0.2041 0.0699 9 0.2409 0.5294 0.1372 0.0925 Average 0.4317 0.2695 0.2202 0.0786 10 0.2464 0.4913 0.1615 0.1007 Variance 0.0061 0.0123 0.0021 0.0003 11 0.1907 0.5105 0.1716 0.1271 12 0.2679 0.5079 0.1565 0.0676 13 0.1254 0.6754 0.1069 0.0923 Table 2-C3 Values of (ecological authorities) 14 0.3548 0.4256 0.1487 0.0708 ଷ ଷ ଷ ଷ Weights for intervieweesܽ inǡܾ theǡܿ ecological݀ authorities type 15 0.2573 0.4838 0.1779 0.0811 Cultivated land Grassland Forest land Other rural land Alternatives Average 0.2084 0.5286 0.1606 0.1024 ( ) ( ) ( ) ( ) Variance 0.0046 0.0070 0.0023 0.0009 1 0.1867 0.3099 0.3808 0.1227 ܽଷ ܾଷ ܿଷ ݀ଷ 2 0.1963 0.3318 0.3740 0.0979

3 0.1685 0.3894 0.3647 0.0774 4 0.2507 0.2620 0.4008 0.0865 5 0.1985 0.3178 0.3836 0.1000 6 0.2718 0.2930 0.3438 0.0915 Average 0.2121 0.3173 0.3746 0.0960 Variance 0.0016 0.0018 0.0004 0.0002

141 142

Appendix 2-C2: Results of Analytic Hierarchy Process Table 2-C2 Values of (farmers)

3 0.1685 0.3894 0.3647 0.0774 4 0.2507 0.2620 0.4008 0.0865 5 0.1985 0.3178 0.3836 0.1000 6 0.2718 0.2930 0.3438 0.0915 Average 0.2121 0.3173 0.3746 0.0960 Variance 0.0016 0.0018 0.0004 0.0002

141 142

Table 2-C4 Values of (economic authorities) Appendix 2-D: Estimation of parameter

ܽସǡܾସǡܿସ݀ସ ࢼ࢏ Weights for interviewees in the ecological authorities type Table 2-D1 Data for welfare weights Cultivated land Grassland Forest land Other rural land Alternatives ௜ ( ) ( ) ( ) ( ) Item Unit ߚ Value 1 0.5768 0.1787 0.1082 0.1364 Population of farmers Person 168514 ܽସ ܾସ ܿସ ݀ସ 2 0.4479 0.3403 0.0848 0.1271 Farmer’s net income in 2012 RMB per capita 6730 3 0.4815 0.2614 0.1194 0.1377 Farmers’ total income Million RMB 1134.10 4 0.3849 0.2982 0.1183 0.1986 Population of herders Person 2986 5 0.3246 0.4247 0.1374 0.1133 Herder’s net income in 2012 RMB per capita 7898 6 0.3669 0.3768 0.1242 0.1321 Herders’ total income Million RMB 23.58 Average 0.4304 0.3133 0.1154 0.1409 Ecological expenditure in 2012 Million RMB 118.96 Rural ecological expenditure share % 70 Variance 0.0083 0.0076 0.0003 0.0009 Ecological expenditure on rural area Million RMB 83.3 Economic expenditure in 2012 Million RMB 1109.42 Rural population share % 81.2 Economic expenditure on rural area Million RMB 900.8 Source: Statistical Bureau, Tai Pusi County in 2012.

Table 2-D2 Value of based on income distribution and labour force of stakeholder types

௜ Herdersߚ Farmers Ecological Economic ( ) ( ) public public Total authorities ( ) authorities ( ) ߚଵ ߚଶ Income in 2012 23.58 1134.099 83.3 ଷ 900.8 ସ 2158.669 (million RMB) ߚ ߚ Income 0.0110 0.5295 0.0389 0.4206 1 distribution( )

Labour force in୧ 1962 55665 200 6601 64428 each stakeholderȾ type in 2012 Labour force 0.0304 0.8640 0.0031 0.1025 1 distribution( ) Source: Statistical Bureau, Tai Pusi County in 2012. Ⱦ୧

143 144

Table 2-C4 Values of (economic authorities) Appendix 2-D: Estimation of parameter

Table 2-D2 Value of based on income distribution and labour force of stakeholder types

Labour force in୧ 1962 55665 200 6601 64428 each stakeholderȾ type in 2012 Labour force 0.0304 0.8640 0.0031 0.1025 1 distribution( ) Source: Statistical Bureau, Tai Pusi County in 2012. Ⱦ୧

