Irrigation, Institutions and Income The impact of water sector reforms on farmer income in Northwest

Msc thesis by: Koen Leuveld, Wageningen University

Supervisors: Nico Heerink and Bettina Bluemling, Wageningen University Liu Tao, Nanjing Agricultural University

Table of Contents Executive Summary...... ii Foreword...... iii 1.Introduction...... 1 2.Theoretical Background...... 4 2.1. The multi-disciplinary nature of irrigation...... 4 2.2. Nested Systems...... 5 2.3.Institutional Economics...... 7 2.4.Performance...... 8 2.5.Impact...... 10 3. Background...... 11 3.1. The Chinese water crisis...... 11 3.2.Minle County...... 12 4.Methodology...... 15 4.1.Research Overview...... 15 4.2.Data Sources...... 16 5.The determinants of WUA performance...... 17 5.1.Variables...... 17 5.1.1.WUA Performance...... 17 5.1.2.WUA Characteristics...... 19 5.1.3.Village Characteristics...... 21 5.2. Results...... 23 6. The impact of WUA performance...... 25 6.1.Variables...... 25 6.1.1.Income...... 26 6.1.2.Household Characteristics...... 27 6.1.3.Farm Characteristics...... 27 6.2.Results...... 29 7.Conclusions and recommendations...... 31 7.1.Conclusions...... 31 7.2.Policy recommendations...... 32 7.3.Discussion...... 33 References...... 34

i Executive Summary According to many observers the world is facing a water crisis. China is one of the countries facing severe water stress, and this has prompted the national government to initiate institutional reforms in the water sector. One aspect of these reforms is the establishment of Water User Associations (WUAs) in which water users are given responsibilities for managing the irrigation system. This study aims to assess the impacts on rural household income of the introduction of these participatory irrigation institutions in Minle County, Province, China. The questions that are examined here are:

1. What are crucial factors for judging WUA performance? 2. Which factors explain differences in WUA performance? 3. How does the performance of the WUA translate at farm household level? I.e. what effect does a better performing water institution have on the water use of a farmer, and farm household income, holding all other factors constant.

In order to answer these questions, this study employs a theoretical framework that aims to combine elements from irrigation literature and institutional economics. The irrigation system (containing both physical and institutional infrastructure) is seen as embedded in wider systems, such as the farm economy and the rural economy. Institutions are seen as elements of this system, that transform inputs (e.g. labour) into outputs (e.g. water delivered to the field) that have an impact (e.g. farm income).

In order to assess WUA performance a set of indicators, both physical and financial, are used. These indicators are used in econometric analyses to estimate both the factors that influence them, and their impact. This is done using data collected for 315 households and 35 WUAs in Minle County.

The results indicate that participation has a positive effect on the performance of water institutions, and that increased performance has a positive impact on cropping income. However, increased performance has a negative impact on off-farm income. This is seen as evidence that farmers shift resources away from off-farm labour in favour of on-farm activities in the face of improvements in water supply. The study did not indicate a large impact on total household income, which can be explained by the fact that the data for the study has been gathered in a year with good rainfall, while the impacts of good water management are expected to be the largest in dry years.

While it is demonstrated that participatory institutions have a positive impact on performance, more reforms are needed in the water institutions in Minle County for the farmers to fully benefit. Evidence of nepotism and adverse policies at a higher level exist. These issues might constrain the possible benefits that farmers are able to derive from water sector reforms.

ii Foreword Before you lies the thesis that is meant to be the final piece of work of my master in International Development Studies at Wageningen University. I wrote this thesis at the Development Economics group. One of the most important themes in development economics these days is the role of institutions. One of the most important themes in my studies has been water. So when the option to perform research on water institutions in a dynamic country like China presented itself, I could not refuse.

In writing this thesis I’ve been greatly helped by a number of people: my supervisors at Wageningen University: Nico Heerink and Bettina Bluemling. At Wageningen Univerity's Development Economics group, Maarten Voors and Marrit van der Berg have provided valuable input for this thesis. At Nanjing Agricultural University Liu Tao, Shi Xiaoping, Feng Shui and the people at the college of international education. In Minle County, the people at the Water Management Bureau, village leaders, and all the farmers and families interviewed were extremely helpful and kind. Special thanks go out to my fellow students Veronica Wachong Castro and Zhou Khan.

All these people helped to make the research, my stay at China, and the writing of my thesis to a pleasurable experience. I hope that reading this thesis will be as enjoyable and interesting to the reader.

Koen Leuveld

iii 1. Introduction As the world’s population grows, so does its demand for food, and hence irrigation water. Questions are arising of whether the world can sustain this increased demand for water. Although 75% of the world’s surface is covered with water, only a small portion of all this water is fresh water that is ready to use for agricultural, industrial or domestic purposes. Moreover, it is threatened by climate change and pollution, leading different scholars to state that water is a scarce good.

This is made all the more pressing by the fact that it is not only the demand for irrigation water that is increasing; a continued rise in living standards fuels a global increase in demand for domestic and industrial water as well. Irrigation is vital to maintain global food supplies and thus to keep basic food affordable to the very poor (Carruthers et al., 1997). As the cost for creating new supplies – e.g. by building dams and other water infrastructure – are increasing, the world looks to use what water is currently available more efficiently. Irrigation is often one of the uses blamed for being very inefficient.

It is for these reasons that the People’s Republic of China has sought to go beyond technical means of increasing water use efficiency in agriculture. Several parts of China, particularly the North and North- West face water shortages, and the costs of improving existing and creating new infrastructure has prompted the Chinese government to employ other methods to improve water management. Since the early 1990s responsibilities in irrigation management are gradually being devolved to water users. These have been organized in so-called Water User Associations (WUAs), in projects throughout the country (Lin, 2003). WUAs are an essential part of a pilot project called “Building a water saving society” in city (note that the word city refers not to the urban area of Zhangye, but the fact that Zhangye is a prefecture level-city, consisting of multiple counties and considerable rural areas), in the Northwestern province of Gansu. Apart from giving more responsibilities to water users by the creation of WUAs, the pilot project involves the implementation of tradable water rights. The purpose of this pilot is to draw lessons that can be implemented at a larger scale throughout China. One county in Zhangye city – Minle – will be the focus of this research. The county is a relatively isolated watershed, as little water flows out of the county. Thus an analysis of decisions made at county level does not need to take into account the complexities of externalities in other counties. This simplifies the analysis of the functioning of WUAs as there is no need to compare their efficiency of use to that of actors far outside the scope of this research. This makes for an excellent research setting.

China's efforts to implement institutional reforms in order to combat water shortages fit in a wider, global trend, in which the issues surrounding water shortages are no longer stated purely as a technical problem. It is often realized that it is not just water that needs to be managed, but to a very large extent, the success of irrigation depends on the management of people. Irrigation depends on people, staff, farmers, experts, engineers etc. and the relationships among those people to make it work. Subramanian (1997) argues that inefficiencies in irrigation arise due to two institutional factors: non- exludability and rivalry. Often, in the face of institutions lacking power and capabilities to enforce rules, actors appropriate an unfair amount of water at the expense of less privileged actors. The classic example of this is the farmer located close to the head works of an irrigation system over-irrigating, while farmers downstream are suffering water shortages. These factors are exacerbated by the complexities in water management, which means that planners have to make decisions on incomplete information. According to Subramanian local actors have the most complete information, and including them explicitly in the decision making process, ensures that adressing issues such as poverty, food security, and public health problems is done as effictively as possible.

WUAs are one of the most prominent types of institutional arrangements proposed to involve water users in the decision making and planning associated with irrigation. The discovery of farmer managed irrigation systems in 1970s (such as the Subaks in Bali) led to the realization that irrigation

1 was not necessarily a task for large scale bureaucracies run by the state. Becker et al. (1995) propose to treat water as a common pool resource, best managed by the users themselves. Water users have a greater incentive to ensure management is done in a good way than state bureaucracies. However the scale of many irrigation schemes makes some state involvement necessary; often sub-divisions of large-scale schemes are turned over to farmer, while the government bureaucracy remains responsible for trunk infrastructure (Dinar et al., 1997).

The research into WUAs focuses broadly on two questions: how can WUAs be designed to perform well; and what are the impacts of the implementation of WUAs. This study will focus on the second question. The first question has received ample attention in the literature . For this study, the crucial part from this literature is how to assess performance. Should it be only assessed by how much water is delivered to farms? Or should the quality of the canals play a role as well? Equity considerations might also be important when discussing these associations. Chapter 2 will draw on literature in order to come to a methodology with which to assess WUA performance. Some general conclusions are that WUA performance is positively affected by community organizers, a cooperative stance by the irrigation bureaucracy, and solid conflict resolution mechanisms (Becker et al., 1995; Dinar et al., 1997; Saleth et al., 1999b; Vermillion, 2004).

On the main topic of this study, the impact of the performance of WUAs, the evidence is mixed. Often, the goal of turning over irrigation management to users is to reduce costs to the governments. This goal is usually achieved (Turral, 1995; Vermillion, 1997). However, the picture on total costs is less clear. In some cases governments simply defer costs to the farmers, reducing the cost to the government while total costs remain unaffected. In other cases institutional reforms lead to increased cost-efficiency, reducing total costs (Vermillion, 1997). In general it can be said that costs to farmers often increase, but that these costs are sometimes offset by increased productivity (Meinzen-Dick, 1997). Another important concern is equity. While some studies report gains in equity (Meinzen-Dick, 1997), others (Van Koppen et al., 2002) report that poor farmers are not as well-positioned as their richer counterparts to capture the benefits of institutional reform.