143 144

Appendix 4-A: DescriptiveTable 4-A1 statistics Descriptive of the statist modelics variablesof the model variables WithAppendix respect 4-B: to The the modelspecification specification of our withmod quadraticels, we use term the quadratic term instead of the reciprocal of land tenure reform variable. That is, is replaced by , as Variables Mean Std. Dev. Min Max Observations ଶ every equation in below. Comparing table 4.3 with table ௜௧ିଵ4-B1, and figure 4.9 with figure௜௧ିଵ 4-B1, 0.36 0.54 -1.5927 3.14 1080 ͳΤ ሺܴ ሻ ሾ݈݋݃ሺܴ ሻሿ model results for these two types of specification are almost the same, which indicates the ݈݋݃ሺܵ௜௧ሻ -0.05 0.49 -2.08 2.93 1080 model specification is reliable and the model results are robust. ݈݋݃ሺܥ௜௧ሻ 1.02 0.81 -0.50 4.23 1080 ݈݋݃ሺܷ௜௧ሻ 1.05 0.79 -2.00 4.17 1080 ௜௧ -1.13 3.30 -4.61 3.26 1080 ଶ ݈݋݃ሺܨ ሻ ௜௧ ௜ ଵ ௜௧ିଵ ଶ ௜௧ିଵ ଷ ௜௧ିଵ ସ ௜௧ିଵ 46.69 49.80 0.04 10028 1080 ݈݋݃ሺܵ ሻൌݏ ൅ܽ ݈݋݃ሺܴ ሻ൅ܽ ሾ݈݋݃ሺܴ ሻሿ ൅ܽ ݈݋݃ሺܯ ሻ൅ܽ ݈݋݃ሺܶ ሻ ݈݋݃ሺܴ௜௧ିଵሻ ହ ௜௧ିଵ ଺ ௧ ଻ ௜ ௧ ௜௧ Ǥ ͳሻܤ൅ᖡ ሺͶ ܻכ ܲ ሻ൅ܽ ܻ ൅ܽ ܩ൅ܽ ݈݋݃ሺ 1035 0.66 -0.53 0.41 0.05 ͳΤ ሺܴ௜௧ିଵሻ 0.12 0.50 -0.45 1.13 1035 ଶ ݈݋݃ሺܯ௜௧ିଵሻ ݈݋݃ሺܥ௜௧ሻൌ݃௜ ൅ܾଵ݈݋݃ሺܴ௜௧ିଵሻ൅ܾଶሾ݈݋݃ሺܴ௜௧ିଵሻሿ ൅ܾଷ݈݋݃ሺܤ௜௧ିଵሻ൅ܾସ݈݋݃ሺܶ௜௧ିଵሻ -0.03 0.06 -0.23 0.10 1035 Ǥ ʹሻܤ௧ ൅Ɋ௜௧ሺͶܻכ ௜௧ିଵሻ൅ܾ଺ܻ௧ ൅ܾ଻ܲ௜ܩ௜௧ିଵሻ ൅ܾହ݈݋݃ሺܤ݋݃ሺ݈ ௜௧ିଵ -0.28 0.42 -2.40 0.01 1035 ݈݋݃ሺܶ ሻ 29 ଶ 12.50 6.93 1 24 1080 ௜௧ ௜ ଵ ௜௧ିଵ ଶ ௜௧ିଵ ଷ ௜௧ିଵ ସ ௜௧ିଵ ሻ൅ܿ ݈݋݃ሺܶ ሻ ܯ௜௧ିଵሻ ݈݋݃ሺܷ ሻൌݑ ൅ܿ ݈݋݃ሺܴ ሻ൅ܿ ሾ݈݋݃ሺܴ ሻሿ ൅ܿ ݈݋݃ሺܩ݋݃ሺ݈ 7.50 8.14 0 24 1080 ହ ௜௧ିଵ ଺ ௧ ଻ ௜ ௧ ௜௧ Ǥ ͵ሻܤ൅ߣ ሺͶ ܻכ ܲ ሻ൅ܿ ܻ ൅ܿ ܩ௧ ൅ܿ ݈݋݃ሺܻ ௧ ଶܻכ ௜ܲ ݈݋݃ሺܨ௜௧ሻൌ݂௜ ൅݀ଵ݈݋݃ሺܴ௜௧ିଵሻ൅݀ଶሾ݈݋݃ሺܴ௜௧ିଵሻሿ ൅݀ଷ݈݋݃ሺܤ௜௧ିଵሻ൅݀ସ݈݋݃ሺܶ௜௧ିଵሻ ହ ௜௧ିଵ ଺ ௧ ଻ ௜ ௧ ௜௧ Ǥ Ͷሻܤ൅ߜ ሺͶ ܻכ ܲ ሻ൅݀ ܻ ൅݀ ܩ൅݀ ݈݋݃ሺ

27 The logarithm of any positive number less than 1 is negative. 28 has the max value of 100 because we assigned a small original value (0.01) to . In this case, it indicates equals to 0. ௜௧ିଵ ௜௧ିଵ 29 YearͳΤ ሺܴ 1985ሻ equals 1, 1986 equals 2, etc. ܴ ܴ௜௧ିଵ 145 146

Table 4-B1 Model results based on Eqs. (4B.1)-(4B.4)

Variables Sheep population model Cattle population model Mutton output model Beef output model

coefficient t coefficient t coefficient t coefficient t ݈݋݃ሺܵ௜௧ሻ ݈݋݃ሺܥ௜௧ሻ ݈݋݃ሺܷ௜௧ሻ ݈݋݃ሺܨ௜௧ሻ -0.13*** -6.86 -0.06*** -3.03 -0.07*** -3.36 -0.04* -1.67

݈݋݃ሺܴ௜௧ିଵሻ -0.03*** -5.80 -0.02*** -3.38 -0.02*** -2.97 -0.01* -1.91 ଶ ሾ݈݋݃ሺܴ௜௧ିଵሻሿ 0.25*** 3.62 na na 0.08 1.11 na na

݈݋݃ሺܯ௜௧ିଵሻ na na 0.18*** 3.84 na na 0.19*** 2.96

݈݋݃ሺܤ௜௧ିଵሻ -0.08 -0.19 1.54*** 4.12 -0.64 -1.52 0.57 1.10

݈݋݃ሺܶ௜௧ିଵሻ 1.06** 2.02 -0.28 -0.58 1.60*** 2.90 0.81 1.17

݈݋݃ሺܩ௜௧ିଵ ሻ 0.04*** 8.47 0.01*** 2.68 0.12*** 22.23 0.10*** 17.14

ܻ௧ -0.001 -0.38 -0.03*** -8.22 -0.02*** -4.38 -0.05*** -10.39

௧ 0.31** 1.98 0.10 0.68 0.21 1.28 0.47** 2.30ܻכ Constantܲ௜

0.30 0.12 0.77 0.60 Observationsଶ 1035 1035 1035 1035 ܴ

147

Figure between 4-B1Relationship 100 105 (Index 75 80 85 90 95 1011121314151617181920212223242526 9 8 7 6 5 4 3 2 1 Year 1=100)

land tenure results intable4-B1 reform andlivestockpr 148 148

oduction basedonthemodel oduction Years beef output mutton output cattle population sheep population

Table 4-B1 Model results based on Eqs. (4B.1)-(4B.4)

Variables Sheep population model Cattle population model Mutton output model Beef output model

coefficient t coefficient t coefficient t coefficient t ݈݋݃ሺܵ௜௧ሻ ݈݋݃ሺܥ௜௧ሻ ݈݋݃ሺܷ௜௧ሻ ݈݋݃ሺܨ௜௧ሻ -0.13*** -6.86 -0.06*** -3.03 -0.07*** -3.36 -0.04* -1.67

݈݋݃ሺܴ௜௧ିଵሻ -0.03*** -5.80 -0.02*** -3.38 -0.02*** -2.97 -0.01* -1.91 ଶ ሾ݈݋݃ሺܴ௜௧ିଵሻሿ 0.25*** 3.62 na na 0.08 1.11 na na

݈݋݃ሺܯ௜௧ିଵሻ na na 0.18*** 3.84 na na 0.19*** 2.96

݈݋݃ሺܤ௜௧ିଵሻ -0.08 -0.19 1.54*** 4.12 -0.64 -1.52 0.57 1.10

݈݋݃ሺܶ௜௧ିଵሻ 1.06** 2.02 -0.28 -0.58 1.60*** 2.90 0.81 1.17

݈݋݃ሺܩ௜௧ିଵ ሻ 0.04*** 8.47 0.01*** 2.68 0.12*** 22.23 0.10*** 17.14

ܻ௧ -0.001 -0.38 -0.03*** -8.22 -0.02*** -4.38 -0.05*** -10.39

௧ 0.31** 1.98 0.10 0.68 0.21 1.28 0.47** 2.30ܻכ Constantܲ௜

0.30 0.12 0.77 0.60 Observationsଶ 1035 1035 1035 1035 ܴ

147

Figure between 4-B1Relationship 100 105 (Index 75 80 85 90 95 1011121314151617181920212223242526 9 8 7 6 5 4 3 2 1 Year 1=100)

land tenure results intable4-B1 reform andlivestockpr 148 148

oduction basedonthemodel oduction Years beef output mutton output cattle population sheep population