In summary, it is not clear what the impacts of institutional reforms in irrigation are. This study aims to provide evidence of this impact, for the case of the reforms in Minle County. The objective of this study is not to have a focus purely on efficiency in terms of water use alone. Irrigation projects are often not only initiated with an aim of food production. Often other considerations play a role when considering new irrigation infrastructure. One such aspect of concern is poverty reduction; most of the world’s poor live in rural areas and irrigation is seen as a good way to increase their incomes. Therefore the main concern here will be what the effect of irrigation water use – and attempts to increase that efficiency– are on the incomes of the people who depend on that water for a livelihood. In order to do so a framework will be used in which irrigation is analysed as sub-system of the wider rural economy. This will allow the estimation of the impact of irrigation institutions on the incomes of their members.

In order to meet this objective the following questions have been identified: 1. What are crucial factors for judging WUA performance? 2. Which factors explain differences in WUA performance? 3. How does the performance of the WUA translate at farm household level? I.e. what effect does a better performing water institution have on the water use of a farmer, and farm household income, holding all other factors constant.

In order to answer these questions, three methodologies have been employed. Firstly, in order to come to a workable definition of the performance of an irrigation institution a literature study has been performed. Secondly, experts around Nanjing Agricultural University have been consulted, to come to hypotheses about the impact of the performance of these institutions on farmer income. Thirdly, factors explaining WUA performance and the effects of institutions have been estimated by means of

2 econometric analysis. The data for these analyses was obtained from results of a survey performed for the “Sustainable Resource Use in Rural China: Institutions, Policies and Markets” (SURE) program. This program is a cooperative research effort by Wageningen University, Nanjing Agricultural University and several other Chinese institutions. Finally, some of the observations from the expert interviews and data analysis were checked during a field visit to Minle county, where government officials and farmers were interviewed.

The structure of the remainder of this report is as follows: Chapter 2 will provide a theoretical background which combines irrigation literature and institutional economics. Chapter 3 will discuss the water situation of China as a whole, and of Minle County in particular. In Chapter 4 the methods used will be discussed, and in chapter 5 the results will be presented. Chapter 6 contains the concluding remarks and policy recommendations.

3 2. Theoretical Background

2.1. The multi-disciplinary nature of irrigation The questions put forward in the introduction are institutional and economic in nature. The theoretical framework which any attempt to answer these questions will use therefore draws on institutional economics. Irrigation is by nature a multi-disciplinary business, and while this study will focus on the institutional economics, mention must be made of where this type analysis touches on the other spheres of science associated with the study of irrigation issues, such as engineering, political science and ecology.

The most basic view of irrigation is a technocratic one. In this view irrigation can be seen as the manipulation by human action of water flows to benefit the production of agricultural crops (Small et al., 1990). It is useful to divide this definition of irrigation in two parts, means and objective. The two parts of the definition, and their limitations will be discussed separately here.

The first part of the definition, the means of irrigation, is “the manipulation of water flows by human action”. This implies that irrigation consist of no more than building dams and canals in order to control where the water flows. However, when studying irrigation it is vital to consider more than just brick and mortar. Of equal importance are the people who are in charge of the water control, and those who depend on it. Mollinga (2001) identifies two ways of examining the social processes by which irrigation water is allocated: political, and institutional.

A good way to prime man’s political instincts is to build a structure to control the flow of a vital resource, and thus control who benefits from that resource. One should not be surprised therefore that political scientists have found irrigation an intriguing area of study. From inter-state conflicts over water to clashes between farmers over irrigation cycles, it appears that political ecology’s defining statement “everything is politics” rings true nowhere more than in irrigation. This is true because of the inherent unequal power relations, asymmetric information and ubiquitous opportunities for rent- seeking in large-scale irrigation. As water flows down through a canal, it passes the head-enders before it reaches the tail-enders. This means that head-enders generally have a stronger say in water allocation than tail-enders. After all, the head-enders are in a physical position to appropriate whatever water the need. However, farmers in general have less to say than irrigation bureaucracies. Irrigation bureaucracies on their turn have less information about what’s happening on the ground than village leaders. These political relationships are as important in the study of irrigation as is the study of what the concrete dams are made of. According to Mollinga (2001), the analysis of politics: “involves looking at the process through which the social relations of power are constituted, negotiated, mediated, reproduced, transformed or otherwise shaped.”

While it is acknowledged that politics play an important role in water management, a different approach to the social side of irrigation is taken here: an institutional approach, as used in institutional economics. This means that the focus will be less on the power relations between the individual actors, and more on the rational self-interest of actors, and ways to make them compatible with the general interest. The resulting “black box” vision on the irrigation institutions, in which they are reduced to a formula transforming inputs into outputs, may seem to be limited and fail to do justice to the important role politics plays in the allocation of water, as described above. However, firstly it is important to remember that political forces remain implicit in these black boxes; the numbers associated with the formulas come from these processes, so while politics are not an explicit part of this research, implicitly they have a role to play, as the defining elements of the processes described here. Secondly, the interest here is not to specifically map the political arena of Minle County. The focus of this study is the effect of the reforms efforts undertaken not only in Minle, but throughout the world. It is therefore important to use methods and variables that are applicable in other settings. “Looking at the

4 process through which the social relations of power are constituted, negotiated, mediated, reproduced, transformed or otherwise shaped” as Mollinga suggests would ultimately limit the application of this research to the Northwest of China, as these processes are very likely to work differently in different settings. Mollinga's realization does have implications for generalizing the results from this research to other settings. As the linkages found here depend to a large on local power relations, the results do not necessarily apply in other settings. Nevertheless, as the research itself can be replicated in other settings, the adoption of an institutional view on the social side of irrigation allows the research to be relevant across different settings.

The second part of the technocratic definition – the production of crops being the goal of irrigation – is also limited. It is posited here that irrigation serves several other functions that might or might not be desirable, but should be considered when analyzing the performance of irrigation. Irrigation systems and their related social and physical infrastructure may be used to produce rents for irrigation agencies, state control over marginal areas, protection against floods or – most importantly for this study – livelihoods for local producers. A core argument of this study is that the main impact of irrigation systems and related institutions is not on yield, but on farmer livelihoods.

2.2. Nested Systems In the previous section it has been discussed that there are technical, social, political and institutional aspects to irrigation. This poses some problems for an analytical framework. While the realization that many disciplines of science are applicable to the challenges posed by irrigation is an important one, for the purposes of this research it is impractical to draw on every single one of them. While the specification used for the construction of reservoirs and dams might have a very strong effect on realities in irrigation districts, they are considered out of the scope of this research. The framework employed must allow differentiation of what to include in the analysis and what not. The object of this study is not to develop one unified theory with which to analyze irrigation systems. Instead irrigation is seen as a nested system, consisting of numerous subsystems. The issues relating to each subsystem will require a smaller set of scientific disciplines to study. This means that one is able to focus on just one set of problems, and the disciplines that relate to it. The nested system approach used here is inspired by the work of (Small et al., 1990).

For the approach to have any meaning, it is important that relevant boundaries between the different subsystems are chosen. This can be done along physical or institutional lines. As this study focuses on the institutional aspects of irrigation it makes sense to adopt institutional boundaries when analyzing the system. These boundaries lie between the body overseeing water resources at the catchment level, which supplies it to the different irrigation districts which supply it to water user associations, which supply it to farm households. For a more comprehensive discussion of the institutional structure of water management the reader is referred to chapter 3. For now it's sufficient to note that the boundaries delimiting the constituent elements of the irrigation system are defined along these institutional boundaries. As can be seen from figure 1, the irrigation system itself is nested in an even larger set of systems. It would be possible to define subsystems of a larger scale (such as the global economy) or systems between the ones mentioned here, but for the purposes of this study this is unnecessary.

5 Rural Economic System

Agricultural Economic System

Irrigation System

Figure 1: Irrigation as a nested system (adapted from Small et al., 1990)

As represented by the arrows in the figure above, each sub-system of a large system draws inputs from outside or other sub-systems, and transforms it into outputs, which can have impact on other systems. The outputs of one subsystem are used as inputs in another part of the system. A further analysis of these relationships is depicted in figure 2.

Inputs Final Outputs Intermediate Impacts Outputs Water, WUA labour, performance: financial meetings, Farmer Financial contributions, maintenance income, viablility, connections, activities etc. run-off investments, etc. etc;

Figure 2: Model of WUA organization as a process (adapted from Small et al., 1990). The main unit of analysis in this study is the WUA. Viewed from this framework it uses water and money to turn it into irrigation service. The output is analogous to performance, discussed above. Because the research area for this study is relatively homogeneous, inputs in terms of labour and money available to WUAs are relatively constant. It is the way in which the WUAs are organized that forms the significant determinant for the level of outputs, and thus is of interest of this study. Hence, the links between “intermediate outputs”, “final outputs” and “impacts” are the concern of this study. The way a WUA is organized (how they run meetings, elections etc.) has an impact on its performance in terms of output; such as investments and canal quality. This performance in its turn has an impact

6 on the members of the association, in the form of improved livelihoods, or more specifically: a higher income.