Summary distinguished, namely, ecological authorities, economic authorities, herders and farmers. The diverging preferences of these four stakeholder types over the different types of land use were quantified using Grasslands play a significant role in the global ecosystem, livestock production and millions of the Analytic Hierarchy Process. Weights for each stakeholder type were derived for three scenarios: household livelihoods. However, grassland degradation has become a worldwide problem, and rural equal weights, weights based on income distribution and weights based on labour force distribution. poverty has become exacerbated in some pastoral areas. The sustainable management and use of Welfare analysis was employed then to determine the individual optimal allocation by maximising the grasslands that maintain the productive and adaptive capacity of ecosystems while providing for the utility function of each stakeholder type. Social optimal allocation was derived by maximising the social well-being of human communities are crucial concerns. In practice, various policy interventions have welfare function, which is the weighted sum of individual utilities. Tai Pusi County, located in an eco- been initiated by governments all over the world to devote to grassland conservation, livestock fragile area of northern China, was taken as a case to present the empirical analysis. Individual optima production and improving local household livelihoods. This research aims to evaluate the effectiveness revealed the degree of divergences among stakeholders, and the social optima revealed the optimal of the policy interventions for grassland ecosystems and livestock production in the pastoral areas of allocation based on social welfare. The research results provide insights on how to achieve an efficient China. The conceptual framework for this research is based on Ostrom’s Social-Ecological Systems allocation of rural land, balancing the ecological and economic benefits of different stakeholders from (SESs) framework. Four specific research questions are addressed. It is hoped that the research findings different types of land. As such, the desirable area of grasslands for different users as well as for the of this thesis can make a contribution to providing some references for policy development in light of whole society is revealed, which further reflects the conflicts of different stakeholders on grassland use. the sustainable management and use of grasslands. Prior to the policy intervention of assigning the grasslands to individual households, grazing was mobile China has around 392 million hectares of grasslands, accounting for 12% of the world’s grasslands and and flexible in China, which allowed herders and their animals to roam over forage and water resources. 41.7% of the national land area. The study about China’s grassland system and its grassland policies Since the start of the land tenure reform in the pastoral areas of China in the 1980s, grassland use rights could provide significant views to understand the grasslands of this world. Inner Mongolia, that includes have increasingly been assigned to individual households and subsequently more grasslands have been the typical pastoral areas of China, is employed as the specific research region. The impacts of China’s in private use. On the other hand, most of the grasslands in China have experienced severe degradation grassland policies on the grassland systems are dissected based on the methodology of econometrics, after the reform. Chapter 3 aims at examining whether land tenure reform plays a significant role in welfare economics and institutional economics and data from remote sensing technology and socio- grassland degradation, i.e. the impact of grassland privatisation on grassland condition. It first outlines economic survey. More specifically, the major policy interventions for the governance of grasslands in the process of land tenure reform in the pastoral areas of China over time and then investigates the China include the land tenure reform during the 1980s-1990s and a series of Ecological Construction changes in grassland condition that are caused by the reform. Grassland condition is presented by Programs occurring mainly after the year 2000. The former aims at privatizing use rights of grasslands grassland quantity and quality using spatial information based on remote sensing. The timing of the to individual households and displacing traditional common use. The latter mainly concerns grassland assignment of grassland use rights and the timing of the actual adoption of private use by households conservation through employing grazing bans and deterministic stocking rates. The effectiveness of differ among counties, which allows to disentangle the impact of the land tenure reform at county-level. these policy interventions is investigated through estimating their impact on grassland condition and A fixed effects model that controls for the factors of climate, agricultural activity and environmental livestock production. The empirical analysis about policy impacts is conducted in chapter 3, 4 and 5, heterogeneity is used to conduct the empirical analysis, based on a dataset involving 60 counties of based on panel data that spans several decades. In addition, a case study based on a county in the eco- Inner Mongolia between 1985 and 2008. The model results show that the private use of grasslands fragile area is conducted in chapter 2 to study the preference for different types of rural land by different following the land tenure reform causes the reduction of the total grassland area and leads to the stakeholders, which mirrors the perception of resource users on grassland use under the social, economic degradation of dense and moderate grasslands into sparse grasslands in the long run. These results point and political settings. to the existence of “a tragedy of privatisation”, as opposed to the well-known “tragedy of the commons”.

More in detail, chapter 2 analyses the divergence in preferences among different stakeholders over the Chapter 4 aims at understanding whether land tenure reform has played a role in the changes in livestock allocation of four types of rural land: cultivated land, grassland, forest and other land. Considering the production of pastoral areas, i.e. the impact of grassland privatisation on local animal husbandry. This heterogeneity of stakeholders that are concerned with land-use decisions, four types of stakeholders are objective is investigated based on county-level data for Inner Mongolia between 1985 and 2008. The

149 150

149 150 changes in livestock production over time are examined by analysing data on changes in livestock Chapter 6 concludes this research with the main findings as well as the contribution to the existing population and meat output. The descriptive analysis shows that livestock production increased at a scientific debates. Moreover, policy implications are suggested based on the findings of this research. higher speed in the crop farming areas, but the development of livestock productivity was faster in the It is finalised with the research limitation and recommendations for future research. pastoral areas. In the empirical analysis, we employed a fixed effects model to disentangle the effects of land tenure reform on livestock production from factors related to market forces, grassland condition, technology development and environmental heterogeneity. The model results reveal that the implementation of land tenure reform had significant and negative effects on the increase in livestock production, although the total livestock production actually increased over the research period. This shows that land reform in itself is unable to offset the impact of other factors that accelerate the increase in livestock production. Moreover, the constraining effect of land tenure reform on the increase in livestock production decreases with the years of implementing land tenure reform and ultimately disappears. Remarkably, the constraining effect of land tenure reform is stronger on the increase of livestock population than that of meat output. This indicates that land tenure reform has led to an improvement in livestock productivity. In conclusion, land tenure reform puts a ceiling on livestock production, which could be a possible reason why it has been difficult to implement the reform on grasslands. However, the reform has improved the livestock productivity of pastoral areas.

A series of Ecological Construction Programs have been initiated to protect the condition of grasslands in China during recent decades. However, grassland degradation is still severe and conditions have not been restored as intended. Chapter 5 investigates the effectiveness of Ecological Construction Programs for protecting the grassland condition in the extensive pastoral areas of China. One Ecological Construction Program that has been implemented widely on the grasslands, specifically the Subsidy and Incentive System for Grassland Conservation (SISGC), is taken as the concrete example for policy evaluation. The Normalized Difference Vegetation Index (NDVI), measured with remote sensing technology, is used to quantify the grassland condition between 2001 and 2014. With data from 54 counties in the pastoral areas of Inner Mongolia, we estimate the impact of SISGC on the grassland condition. A fixed effects model is employed to control for livestock production, climate, time trends and environmental heterogeneity. The model results provide quantitative evidence that the condition of the grasslands has improved significantly due to SISGC, but that the effectiveness of SISGC was offset to some extent by other socio-economic and climate factors, such as increased producer prices and high temperatures. This may explain why the actual grassland degradation has not been prevented as effectively as was expected. In addition, the impact of SISGC was stronger in counties with a worse initial grassland condition. Furthermore, the effects of producer prices and climate changes were also more pronounced in these counties.

151 152 changes in livestock production over time are examined by analysing data on changes in livestock Chapter 6 concludes this research with the main findings as well as the contribution to the existing population and meat output. The descriptive analysis shows that livestock production increased at a scientific debates. Moreover, policy implications are suggested based on the findings of this research. higher speed in the crop farming areas, but the development of livestock productivity was faster in the It is finalised with the research limitation and recommendations for future research. pastoral areas. In the empirical analysis, we employed a fixed effects model to disentangle the effects of land tenure reform on livestock production from factors related to market forces, grassland condition, technology development and environmental heterogeneity. The model results reveal that the implementation of land tenure reform had significant and negative effects on the increase in livestock production, although the total livestock production actually increased over the research period. This shows that land reform in itself is unable to offset the impact of other factors that accelerate the increase in livestock production. Moreover, the constraining effect of land tenure reform on the increase in livestock production decreases with the years of implementing land tenure reform and ultimately disappears. Remarkably, the constraining effect of land tenure reform is stronger on the increase of livestock population than that of meat output. This indicates that land tenure reform has led to an improvement in livestock productivity. In conclusion, land tenure reform puts a ceiling on livestock production, which could be a possible reason why it has been difficult to implement the reform on grasslands. However, the reform has improved the livestock productivity of pastoral areas.