2.3. Institutional Economics Farmers do not construct dams and canals all by themselves. A set of institutions have been devised to oversee the transformation of melt water in the mountains, through dams, weirs, canals and gates into soil moisture in the crop root zone that can be used for crop evapo-transpiration, by the use of which the plants form a yield. The level of coordination between the elements of this set of institutions determines how effectively the water is transformed into crops; a smooth cooperation between farmer, WUA and water management bureau ensures that water flows arrives at the field when needed and expected. If one were to adopt a purely neo-classical way of thinking, the best way to ensure an optimal level of coordination would be to introduce a free water market. And indeed, much work has gone into research on how such markets can be established (Howe et al., 1986), but also why such markets won’t work (Moore, 1989). A crucial observation regarding the limitations of free markets in general is the Coase Theorem: If there are no transaction costs (the concept of transaction costs will be thoroughly discussed in the next paragraph), the allocation of property rights does not matter for the realization of an efficient outcome. Or conversely, if transaction costs are present, the initial allocation of property rights does matter (Allen et al., 2002). In such a case a free market would not be the most efficient way to allocate scarce resource. If the initial allocation of resources is inefficient, transaction costs could severely limit the market’s ability to come to an efficient allocation. A large literature on water sector reform exist (see Saleth et al. (2004) for a thorough review), and the consensus is that the allocation of property rights does seem to matter. Systems in which farmers at least partially own the water, or the water system, perform better than fully state owned systems (for evidence in China, see Lin 2003). If the initial allocation of property rights matters, according to the Coase Theorem, there must be significant transaction costs.

What are these transaction costs? Allen et al. (2002) consider two types of definitions: The neo- classical, and the property-rights approach. The neo-classical definition was first put forward by Demsetz (1968). It focuses purely on “the cost of exchanging ownership titles”. This is a narrow definition as after the transfer has been made, monitoring and enforcement can impose significant costs as well. Allen et al. (2002) expand upon this definition to come to what they call “the property rights approach to transaction costs”. In this approach the neo-classical definition is expanded by including ex-post costs, like monitoring and enforcement, and define transaction costs as: “the costs of enforcing and maintaining property rights” Allen et al. (2002). It might appear paradoxical that this definition does not in any way mention transactions. It is not just the transaction itself that brings about costs, as Demsetz proposed. Before ownership titles can change hands, actors need to find each other, and all the relevant information surrounding the transaction. After the transaction, maintaining property rights can be costly. If the context of the rights is uncertain, unforeseen circumstances can arise, which can lead to conflicts. Resolution of conflicts can be very costly, and these types of costs should also be included in a definition of transaction costs. The transaction itself is thus merely one stage in the process in which transaction costs can arise.

In a water setting this realization is particularly useful, as water allocation is notoriously uncertain, due to the unpredictable nature of water flows. (Young, 1986) defines transaction costs in a water setting as “The resources required to establish, operate, and enforce a system to govern resource allocation.” Market transactions often do not take place with irrigation water, but irrigators do perceive to have certain rights to water. The enforcements of these rights can be difficult. It can be difficult to observe the effort that an irrigation agency puts into the proper allocation of resources (both physical and financial). Water flows can be variable, but the cause of this variability can be difficult to establish. Is reduced flow of water caused by less precipitation high in the mountains or due to poor management by the water institution?

7 The level of transaction costs is for a great part determined by the institutional arrangement governing the specific transaction (Coase, 1937; Coase, 1960). Institutions can be defined as “collective conventions and rules that establish acceptable standards of individual and group behaviour” (Bromley, 1982). Coase’s observation was that several forms of organizing transaction (e.g, the free market, or the firm) exist in order to economize on transaction costs. Williamson (1981) expanded this idea by detailing where exactly these transaction costs came from. Two of these origins lie in human nature: human actors are boundedly rational, and exhibit opportunistic behaviour. This means that no contract (at least one drafted by humans, who are limited by their bounded rationality) can never contain all contingencies, so there will always enforcement costs. In some situations these costs will be negligible, but if transactions grow more complex, transaction costs rise. Williamson identified three sources for this complexity: frequency, uncertainty and the degree to which the transaction prescribes assets that are specific to that transaction.

In the realm of business, for which this theory was originally developed, organizations that feature low transaction costs gain a competitive advantage, and so firms have an incentive to economize on their transaction costs, for otherwise they will be competed out of the market. In the water sector, different water providers often do not compete head-to-head. As a farmer it is often not possible to choose a water provider for irrigation services (though sometimes surface water can be supplemented by ground water). This results in a lack of incentives to economize on transaction costs. It is therefore of little surprise that there is a rich variety in the institutional arrangements in the water sector; none has been able to make the others irrelevant, as would happen in a business setting.

Dinar et al. (1997) classify this diversity into four distinct groups: Marginal cost pricing; state allocation; water-markets; and user-based allocation. Marginal cost pricing rests on the notion that for an efficient allocation of water each user pays the full economic opportunity cost of the amount he uses. Technically this can be hard to achieve. Water can also be allocated by the state. This allows equity considerations to be included in the allocation-process, but some efficiency might be lost. Rent- seeking behaviour is also a risk. Water-markets involve the buying and selling of water rights. This can improve efficiency, while at the same time respecting farmer’s historic rights, to a certain extent. User-based allocation can effectively be employed at small scales, to allocate water according to specific local needs.

None of these can be universally applied; each setting requires a specific institutional arrangement that minimizes transaction costs. However, as noted in Chapter 1., participatory institutions have been gaining preference over the last two decades. From an institutional perspective, it might be hypothesized, that participatory institutions do have incentives to economize on transaction costs on behalf of their members. In theory, this would promote greater allocative efficiency, and therefore greater welfare.

2.4. Performance The thesis that participative institutions promote allocative efficiency is somewhat esoteric. The nested system framework proposed above provides ways of making this more concrete. WUAs have a certain task to fulfill in the system, and how well they perform this task can be measured. So in order to measure their performance, it is first needed to define performance. Bos et al. (1993) define the performance of an organization as follows: • The degree to which an organization's products and services respond to the needs of their customers or users • The efficiency with which the organization uses the resources at its disposal.

This is a very broad definition. It merely states that one should consider the resources used (in this case water) and how well the services created with that water (irrigation) correspond to the needs of

8 the users (farmers). But a ready-to-use definition needs to be more specific than this. Molden et al. (1998) discuss the criteria performance indicators should satisfy:

• The indicators are based on a relative comparison of absolute values, rather than being referenced to standards or targets. • The indicators relate to phenomena that are common to irrigation and irrigated agricultural systems. • The set of indicators is small, yet reveals sufficient information about the output of the system. • Data collection procedures are not too complicated or expensive. • The indicators relate to outputs and are bulk measures of irrigation and irrigated agricultural systems, and thus provide limited information about internal processes.

While most of this might be obvious, it is still a good set of criteria to follow. Still, it is not specific enough to be a set of ready-to-use indicators. Molden et al. (1998) continue to propose a set of indicators. Most have the form of: Gross output/technical indicator (such as unit command, or amount of water applied). The problem with this is that gross output is determined by many more things than just irrigation. In this study, the output of irrigation is viewed as just one input to the process that in the end determines farm household income.

A ready-to-used definition that complies to the requirements put forward by Molden et al. (1998) is proposed by Saleth et al., 1(999a). They distinguish between four kinds of performance:

• The physical performance of the water sector is evaluated in terms of the following aspects: (a) Demand-supply gap, b) Physical health of water infrastructure, (c) Conflict resolution efficiency (low-cost and less time), and (d) Smoothness of water transfers across sectors/regions/users.

• The financial performance of the water sector is evaluated in terms of the following aspects: (a) Investment gap (actual vs. required) and (b) Financial gap (expenditure vs. cost recovery).

• The economic efficiency of the water sector is evaluated in terms of the following aspects: (a) Pricing gap (water prices vs. supply cost) and (b) Incentive gap (water prices vs. scarcity value of water).

• And, finally, the equity performance of the water sector is evaluated in terms of the following aspects: (a) Equity between regions, (b) Equity between sectors, and (c) Equity between groups.

This will be the definition used here. However due to data constraints, no estimates of the economic efficiency and the equity performance could be made. There is not likely to be a great difference in these within Minle. Economic efficiency depends on the allocation vis-à-vis other sectors. This can reasonably assumed to be equal across the county as the policies defining this type of efficiency are set at county-level by the water management bureau. For similar reasons conflict resolution and smoothness of transfers have been left out of the analysis. As for equity, the main concern in irrigation in this respect is the question of head-enders versus tail-enders. In Minle most farmers have holdings both in the head-end of the canal and in the head-end, so at WUA level, equity is not a concern. This

9 leaves two measures for physical performance, and two for financial performance. Detailed descriptions of how these indicators have been operationalized are provided in Chapter 5.

2.5. Impact The performance above relates a set of inputs to a state the WUA is in: the WUA uses the inputs at its disposal in a certain way to produce canals, investment activities etc (see Figure 2). However, these factors reflect only the state in which the irrigation system is in. It is argued here that the irrigation system is a nested subsystem of a larger system (see Figure 1), and each subsystem has an impact on the part of the system in which they are nested. The irrigation system is nested in the agricultural economic system. It is expected that if the irrigation system performs better, this has a positive impact on the rural economy. This has several reasons, both direct and indirect. A direct reason is that increased performance by definition means increased physical health of the irrigation infrastructure, so more water reaches the fields, entailing potential benefits to crop production. Moreover, this water can be timed more precisely, allowing each unit of water to be used more efficiently. Indirect effects are more intangible: because the water supply is more predictable, farmers can make better crop choices, as they have less reason to take into account the risk of crop failure due to water shortages. They can focus more of their resources on the most profitable activity, without the risk of losing all their income in the case of severe water shortages.

10 3. Background This section will provide the background for the current study. Firstly, the current water situation in China, and responses to this situation are discussed. Then, the situation in Minle county will be discussed. After that, some details on how the data for the research was gathered in Minle will be presented.

3.1. The Chinese water crisis China is widely believed to experience a water crisis. Although the country possesses large endowments of water resources, an estimated 2812 billion cubic metres in 2004 it is also the most populous. This results in a per capita water availability of 2206 cubic metres per year (Shalizi, 2006). However, the diverse nature of the country means that this figure varies widely between regions; in some basins in the North of China, water availability is as low as 427 cubic metres per person per year, the limit for water stress being 1700 cubic metres (Hu et al., 2006). The problem is exacerbated by the fact that the precious supplies of fresh water are distributed very unequally across the country. 81% of the country’s water resources are located in the South of the country, while 65% of all arable land is located in the North (the dividing line between north and south being the Yangtze river) (Hu et al., 2006). This leads to severe water stress in the Northern part of the country the area called the North China Plain in particular (Lohmar et al., 2003; Varis et al., 2001; Yang et al., 2001; Yang et al., 2003). Water demand is not decreasing, so the problem is only expected to get worse. Total water use in China increased fivefold between 1949 and 1998, and it is still rising (Lohmar et al., 2003).

The largest water user by far in China is irrigation: it accounts for two thirds of water use (Nickum, 1998). This water is crucial to feed the growing Chinese population. It is, however, used in a wasteful way: the efficiency reached by Chinese irrigators is 25-50% lower than of their counterparts in the developed world (Liu et al., 2002). This waste stems from the fact that most of China’s irrigation infrastructure is ageing badly. Most of it was constructed between the 1950s and the end of the 1970s, with relatively primitive technologies (e.g. earthen dams, rather than concrete ones) (Nickum, 1998). Talking of efficiency in terms of waste can be quite deceptive. After all, water that seeps from canals is not destroyed, but re-enters the hydrological cycle, where it could, for example, recharge aquifers or flow to downstream users (Seckler et al., 2003). However, irrigation water can be used by the plants it is intended for if its application is timed correctly. And while seeping water is not destroyed, its timing is altered in such a way that could make it less useful for crop production (Lohmar et al., 2003). This means that at the water utilization levels realized in China, where every litre of water must be used optimally, mis-timed water means a loss to society. It should be noted here that while water use efficiency is lower in China than in many other water scarce countries, as described above, recent trends are somewhat positive. Due to increased use of sophisticated irrigation technololgies (e.g. drip irrigation) water use by agriculture has decreased by about 5 percent, while irrigated area has been increased by 5,5 percent (World Bank, 2006).

Before discussing the possible solutions to China’s water crisis, it must be said that not all experts agree that the situation is as dire as the numbers above suggest. There has been sufficient water to expand irrigated area while not disrupting industrial output (Lohmar et al., 2003; Nickum, 1998).Institutions can be expected to adapt to the situation, although this might be a slow process (Nickum, 1998; Nickum, 2005). And while China might have issues, on a global level there are regions faring worse, such as South-Asia and Central Mexico (Alcamo et al., 2002). China is nevertheless searching for an answer to the challenges its water situation poses, and to mitigate the costs water scarcity imposes on society – it is estimated that the drying up of the Yellow River cost the Chinese economy 4 billion Yuan annually in the 1990s (Liu et al., 2002). One of the most eye- catching efforts of the Chinese government to deal with the problems is the multi-billion dollar construction program to bring water from the South of the country to the parched North (Liu et al., 2002; Wang et al., 2005). The construction of new dams and other infrastructure projects is unlikely to

11 be the only solution to the problem though; in some basins water utilization levels exceed 60% or even approach 100%, where 30-40% is recommended by experts (Yang et al. 2003). Most scholars recognize that institutional reforms are necessary to overcome the crisis (Liu et al., 2002; Nickum, 1998; Nickum, 2005; Varis et al., 2001).

This is not to say that technical interventions should not play a role. New technologies can assist in increasing the efficiency of irrigation. Most farmers in China rely on flood irrigation, but with more sophisticated technologies such as drip or sprinkler irrigation allow for the amount of water applied to be more finely adjusted to crop needs. This would enhance efficiency, as farmers are able to achieve more production with less water. Such innovations are costly however, and currently farmers face few incentives to save water. In large parts of the county, they are assigned a certain amount of water, and pay for it. Using less water does not deliver any benefits to the farmers. Charging each farmer individually for the amount of water he uses would provide such an incentive. However to this in surface irrigation systems is complex, certainly if such a thing is to be done for the millions of farmers with fragmented lands in China. One institutional way around this issue is to organize farmers in groups that collectively negotiate with the irrigation district over the terms of water supply. Such a reform is what is being implemented in Minle County, Gansu Province and is what this study will focus on.

3.2. Minle County Minle County is located in the upper reaches of the Hei river basin. The Hei river is an inland river: its water does not flow to the sea. Instead it flows to a the Juyanhai lake, in the northern part of its basin in the Inner-Mongolia Autonomous Region (see Figure 3 for a map). This lake does not have any connection to the sea. All water that flows into it leaves it by evaporation. The main source of water for the Hei River melting snow from the , located in Province in the southern part of the river basin. The tops of these mountains can reach over 5000m, and water (in the form of ice and snow) is abundant there. This means that the river has a relatively high runoff (Li et al., 2001). The water from the river turns the Hexi corridor –where the Silk Road passed between the Gobi desert and the Qilian Mountains– into a patchwork of fertile oases, allowing the cultivation of crops, turning an area that would otherwise be dry and barren to be one of the major commodity grain producing regions of China (Zhang, 2007). Major crops produced in the area are wheat, barley, maize and potatoes.

Minle County itself is spread between the foothills of the Qilian Mountains and the lower lying Hexi corridor. This results in different agro-ecological conditions: at the higher elevations, rainfall is relatively abundant, while at the lower altitudes it is almost non-existent. For water management, three zones are recognized, each with its own water requirements: zone 3 is the highest. It does not really need irrigation as badly as the other two regions, as rainfall is quite abundant. Its standard allocation of water is 72 m3/mu. Zone 2 is in between and has a standard allocation of 77 m3/mu. Zone 1 is the lowest zone. This means it is at the tail end, and the driest part of the county. Its standard allocation is 80 m3/mu.

As in other parts of the North of China, water resources have become overstretched; due the high amounts of water withdrawn from the river, Juyanhai Lake, dried out in 1992. This led to vast desertification, which has a grave impact on the local ecosystem as well as livelihoods (Zhang, 2007). The shortage of water can also lead to conflicts among users: in 2002, 4 people died and 30 were injured in a dispute over water between two villages.

12 Figure 3: Location of Minle County and the Hei River (Heihe) Basin within China.

Low precipitation in the area is partly to blame for the water shortages. Average annual precipitation in Zhangye city is less than 300mm, while evaporation can be as high as 1700mm (Li et al., 2001; Zhang, 2007)). Inefficient infrastructure also takes its toll. Only 20 to 30% of water diverted for agriculture is used productively (Zhang, 2007). The rest is lost through relatively primitive irrigation technologies, leakages, and evaporation.

The shortage of water has induced the Ministry of Water Resources to initiate a pilot program in Zhangye city, to which Minle county belongs, called “Building a water-saving society in Zhangye City”. One of the key aspects was the introduction of water use rights, and water user associations. The water use rights theoretically allow farmers to buy and sell rights to water, which should provide an incentive to use water more efficiently. In practice however, water use rights are hardly ever traded (Zhang, 2007; Zhang et al., 2009).

The new institutional arrangement for water allocation is detailed in Figure 4. The highest authority at County level is the Water Resource Bureau. The ministry decides on the inter-sectoral allocation of

13 water. Currently about 70% of water resources in Minle are devoted to irrigation, 10% to industry and 15% to domestic use (Minle County Ministry of Water, Pers. Comm.). The irrigation area Water Management Bureaus (WMBs) are responsible for the management of the main conveyance infrastructure in the irrigation sheds of Minle county. There are four WMBs in Minle County: two large ones (>300.000 mu), and two smaller ones. The WUAs are organized at village level, and are responsible for the allocation of water among their members, and the maintenance of secondary and tertiary canals.

County Ministry of Establishes the institutions relating to water management Water Resources

Participants: leaders of the County Ministry of Water Resources, Irrigation Area Water Management Bureau, and WUA. Large-scale meeting Irrigation Area Water to discuss the water Management Bureau allocation Purpose: To decide the time and quantity of flooding, etc.

Small-scaled meeting Participants: leaders of the village, WUA, as well as the villagers. WUA to discuss the water allocation

Purpose: To communicate the instructions of the above large-scaled meeting, arrange the time of flooding, and allocated the water quantity to each household.

Farm households Report their water use rights areas

Figure 4: Organization of water management in Minle County This type of reforms has had its success in Zhangye city (of which Minle County is a part), where withdrawals from the He river significantly reduced (Zhang, 2007). This has restored the flow to lake Juyanhai, which has returned to the size it had in the 1950s. It must be remarked that surface water use has been partly substituted by ground water use. This would mean that one environmental problem is exchange for another. For Minle county, this is not a large problem: ground water is either too deep, or it is prohibited from being exploited.

Some doubts to the effect of the introduction of WUAs do remain. One reason for this is the ease with which existing collective village institutions can be “rebranded” as a WUA (Wang et al., 2005). The village leader can easily assume the title of WUA leader, and still take the most important decisions. Another reason is that the potential benefits of WUAs are less tangible. Increased transparency, accountability and security are very desirable, but hard to measure.

14 4. Methodology This section will provide a brief overview of the methods used in order to answer the questions identified in Chapter 1. Firstly relevant variables and their interrelations will be identified to provide a stylized overview of the research performed in this study. Secondly, the data sources on which this study draws will be identified.