151 152

Acknowledgements Next, I would like to express my gratitude to my best roommates, Anouschka, Valentina and Nadja. Thank you all for accompanying me, supporting me and taking care of me. Anouschka, thank you very During these four years of study at Wageningen University, I have benefited a lot from the guidance much for your sharing, listening and supporting, and also being the paranymph of my PhD defence. I and support of many people. Their contributions helped me finish my PhD training and grow into a “big am absolutely delighted to have a thoughtful, respectful and trustworthy close friend like you. Valentina, girl”. I would like to express my heart filled gratitude to all of them and mention a few particularly. thank you very much for being my sister. You accompanied me to get through the tough period of my First of all, I would like to thank my four supervisors, Prof. Wim Heijman, Prof. Jikun Huang, Dr. PhD and taught me to confront the challenges optimistically. Nadja, thank you so much for supporting Liesbeth Dries and Dr. Xueqin Zhu. Wim, thank you very much for your trust and encouragement in me and teaching me how to start my PhD. You helped me get through the confused period of my PhD the course of my PhD training. No matter how difficult the issues I met, you encouraged me and helped when I started my oversea life alone. I am so grateful for having all of you such fabulous girls. Also me overcome them. Your international horizon taught me how to confront the differences of this world thanks to the girls’ boyfriends, Hildert, Thomas and Robert. Thank you for giving me the happy girls and inspired me to think about the same question from different perspectives. Most importantly, you during work time. helped me to start my journey of exploring this beautiful world. The same gratitude goes to Sonja, Many thanks to my other close friends who are around me during the past several years, Oriana, Elena, Wim’s wife. Thank you for your very warm reception when I started my study in the Netherlands. Jikun, Yu Hong, Li Meng, Chuanlan, Shiyi, Zihan, Huimin, Haibo, Haoran, Tian Yu, Jin Zhang, and Yu Wang. thank you so much for your professional guidance and generous support on the road of my academic Thank you very much for standing by me and listening to me. Many thanks go to my corridor mates, career. You saved me from the most difficult position of my PhD study. You promised to help me when Anne, Maikel, Lana, Susan, Rung and Elia. Thank you for introducing the Dutch culture to me and you did not even know me, only because you thought I am a student who needed help. You set a very teaching me how to bake. You eventually do not need to worry that I may burn the kitchen by accident good example to me on how to be a teacher. You also set a very good example to me on research by when I cook. The special gratitude goes to the best PhD students in Agricultural Economics and Rural treating your work very conscientiously and enthusiastically. Liesbeth, thank you very much for being Policy Group, Sabrina, Esther, Qianqian, Jinghui, Evert, Thomas, Hoyga, Anastasia, Dadan, Eko, my supervisor as well as my close friend. You had a great contribution to my PhD study and my life in Undrakh, Niccolo, Shanshan, Chuka, Mohammed, Bek and Praxedis. Thank you all for making me feel the Netherlands. You were always the first person who came to me and stood by me when I had any not alone on the tough road of pursuing a PhD degree. Extra thanks to the other paranymph of my PhD troubles or successes. My wish is to be a female scholar as elegant as you are. Xueqin, thank you so defence, Evert. Many thanks go to my colleagues at AEP group, Justus, Jack, Koos, Rico, Dusan, Roel much for your guidance. You taught me to be independent, which is one of the most significant things and Alexander. Thank you for teaching me and guiding me. The other special thanks go to the I learned from my PhD training. I sincerely appreciate my amazing supervision team and remain deeply supporting team of AEP group, Karen, Dineke, Betty, Marian and Gre. Thank you for giving me a lot grateful to all of you. of help to make my life in the Netherlands much easier. Extra appreciation goes to Karen for helping Then, I would like to thank my other two supervisors who supervised me during my internship at the me and supporting me with your optimism. I also would like to thank my friends and colleagues who World Bank, Dr. Klaus Deininger and Thea Hilhorst. Thank you very much for giving me the chance are or were working in Chinese Academy of Sciences and Peking University, Linlin Hou, Xiaobing to work with you and being patient to teach me about the new research domains. You are so helpful and Wang, Wei Xie, Chengfang Liu, Renfu Luo, Tianhe Sun, Dan Wang, Cheng Xiang, Jiliang Hu, Yulan thoughtful to guide me to develop my next career. Many thanks go to the other colleagues at World Bu, Shukun Wang, Guolei Yang, Xia Wang, Yuanning Hu, Jie Zhou, Mi Tian, Yangjie Wang, Shi Min, Bank, Daniel, Chala, Futoshi, Estella, Xinxin and Faye, a group of brilliant friends at Washington DC, Shaoze Jin, Yunhua Wu, Yuxian Lin and Li Wang. Thank you all for any help when I went back to Lily, James, Sofia, Hadrien, Dimitra, Cindy, and Laith, and the awesome family, Jenn, Karen, Marie, China to work. Many thanks go to a lot of friends who I did not mention here yet. Barbara, Toly and Clara. Thank you very much for making me have a wonderful life in the United Moreover, I would like to acknowledge the institutes who funded and supported my PhD program, States. I also express my gratitude to Dr. Nico Heerink, Prof. Xiaoping Shi, Prof. Xiangzheng Deng, China Scholarship Council, Center for Chinese Agricultural Policy (CCAP), LEB Foundation and Prof. Linxiu Zhang, Prof. Jinxia Wang, Prof. Fujiang Hou, Dr. Zeng Tang, Prof. Shunbo Yao, Prof. Wageningen School of Social Sciences. Xuexi Huo, Prof. Minjuan Zhao, Prof. Xianli Xia, Prof. Tianjun Liu and Dr. Jundi Liu for your any suggestions and supports to my research and my PhD study. Last, but certainly foremost, I would like to appreciate my parents. Thank you so much for your understanding and supports. Being your daughter is the happiest thing of my life, love you always. 