4.1. Research Overview The research has been done in three steps, each corresponding to one of the research questions identified in Chapter 1. The first question has been mostly answered by means of literature study, and has been described in Chapter 2. It identified important characteristics that affect how well the inputs from figure 2 (labour, financial contributions etc.) are transformed into performance. It is expected that participative institutions perform better than non-participative ones. Finally some measures for performance were obtained from the literature, that is helpful in comparing the performance of different WUAs.

The second step is then to determine to what extent each of these characteristics matters in shaping performance. To this end a regression analysis has been performed. As the research area is relatively homogenous, the difference in the inputs available to the WUAs are not large, and it is expected that the way in which the WUAs are organized has a significant impact on performance. This is schematically represented in the left half of figure 5. Village characteristics contain the inputs to the process that are not expected to be homogeneous (e.g. some villages have better political connections, or are located in different agro-ecological zones, or further away from the main canal). WUA characteristics are the variables that are the most important in this step; they contain indicators on how the WUA is organized, and on how participative the WUA is. A more detailed definition of each of these indicators is provided in Chapter 5.

The second step is purely related to the irrigation system. The aim of the third step in the research is then to estimate the impacts of performance on the rural economy. The way this is done is schematically represented in the right half of figure 5. Income is hypothesized to be affected by several sets of factors. Both on-farm and off-farm income will be analyzed, as WUA management might affect labour requirements for irrigation and thus have an effect on both types of income. Apart from WUA performance (as discussed above), a set of traditional determinants of income (e.g. amount of land) will be used. These controls are called farm and village characteristics in figure 5. A more detailed discussion of the relationships between these variables will be provided in Chapter 6.

A chain of relations exist from left to right, of figure 5, from WUA characteristics to farmer income. This will allow conclusions to be drawn about the impact of WUA characteristics – such as participation – on income.

15 Village Village/Farm Characteristics Characteristics

WUA WUA Household Characteristics Performance Income

Figure 5: Schematic representation of relationships to be estimated

4.2. Data Sources For collection of data this study draws on three sources: literature study; expert interviews at Nanjing Agricultural University (NJAU) and in Minle County; and survey results from Minle County.

The expert interviews at NJAU were performed at the early stages of research. Hypotheses from the literature were verified with scholars at NJAU in order to verify if they were applicable in a Chinese setting. With this input, variables and equations were specified. The preliminary results from these equations were checked during a field visit to Minle County in December 2008. During this field visit representatives from the county water ministry, WUAs and farmers were interviewed.

The quantitative data for the research was gathered in Minle County in May 2008. There are 53,731 households, in 211 villages which are in turn located in 10 townships in the research area. As these townships are located in three different agro-ecological zones and six different irrigation districts the sample was stratified according to the townships to ensure the sample represents a proper spread over all agricultural conditions. This ensures that the sample represents the differences between the townships and irrigation districts, the administration of which may have a profound impact on water management. In each township, 10% of the villages were randomly selected, resulting in 21 villages being visited. Within each of these villages, 15 households were randomly selected, and interviewed. The total sample size is therefore 21*15 = 315.

Furthermore, in each village the WUA leader was interviewed about the management of the WUA. To increase the number of observations at this level, 14 additional villages were visited

16 5. The determinants of WUA performance One of the goals of involving water users in the management of irrigation systems is to increase the performance of these systems. In Chapter 2 a definition of this performance has been given. This definition is used here in order to verify the commonly held view that participation increases performance. First the variables and model specification used for the analysis will be discussed, followed by a discussion of the results.

5.1. Variables As can be seen from figure 5, two sets of variables are thought to influence performance: WUA characteristics and village characteristics. The WUA characteristics include a number of issues such as: farmer participation in decision-making, the election mode etc. These are thought to influence the quality of WUA decision making, and thus performance. The second set of variables is a set of village characteristics. These are included because it was noted that some villages might have been advantaged by the Water Management Bureau because of several factors that are unrelated to the quality of WUA decision-making, e.g. being on location on a main road (and hence very visible) or being the birthplace of an important regional leader. Both of these could affect the assistance a village gets from higher-level governments.

The way this impact is estimated is to perform an OLS regression. As there are five indicators for WUA performance, there will also be five separate regression, all in the form:

WP = c + βwc’WC + βvc’VC

Where: WP is a vector of 5 indicators of WUA performance; WC and VC are matrices of WUA and Village Characteristics respectively; Both β-terms are vectors of the parameters to be estimated.

The remaining part of this chapter will further describe the variables – WUA performance, WUA characteristics and village characteristics.

5.1.1. WUA Performance Some details on how to measure the performance of a water institution have already been given in chapter 2.4. The relevant indicators selected there were: demand-supply gap, physical health of water infrastructure, investment gap, and financial gap. These have been operationalized in the following ways:

Water demand-supply gap: Water per mu A gap of between supply and demand would imply that the value for the indicator can assume both positive and negative values. A negative gap would imply that more water is supplied than strictly necessary. However, as will be discussed in chapter 6., the household level analysis requires natural logs to be taken of all indicators. As this is impossible for negative values, a ratio has been chosen, where the supply (in cubic meters per mu) has been divided by the demand.

One complicating factor in this is that there are multiple rounds of irrigation, up to four in the research year, but not all these rounds where used by the WUAs. This depends on local conditions, and crop choices. Therefore an average over all the cycles used by the WUA has been used. There are also rounds of irrigation after the growing season, but as the effects of these rounds are not comparable with the regular rounds these have been left out of the consideration.

17 The figure used to determine demand is the standard water allocation for the agro-ecological zone the WUA is located in.

Physical health: Secondary and tertiary canal quality A canal system is divided in several levels of canals. Primary canals feed water from the reservoir, to secondary canals, which branch off these main canals. Secondary canals feed to their branches: tertiary canals, and so on, until the water flows into the farmers field. These fields are often located at small, quaternary canals. WUAs have responsibility over the secondary and tertiary canals in their village, the quaternary canals are the responsibility of the farmer group with land at that canal. So only the secondary and tertiary canals will be examined here. They will be examined separately, as the health of the types canal might not have the same impact: the water to multiple tertiary canals flow through the same secondary canal.

These two types of canals can be subdivided in canals of four types of quality: canals with no lining whatsoever, canals lined with stones, canals lined with stones and concrete, and canals lined entirely with concrete. Concrete canals are the smoothest, and so deliver the water quickest from one end of the canal to the other, resulting in the smallest conveyance losses. In order to quantify quality for the purposes of this research, the ratio of canals treated with concrete has been determined. The data on this has been provided by WUA representatives.

Investment gap: Investment Density As it’s very hard to determine how much investment would be needed, the amount of investment in 2007 is taken, corrected for the total length of the canals in a village. Both farmer’s work, and government aid has been taken into account in this. Farmers’ work has been valued at the average wage in the county.

Financial gap: Financial ratio Here a gap measure is the only appropriate measure. The total expenditures over 2007 have been subtracted from the total revenues in the same year. Some WUAs have a larger budgets than others, simply because they have more members to contribute water fees. In order to correct for this, the resulting gap has been expressed as a share of the total WUA income.

Table 1: WUA Performance, descriptive statistics

Indicator Mean Std. Dev. Min Max

Investment Density 103 224 0.00 907

Secondary Canal Quality 0.73 0.34 0.00 1.00

Tertiary Canal Quality 0.4 0.44 0.00 1

Water per Mu 2.33 1.12 0.71 5.56

Finance Ratio 1.66 2.43 -0.15 10

18 As can be seen from the table above, there is a certain amount of variance in the performance of the WUAs. One of the striking features is the fact that WUAs provided more than twice the standard amount of water (average “Water per mu is 2.33). This is because 2007 was characterized by abundant rainfall, and hence there was plenty of water to supply more than the standard amount. Still, some WUAs did not achieve this, resulting in a figure lower than one (the minimum is 0.71). Variance within the investment density is large; the average is less than one half standard deviations above zero. The fact that secondary canal quality is higher than tertiary canal quality makes sense, as larger amounts of water flow through the secondary level.

5.1.2. WUA Characteristics There is a substantial literature on what makes a water institution work well. A main conclusion of this literature – and indeed part of the justification for the world wide adoption of WUAs – is that an institution that is participatively run performs better than one that is bureaucratically or technocratically run ((Becker et al., 1995; Meinzen-Dick, 1997; Vermillion, 1997; Vermillion, 2004). This hypothesis will be tested in this research by including variables in the WUA-level regression that reflect how participative the WUA is.

Democratic elections The first such variable is whether or not the WUA leader was elected democratically. In the majority of the cases, the village leader acts as the WUA leader. Only in two villages is this not the case. Remarkably, these two WUA leaders were among those not appointed by democratic ways (i.e. not elected by member households or their representatives).

Decision making participation The process by which the association leader is elected is important, but after a leader is elected, there are many decisions to be taken. These operational decisions can be made in a participative way, or not. Most studies seem content with a view of participation as a dichotomous variable: irrigation is either participative or not. From there these studies go on to assess the advantages and disadvantages of participation ((Lin, 2003; Subramanian, 1997; Vermillion, 1997). In this case, this dichotomous distinction is not relevant: some WUAs devolve many decisions to their members, some only give them indirect control over a number of decisions, while still others might not give their members any say. The data used for this research contains information on how seven types of decisions – ranging from irrigation order, to the assigning of guards to oversee water allocation – are taken. Nine ways of taking decisions were identified in the research, decisions can be taken by: 1. WUA director 2. WUA director, approval of general meeting of all WUA households needed 3. WUA director, approval of meeting of representatives of WUA households needed; 4. General meeting of all WUA households 5. Meeting of representatives of WUA households 6. User groups in the WUA; 7. Individual households; 8. Water Management Bureau; 9. Party of the village

Any of these that give at least some voice to water users have been considered participative for the purposes of this study. These include all but numbers 1, 8 and 9 in the list above. The non- participation category also includes non-response by the WUA. Table 2 provides the number of WUAs that take a decision partcipatively for each type of decision considered in the research.