153 154 Acknowledgements Next, I would like to express my gratitude to my best roommates, Anouschka, Valentina and Nadja. Thank you all for accompanying me, supporting me and taking care of me. Anouschka, thank you very During these four years of study at Wageningen University, I have benefited a lot from the guidance much for your sharing, listening and supporting, and also being the paranymph of my PhD defence. I and support of many people. Their contributions helped me finish my PhD training and grow into a “big am absolutely delighted to have a thoughtful, respectful and trustworthy close friend like you. Valentina, girl”. I would like to express my heart filled gratitude to all of them and mention a few particularly. thank you very much for being my sister. You accompanied me to get through the tough period of my First of all, I would like to thank my four supervisors, Prof. Wim Heijman, Prof. Jikun Huang, Dr. PhD and taught me to confront the challenges optimistically. Nadja, thank you so much for supporting Liesbeth Dries and Dr. Xueqin Zhu. Wim, thank you very much for your trust and encouragement in me and teaching me how to start my PhD. You helped me get through the confused period of my PhD the course of my PhD training. No matter how difficult the issues I met, you encouraged me and helped when I started my oversea life alone. I am so grateful for having all of you such fabulous girls. Also me overcome them. Your international horizon taught me how to confront the differences of this world thanks to the girls’ boyfriends, Hildert, Thomas and Robert. Thank you for giving me the happy girls and inspired me to think about the same question from different perspectives. Most importantly, you during work time. helped me to start my journey of exploring this beautiful world. The same gratitude goes to Sonja, Many thanks to my other close friends who are around me during the past several years, Oriana, Elena, Wim’s wife. Thank you for your very warm reception when I started my study in the Netherlands. Jikun, Yu Hong, Li Meng, Chuanlan, Shiyi, Zihan, Huimin, Haibo, Haoran, Tian Yu, Jin Zhang, and Yu Wang. thank you so much for your professional guidance and generous support on the road of my academic Thank you very much for standing by me and listening to me. Many thanks go to my corridor mates, career. You saved me from the most difficult position of my PhD study. You promised to help me when Anne, Maikel, Lana, Susan, Rung and Elia. Thank you for introducing the Dutch culture to me and you did not even know me, only because you thought I am a student who needed help. You set a very teaching me how to bake. You eventually do not need to worry that I may burn the kitchen by accident good example to me on how to be a teacher. You also set a very good example to me on research by when I cook. The special gratitude goes to the best PhD students in Agricultural Economics and Rural treating your work very conscientiously and enthusiastically. Liesbeth, thank you very much for being Policy Group, Sabrina, Esther, Qianqian, Jinghui, Evert, Thomas, Hoyga, Anastasia, Dadan, Eko, my supervisor as well as my close friend. You had a great contribution to my PhD study and my life in Undrakh, Niccolo, Shanshan, Chuka, Mohammed, Bek and Praxedis. Thank you all for making me feel the Netherlands. You were always the first person who came to me and stood by me when I had any not alone on the tough road of pursuing a PhD degree. Extra thanks to the other paranymph of my PhD troubles or successes. My wish is to be a female scholar as elegant as you are. Xueqin, thank you so defence, Evert. Many thanks go to my colleagues at AEP group, Justus, Jack, Koos, Rico, Dusan, Roel much for your guidance. You taught me to be independent, which is one of the most significant things and Alexander. Thank you for teaching me and guiding me. The other special thanks go to the I learned from my PhD training. I sincerely appreciate my amazing supervision team and remain deeply supporting team of AEP group, Karen, Dineke, Betty, Marian and Gre. Thank you for giving me a lot grateful to all of you. of help to make my life in the Netherlands much easier. Extra appreciation goes to Karen for helping Then, I would like to thank my other two supervisors who supervised me during my internship at the me and supporting me with your optimism. I also would like to thank my friends and colleagues who World Bank, Dr. Klaus Deininger and Thea Hilhorst. Thank you very much for giving me the chance are or were working in Chinese Academy of Sciences and Peking University, Linlin Hou, Xiaobing to work with you and being patient to teach me about the new research domains. You are so helpful and Wang, Wei Xie, Chengfang Liu, Renfu Luo, Tianhe Sun, Dan Wang, Cheng Xiang, Jiliang Hu, Yulan thoughtful to guide me to develop my next career. Many thanks go to the other colleagues at World Bu, Shukun Wang, Guolei Yang, Xia Wang, Yuanning Hu, Jie Zhou, Mi Tian, Yangjie Wang, Shi Min, Bank, Daniel, Chala, Futoshi, Estella, Xinxin and Faye, a group of brilliant friends at Washington DC, Shaoze Jin, Yunhua Wu, Yuxian Lin and Li Wang. Thank you all for any help when I went back to Lily, James, Sofia, Hadrien, Dimitra, Cindy, and Laith, and the awesome family, Jenn, Karen, Marie, China to work. Many thanks go to a lot of friends who I did not mention here yet. Barbara, Toly and Clara. Thank you very much for making me have a wonderful life in the United Moreover, I would like to acknowledge the institutes who funded and supported my PhD program, States. I also express my gratitude to Dr. Nico Heerink, Prof. Xiaoping Shi, Prof. Xiangzheng Deng, China Scholarship Council, Center for Chinese Agricultural Policy (CCAP), LEB Foundation and Prof. Linxiu Zhang, Prof. Jinxia Wang, Prof. Fujiang Hou, Dr. Zeng Tang, Prof. Shunbo Yao, Prof. Wageningen School of Social Sciences. Xuexi Huo, Prof. Minjuan Zhao, Prof. Xianli Xia, Prof. Tianjun Liu and Dr. Jundi Liu for your any suggestions and supports to my research and my PhD study. Last, but certainly foremost, I would like to appreciate my parents. Thank you so much for your understanding and supports. Being your daughter is the happiest thing of my life, love you always. 153 154 Biography Min Liu