19 Table 2: Number of WUAs that take decisions participatively

Decision Particpative Non- participative

Irrigation order 27 8

Water price 11 24

Volunteer labour 15 20

Tertiary canal maintenance 27 8

Rules regarding groundwater use 24 11

Assignment of guards overseeing water the water allocation process 8 27

Sanctions against members breaking the rules 26 9

The indicator for participation that has been used in the regression is a count measure indicating for each WUA how many decisions are taken participatively. The higher this number, the higher the participation, and – according to the theory – the better the performance of the WUA. Table 3 breaks down the WUA into the ones with low and high participation in order to compare their performance. Some trends are apparent from the comparison; Investment is higher in participative WUAs, while secondary canal quality is lower. These trends will be further examined in section 5.2.

20 Table 3: WUA performance and participation

Participation Participation <4 ≥ 4

Number of obs. 16 19

Water per Mu 2.42 2.24

Investment Density 77.14 124.97

Secondary Canal Quality 2.02 1.63

Tertiary Canal Quality 1.13 1.17

Finance Ratio 1.70 1.63

Other characteristics Several other variables that typify WUAs have been employed as well. While these are not expected to be unimportant, they do not relate directly to participation, and are thus of less interest to the hypothesis that participatory institutions perform better. They include characteristics of the leader (e.g. age and education), as well as the number of farmer groups that comprises a WUA and the number of years that have passed since the establishment of the association. These are explained in table 4.

5.1.3. Village Characteristics In order to control for factors that lie outside WUA control, a number of variables have been included. Regional leaders might pass extra resources to the village in which they were born. In order to account for this a dummy variable has been included. Also, a village close to the county capital might be advantaged, because it is more visible to high officials from the water management bureau. Agro- ecological zone (dry, medium, or wet) might affect water management, as might the location of the village with respect to the main canal. Table 4 lists all these variables and the ones mentioned above, and provides brief summaries of how they have been measured.

21 Table 4: Variable specification for the first stage of the regression (at village level)

Variable Type Definition Expected Effect

Independent Variables

Investment Density Continuous Money spent maintenance, construction, and upgrading of canals by farmers and government divided by total length of secondary canals plus tertiary canals

Secondary canal Continuous Ratio of improved canals. quality

Tertiary canal quality Continuous See secondary canal quality.

Amount of water per Continuous The amount of water delivered per mu of Mu agricultural land, per rotation, divided by the standard amount of the agro-ecological zone.

Financial Ratio Continuous (Total income – total expenditure)/total income (All in RMB)

WUA Characteristics

Leader Age Continuous Age in years of the WUA leader +

Leader Education Continuous Years of education of the leader +

Decision-making Discrete Number of decisions taken in a participative + Participation way.

Democratically Dummy Democratically elected leader=1, otherwise 0. + elected leader

WUA age Continuous Number of years since the establishment of the + WUA.

WUA size Continuous Number of farmer groups in the WUA. ?

Village Characterisitcs

Leader born Dummy 1 = Important regional leader was born in the + village.

Distance from main Continuous Number of meters from main canal to village. - canal

Distance from county Continuous Number of km from village to the county - capital capital, Hongshui.

22 Zone 2 Dummy 1 = The village is located in the middle zone. ?

Zone3 Dummy 1 = The village is located in the highest (and ? wettest) zone.

5.2. Results This section will present the results from the analyses discussed above. A more in depth discussion of the implications of the findings will be given in chapter 7. Table 5 presents the findings from the first stage of the research. Each column of the table represents a different regression. The variables in the top row are the independent variables; the different indicators used for WUA performance.

Table 5: Results from the village level regression.

Secondary Tertiary Investment Canal Canal Finance Density Quality Quality Water Per mu Ratio WUA Characteristics Decision -3.02 -0.11 ** 0.00 -0.04 -0.11 Participation -(.1) -(2.73) (.14) -(.26) -(.54) Democratic 277.18 ** 0.23 * 0.08 -0.61 -1.06 Elections (2.19) (1.84) (.42) -(.89) -(1.61) Leader Age 3.94 0.00 0.00 -0.02 0.02 (.74) -(.54) -(.04) -(.49) (.26) Leader Education 7.13 0.03 0.04 * -0.02 -0.14 (.51) (1.36) (1.8) -(.23) -(.87) WUA Age 12.94 0.01 0.08 ** 0.11 -0.10 (.92) (.57) (2.26) (.88) -(.5) WUA Size -6.35 0.00 0.01 0.02 0.01 -(.79) (.44) (.36) (.5) (.11) Village Characteristics Leader Born 170.64 0.43 *** 0.09 0.84 -1.56 (1.38) (4.2) (.37) (1.15) -(1.13) Distance Main -2.13 0.00 ** 0.00 0.03 *** 0.02 ** Canal -(1.66) -(2.8) (.59) (3.94) (2.72) Distance County -2.11 0.00 0.00 0.00 -0.06 Capital -(.56) -(1.01) (.24) -(.11) -(1.44) Zone 2 39.53 -0.16 -0.11 0.93 0.39 (.4) -(1.28) -(.42) (1.67) (.4) Zone 3 -58.94 -0.61 *** -0.38 1.43 0.48 -(.55) -(3.51) -(1.25) (1.55) (.29) Regression Statistics R2 0.37 0.63 0.44 0.50 0.26 n 35 30 33 35 35 Note: T-values are given in parentheses. * Indicates a variable significant at the 10% level, ** at the 5% and *** at the 1% level.

23 One of the main trends one would have expected to see in this table would be that more participative institutions perform better. The two variables chosen to represent participation – democratically elected leaders and decision-making participation – indeed seem to have a significant impact on WUA performance. However, this impact is not uniformly positive.

Having democratically elected leaders is expected to lead to an increase in performance. After all, the fact that a successful leader is more likely to be re-elected forms an incentive for the leader to make sure that the association is running smoothly. The results support this hypothesis; in villages that elect their WUA leader democratically, secondary canal quality is higher and more investments in the quality of the water infrastructure are made.

However, contrary to expectations, increasing decision making power of association members does not appear to lead to an increase in performance. On the contrary, the more decisions farmers are involved in, the lower secondary canal quality is. Two explanations for this effect can be put forward: firstly, farmers are unwilling to invest in secondary canals, because benefits of good-quality canals can not be fully captured by the individual farmers. This would lead participative institutions to under- invest, as each farmer would rather have the others invest. A second explanation would be more positive about the attitudes of the farmers towards collective action: in villages with poor quality canals, farmers are in a stronger position to demand more say in the decisions taken. We would be seeing reversed causality here. The fact that decision making participation does not have a significant negative impact on investment might support this position. However, this argument does reveal one problematic aspect of canal quality as an indicator for WUA performance. Canals predate the establishment of water user associations in the research area. As such, canal quality depends to a large extent on past investments. This would suggest that WUA characteristics are of limited importance to the physical health of the infrastructure.

Of the WUA characteristics not directly related to participation, both leader education and WUA age have an effect on tertiary canal quality. Especially the fact that WUA age positively affects performance is an indication that the reforms initiated in Minle are beneficial, and that some of the benefits take some time to materialize.

When considering the village characteristics – which were added to the regressions as controls – three variables are found to have a significant effect on performance. A regional leader being born in the village has a positive effect on secondary canal quality; the distance from the main canal matters for the the secondary canal quality, amount of water delivered and the financial situation of the WUA; and finally the zone in which the village is located matters for canal quality.

The effect of regional leaders indicates some nepotism. Villages receive assistance in the construction and upgrading of canals from higher governments. Officials in these higher governments are more likely to channel these funds to their native villages, than to other villages.

The distance from the main canal seems to have three effects. That secondary canal quality is lower, which can be explained by the fact that the canals are longer, and thus more costly to upgrade. The fact that villages located far away from the main canal are able to supply more water than those close to the canals is explained by the way the variables are defined. The amount of water supplied is the amount the WUA obtains from the WMB, at the branch of the secondary canal from the primary canal. As the WMB compensates the WUA for conveyance losses in the stretch between the primary canal and the village, villages located far from the canal receive more water without having to closely manage their water. The effect on financial situation is more mysterious. One would expect such an effect to be negative, if anything, due to the extra expenses more canals bring with them.

24 6. The impact of WUA performance The question this chapter aims to answer is: “What is the impact of WUA performance on household income?”The structure of this chapter largely follows that of the previous one: First the variables used in the analysis will be discussed, and then the results will be presented. However, the level of analysis is not at the village level in this chapter, but at the household level.

6.1. Variables Three reduced form equations will be estimated in order to examine the importance of WUA performance relative to the other determinants of farm income, off-farm income and total income. While the link between irrigation management and farm income is fairly obvious, the link between irrigation management and off-farm income might not be. However, a properly functioning irrigation system might have two opposing effects on the off-farm income of a household. Firstly, the reduced labour needs for irrigation-related tasks that canal improvements effect, might increase opportunities for off-farm labour. However, a second effect is that if irrigation is managed well, it reduces the uncertainties associated with farming, and thus leads to households committing more resources to farm activities.

The variables for WUA performance used in the equations are the same as the dependent variables in the first stage. However, for the set of regressions that make up the second stage of the research, they will serve as independent variables. These variables are not the only factors that influence household income. Three studies that analyze irrigated agriculture and its impacts on farm household income serve as a guide here as to what variables to select: Estudillo et al. (2001), Huang et al. (2006), and Kamara et al. (2002). These three studies were selected because the goals were similar to ones of the study presented here. They were not selected to be representative for the whole literature on income functions.

The variables selected fall into four categories: household characteristics – covering aspects of the household members, such as age and education, farm characteristics – which cover aspects such as land holdings and soil fertility, village characteristics – as defined in Chapter 5, and finally WUA performance. Note that prices of inputs and outputs are not included in the analysis. This is a cross- sectional analysis, and prices are not expected to vary widely in the research area, so their effects are held constant without adding them as a variable.

Firstly farm income is likely to be determined by the amount of labour available and the skill of the head of the household. As farmer skill is hard to measure, education is taken as a proxy. While both are not found to have a significant effect on farm income in any of the three studies, both were found to have a significant effect on off-farm income by Estudillo et al. (2001).

A second type of factors influencing income of farm households pertains to the farm holdings of the household. The factor in this category that is expected to have the greatest influence is the amount of land held by the households. Other factors that are considered relate to the of the quality of the land (e.g. its fertility), an aspect on which Huang et al. (2006) focus.

The third category considered here does not relate to the household itself, but rather to the village of which it is part. Neither Estudillo et al. (2001) or Kamara et al. (2002). mention these in their analysis, and Huang et al. (2006) merely mention them before explaining that it is unnecessary for them to consider them due to their methodology. For this study it is however important to consider the effects of factors that are determined at village level – such as market access – because they coincide with the actual variables of interest: WUA performance. Consider the case in which a well performing WUA is located in a village that lacks proper market access, which has a detrimental effect on income. Should

25 the WUA performance be the only variable at village level, the coefficients for WUA performance would have downward bias.

The resulting set of equations is:

Y = c + βhc’HC + βfc’FC + βvc’VC + βwp’WP

Where: Y is a vector of income: On-farm, off-farm and total. HC, FC and VC are matrices of household, farm and village characteristics respectively; WP is a matrix of WUA performance indicators; β-terms are matrices of the parameters to be estimated.

Income consists of two components: on-farm and off-farm income. The number of household in the sample that have no or negative income from farming is so small as to be negligible (n=16). Things are different for off-farm income. Not every household in the sample is engaged in the off-farm labour market. And no household make a loss at their off-farm activities. Only for the households that do engage in the labour market have effects been estimated, using a truncated regression.

To both make regression estimations consistent with a Cobb-Douglas production function and mitigate the effects of potential outliers, the natural logs of the variables on both sides of the equation will be taken.

6.1.1. Income Even though most of the households in the research area can be typified as farm households, many also engage in income generating activities that are not directly related to farming activities in the region. As Table 6 indicates, a significant portion of the income in Minle County is derived from off- farm sources. While irrigation management is likely to affect farming income, it could also indirectly affect off-farm income. Improved irrigation canals could, for example, lead to lower labour requirements for irrigation, and thus free up labour for off-farm activities. These two forms of income are therefore both analyzed.

Table 6: Mean Incomes in Minle County

Mean Std. Dev. Min Max

Cropping Income 9018 11138 -5804 96904

Off-farm Income 8310 15739 0 210000

For farm income, only cropping income has been considered. While there is some livestock raising, this is likely to be affected differently by irrigation than cropping. Livestock is more dependent on large areas of unirrigated land, and thus of less interest here. The produce used for own consumption has been valued at market prices and added to income.

Off-farm income comprises three sources: labour or business outside the farm (but in the region), income from migrant labour – many women go to Xinjiang in the far west of China to pick cotton every year – and remittances from family members living elsewhere in China.

26 6.1.2. Household Characteristics The composition of the household matters greatly to the ability of the household to generate an income. Aspects that are included are: the total members in the family, the ratio of men to women, and the age of the household members. To account for the farming ability of the household, the age and education of the household head have been included.

6.1.3. Farm Characteristics Not only the demographic characteristics of the household members matter, their assets also need to be taken into account, which are grouped under the header of farm characteristics here. The amount of land in mu, how much of their land is sloped, the fertility of the land, the irrigation status of the land, the amount of capital and livestock are all included.

27 Table 7: Variable specification for the household level regression

Variable Type Definition Expected Effect

Independent Variables

Farming Income Natural log Income generated from cropping

Off-farm Income Natural log Income generated from off-farm activities

Total Income Natural log Off-farm plus farm income

Household Characteristics

Labour Natural log Number of persons in the household +

Gender Ratio Ratio of men in the household +

Age1 Ratio Ratio of household members between the age + of 16-55

Age2 Ratio Ratio of household members aged over 55 +

Education Head Natural log Years of education of the head of the + household

Age head Natural log Age of the head of the household +

Farm Characteristics

Farm Size Natural log The area in mu planted with crops +

Irrigation Status Ratio Ratio of land irrigated +

Slope Ratio Ratio of land on slope -

Fertility Continuous Average fertility of the land: 3 means bad + quality, 1 means good.

28 6.2. Results In this section, the results from the regression at household level will be presented. They will be briefly discussed, but a more thorough analysis of the implications of the results with respect to policy recommendations will be given in chapter 7.

Judging from Table 8, the main determinants of income seem clear, and are as expected: more land, capital, and irrigation rights lead to more farm income, while labour is the main determinant of off- farm income.

The variables of interest in the table are the WUA Performance indicators. A positive impact on farm income was expected. After all, the increased security improved water management offers is expected to improve allocative efficiency on farms and therefore increase cropping incomes. There is some evidence for this. The indicators for investment and financial performance have a positive impact on income. Not all indicators were found to have a positive effect: no impact of canal quality was observed. The amount of water provided per mu even carries a negative sign. It is possible that both these effects are due to the fact that water allocation is often intended to compensate for adversarial circumstances. In drier, riskier areas, more water per mu is provided. In villages that are located farther from main canals, canal quality is often higher (see table 5).

As for off-farm income, the indicators for investment and secondary canal quality are associated with negative coefficients. This could indicate that farm households in better performing associations are willing to rely more on their farming activities, and thus seek less off-farm employment. In such a case one would expect farm income to be higher, after all, the household would only shift away labour from off-farm activities in favour of farm activities if it were compensated by means of extra income. As observed above there is some evidence for this effect of WUA performance on farm income.

When one adds up cropping and off-farm income, and runs the regression, the effects are as in the third column of table 8. Very little positive impact of the WUA performance indicator remains visible. This might be due to the fact that all data for the research comes from a wet year. The direct benefits from improved WUA performance, more water that is timed better, are not very large: after all there is more than enough water to make up for conveyance losses in poorly maintained canals.

29 Table 8: Results for the regressions at household level

Off-Farm Farm Income Income Total Income Household Charactheristics Labour -0.04 0.99 *** 0.35 ** -(.29) (2.84) (2.1) Gender -0.09 -0.02 0.23 -(.3) -(.04) (.6) Age1 -0.02 0.91 ** 0.26 -(.06) (2.) (.9) Age2 0.01 -0.02 0.04 (.03) -(.04) (.12) Age Head 0.00 0.00 0.00 (.19) -(.11) (.47) Education Head 0.01 0.02 0.02 (.44) (1.06) (1.24) Farm Characteristics Land 1.07 *** -0.21 0.58 *** (12.78) -(1.26) (4.82) Irrigation Status 0.12 ** -0.31 0.10 (2.52) -(.98) (.93) Slope -0.14 -0.52 -0.15 -(.88) -(1.51) -(1.) Fertility 0.05 0.12 0.12 (.59) (.94) (1.64) Capital 0.05 ** 0.05 0.03 (2.21) (.98) (1.04) Village Characteristics Distance county -0.06 *** 0.00 -0.05 capital -(2.91) -(.06) -(1.58) Distance main 0.00 0.03 0.00 canal -(.14) (.81) -(.13) Zone2 0.15 0.66 *** 0.30 ** (1.14) (3.02) (2.07) Zone3 0.38 ** 0.11 0.20 (2.21) (.36) (1.06) WUA Performance Water Per mu -0.28 *** 0.34 -0.01 -(2.77) (1.49) -(.07) Investment 0.03 * -0.08 ** 0.00 Density (1.95) -(2.19) -(.14) Secondary Canal -0.18 -0.53 * -0.23 Quality -(1.27) -(1.87) -(1.44) Tertiary Canal 0.11 -0.40 -0.09 Quality (.72) -(1.34) -(.46) Finance Ratio 0.05 ** 0.01 0.04 * (2.28) (.24) (1.81) Constant Term 5.85 *** 6.55 *** 6.59 *** (12.22) (6.37) (8.89) Regression Statistics R2 0.54 0.10 0.31 n 301 258 312

Note: T-values are given in parentheses. * Indicates a variable significant at the 10% level, ** at the 5% and *** at the 1% level. 30 7. Conclusions and recommendations This study has combined elements from irrigation literature and economics literature to explore the relationships between irrigation and the rural economy of Minle County, China. The integration of these two disciplines has allowed this study to look beyond a technocratic view of irrigation merely being an input to production, but rather as an important element within the local institutional and economic system. This section will discuss the results from this analysis. Firstly the conclusions from the analysis provided in Chapters 5. and 6. are given. These conclusions imply some policy recommendations, which are given in the subsequent section. The final section will discuss the limitations of this study, and provide recommendations for further research.