Min Liu was born on the 26th of April, 1988 in Hanzhong City, Shaanxi Province, China. She studied Wageningen School of Social Sciences (WASS) Tourism Management at Northwest Agriculture and Forestry University, China from 2006 to 2010 and Completed Training and Supervision Plan got her Bachelor degree of management. She then was recommended for Master of Agricultural *One credit according to ECTS is on average equivalent to 28 hours of study load Economics and Management in Northwest Agriculture and Forestry University, and obtained her Master degree in 2012. Name of the learning activity Department/Institute Year ECTS* A) Project related competences In October 2012, Min started to study at Wageningen University, the Netherlands, funded by Chinese Advanced Microeconomics (ECH-32306) WUR 2013 6 Theories and Models in Environmental WUR 2013 6 Scholarship Council to pursue her PhD degree under the supervision by Prof. Wim Heijman, Prof. Jikun Economics (ENR-30306) Huang, Dr. Liesbeth Dries and Dr. Xueqin Zhu at Agricultural Economics and Rural Policy group. Rural Economic Analysis (AEP-31306) WUR 2013 6 During the final year of her PhD, she joined in the Development Research Group of World Bank, Advanced Econometrics (AEP-60306) WUR 2014 Cost-Benefit Analysis and Environmental WUR 2014 Washington D.C. to work with Dr. Klaus Deininger. Her current research interests focus on social- Valuation (DEC-31306) ecological systems, grassland degradation, livestock production, eco-environmental policies, land use Agricultural Economics & Policy WASS 2016 2 B) General research related competences and management, land tenure, rural household livelihoods and rural transformation. Workshop: interviewing professionals WASS 2012 0.5 Research Methodology: From topic to proposal WASS 2013 4

Doing systematic literature review WASS 2013 3.7 WASS Introduction Course for new PhD WASS 2013 1.0 candidates Techniques for Writing and Presenting a WASS 2015 1.2 Scientific Paper ‘Tragedy of the commons or tragedy of - EAAE PhD Workshop, Roma 2015 1 privatization? The impact of land tenure reform - WASS PhD day on grassland condition in Inner Mongolia, - International Conference "Food in the Bio-based China’ Economy", Wageningen - 29th International conference of agricultural economists (ICAE), Milano ‘What is the impact of land tenure reform on - Workshop on tenure reforms in different land use types 2015 1 livestock production in pastoral areas? An in China, Beijing, China empirical study conducted in Inner Mongolia, - The Fourth International Conference on Natural China’ Resources and Agricultural Economics), Hohhot, China

‘Individual and Social Optima of Rural Land The 29th International conference of agricultural 2015 1 Allocation by Stakeholders: A Case Study on economists (ICAE), Milano, Italy Eco-fragile Areas of Northern China’

‘The impact of Ecological Construction - The 2016 China Conference of the Chinese Economists 2016 1 Programs on grassland conservation in the Society, Shenzheng, China pastoral areas of China’ - Lunch seminar at the Department of Agriculture, Food and Environment, University of Pisa, Italy C) Career related competences/personal development Internship Development Research Group, World Bank 2016 6

Total 40.4

155 156

Biography Min Liu

Total 40.4

155 156