7.1. Conclusions The questions this study set out to answer were: 1. What are crucial factors for judging WUA performance, both in terms of financial efficiency (how much money does a WUA add to the basic cost price of water), and in terms of average water use per hectare? 2. Which factors explain differences in WUA performance? 3. How does the performance of the WUA translate at farm household level? I.e. what effect does a better performing water institution have on the water use of a farmer, and farm household income, holding all other factors constant.

The first question has been answered by means of literature study. Irrigation is not only employed as a tool for water management, but also for rural development. When judging the performance of irrigation institutions this should be reflected in the indicators chosen. This study follows the recommendations of Saleth et al., (1999b). Crucial indicators are physical, financial, economic, and equity considerations. Physical indicators relate to the health of the infrastructure and to the amount of water delivered. Financial indicators are used to assess the financial health of the institution. Economic indicators compare the irrigation sector to other water using sectors, and assess whether it might be more efficient to assign water to other uses. Equity indicators are used to assess whether the allocation is fair, or certain groups are advantaged over others. Unfortunately, due to data limitations this study has only focused on the first two types of factors. The challenges facing the water sector in Minle County mainly pertain to these factors, so for the purposes of this research this in not problematic.

As for question two, it is a widely held believe that the level of user participation is a key determinant of performance. The results as presented in chapter 5. do not unambiguously support this position, It has been shown that increasing participation by having democratically elected leaders has a positive effect on WUA performance, by increasing investments and improving canal quality. However, increasing direct decision making participation did not show a positive effect. Apart from participation, other factors were important as well, such as agro-ecological location, or being the home village of an important regional leader. Does this imply that participation is not as desirable as is often claimed? No. The clear positive effect of democratic elections demonstrate the potential of holding management accountable to association members. The fact that apart from increasing participation, being a home village of a regional leader leads to better water management performance, does not demonstrate the limitations of participation. It demonstrates that leaders at higher levels of water management should also be made more accountable. The fact that older WUAs have better tertiary canals demonstrate that the reforms have not had their full effect in all villages yet, and that improvements are to be expected even without further reform.

Results at the household level show a positive impact of some indicators of WUA performance (financial health of the WUA and investment density) on household farm income, but also a negative

31 impact of an other, the amount of water supplied per mu. As noted in chapter 6, this negative impact probably has more to do with compensating for harsh circumstances than with more water leading to lower incomes. So it can be said that WUA performance is positively correlated with cropping income. On the other hand WUA performance is found to be negatively correlated with off-farm income. This is explained by the fact that increased WUA performance leads to higher farm productivity, and thus form a disincentive for engaging in off-farm employment. One the whole, one would then expect WUA performance to have a positive impact on the combination of off-farm and cropping income. This effect was not found. This is possibly explained by the fact that the data used for this study stems from one year (2007), which happened to be a year characterized by particularly abundant rainfall. In such circumstances a well functioning irrigation sector is of less importance, because the more abundant water is, the less important economizing on it becomes.

7.2. Policy recommendations Current reforms are a step in the right direction. However, the implementation of the concept of WUAs will not do anything to change conflicts of interests within the chain of water management. A good example is the decision to open a potato processing plant, and accompanying ambitions to expand the area planted with potatoes. For this policy, potatoes are given priority over all other crops in assigning water rights. This means that in a dry year, when there is not enough water for all fields, the fields planted with potatoes are guaranteed of water. Officials at the county water management bureau insisted that potatoes are a very water efficient crop, a statement denied by every other actor interviewed (village officials and farmers). In an environment where officials have such diverging views, giving farmers some more say might not have a large effect. Village-level decisions are embedded within the greater contexts of the county and catchment levels. If allocation decisions are made in a poor manner at these high levels, and these poor decisions can go unchallenged, village level decisions can go only so far to positively affect farmer income. Another constraint on the extent to which village level decisions can make an impact is the fact that regional leaders seem to favour their village of birth.

One aspect not taken into account in this study has been the cost of reforms. Williamson (1981) noted that a smoothly operating system is not desired for its own sake. If the costs of achieving a smoother operation are higher than its benefits, there’s little reason to engage in reforms. Firstly this implies that even if reforms are to have a positive impact on livelihoods, this impact needs to be compared to the impact that could have been made if investments in reforms were invested elsewhere (e.g. better infrastructure or education). Secondly it implies that whatever the outcomes here, they are hard to generalize, as transaction costs and benefits differ in each setting. It might be more costly to organize farmers in other settings, increasing the transaction costs of WUAs, or the irrigation bureaucracy might be functioning very efficiently, decreasing its transaction costs. This would make the switch from a centrally planned system to a decentralized system less economical. Likewise, the benefits of increased irrigation efficiency are likely to be smaller in situations when water is less scarce.

One aspect to be considered is that participative institutions are not merely desirable for their efficiency benefits, although this is intended to be their main benefit. Participatory institutions are also better at coping with dynamic circumstances. Should circumstances and thus transaction costs change, rules and regulations governing water allocation should change as well. These benefits would only manifest themselves over a long time span, and therefore it has been impossible to include them in this research, which has only focused on one year.

32 7.3. Discussion There exists plenty of room to expand the current approach. The research has been limited by the fact that data was only available for one year. Having extended data would improve the research in a number of ways. It would allow for a more detailed analysis of the effects of investments in infrastructure. Investments take time to have an effect. In the current research it is only observed whether investments are being made, not if they have any effect. Moreover, performing the research over multiple years would allow the comparison between wet and dry years. Dry years put greater demands on water management than wet years, so it is hypothesized that the effects of water sector reform are more pronounced in dry years.

Also, the focus has been restricted to the village level, but what happens within irrigation districts and at the county water management bureau is very likely to have a profound effect on what happens “on the ground”. This would require the expansion of the framework to include indicators for water management bureau and regional government performance. It would also require the expansion of the scale of the study to include villages from other counties.

Another crucial aspect that future studies should take into account is a more objective measurement. Much of the information about water and investment flows has been provided by WUAs themselves, without the means to verify these. This is problematic, as WUA officials might not have an incentive to divulge all information they have. If some of this information can be obtained by objective measurement, it would allow for a means of verifying the data.

Regarding the measures for performance, these can be expanded upon. Equity consinderation were no part of this research, nor was the economic efficiency of the irrigation sector vis a vis other sectors. These are important considerations when assessing WUA performance, and while there were reasons why it was allowable not to include these, it would have been better if the possibilities has existed to do so. Another issue has to do with the indicator called investment density. While investment sounds like it is always a good thing, more investment is not necessarily an indications of a well performing WUA. Poor quality earthen canals require more maintenance than canals lined with concrete. Combined with the fact discussed above that the benefits of investments in the upgrading of canals materialize over the course of many years, where only one year has been considered here, it is clear that the use of investment data in this study is not without issues. However, there simply was no data available about how the WUAs did their work, nor was there data for multiple years. So for this research the assumption that investment is always a good thing is a necessary one. And given the fact that canal quality is also considered in the research it might not even be a problematic one.

33 References Alcamo, J. & Henrichs, T. (2002). Critical regions: A model-based estimation of world water resources sensitive to global changes. Aquatic Sciences - Research Across Boundaries, 64(4) 352-362. Allen, D. & Lueck, D. (2002). The nature of the farm, contracts, risk and organization in agriculture. The MIT press, Cambridge. Becker, D. & Ostrom, E. (1995). Human Ecology and Resource Sustainability: The Importance of Institutional Diversity. Annual Review of Ecology and Systematics, 26() 113-133. Bos, M. G., Murray-Rust, D. H., Merrey, D. J., Johnson, H. G. & Snellen, W.B. (1993). Methodologies for assessing performance of irrigation and drainage management. Irrigation and Drainage Systems, 7(4) 231-261. Bromley, D. W. (1982). Land and Water Problems: An Institutional Perspective. American Journal of Agricultural Economics, 64(5) 834-844. Carruthers, I., Rosegrant, M. W. & Seckler, D. (1997). Irrigation and food security in the 21st century. Irrigation and Drainage Systems, 11(2) 83-101. Coase, R. H. (1937). The Nature of the Firm. Economica, 4(16) 386-405. Coase, R. H. (1960). The problem of social cost. The journal of Law and Economics, 3(1) 1. Demsetz, H. (1968). The cost of transacting. Quarterly Journal of Economics, 82() 33-53. Dinar, A., Rosegrant, M. & Meinzen-Dick, R. (1997). Water allocation mechanisms, principles and examples. Policy research working paper 1779, World Bank and IFPRI. Estudillo, J. P., Quisumbing, A. R. & Otsuka, K. (2001). Income distribution in rice-growing villages during the post-Green Revolution periods: the Philippine case, 1985 and 1998. Agricultural Economics, 25(1) 71-84. Howe, C. W., Schurmeier, D. R. & Shaw, W.D. (1986). Innovative Approaches to Water Allocation: The Potential for Water Markets,. Water Resources Research, 22(4) 439–445. Hu, J., Wang, S. & Yeh, F. (2006). Total-factor water efficiency of regions in China. Resources Policy, 31(4) 217-230. Huang, Q., Rozelle, S., Lohmar, B., Huang, J. & Wang, J. (2006). Irrigation, agricultural performance and poverty reduction in China. Food Policy, 31(1) 30-52. Kamara, A. B., van Koppen, B. & Magingxa, L. (2002). Economic viability of small-scale irrigation systems in the context of state withdrawal: the Arabie Scheme in the Northern Province of South Africa. Physics and Chemistry of the Earth, Parts A/B/C, 27(11-22) 815-823. Li, X., Lu, L., Cheng, G. & Xiao, H. (2001). Quantifying landscape structure of the Heihe River Basin, north-west China using FRAGSTATS. Journal of Arid Environments, 48(4) 521-535. Lin, Z. (2003). Water User Association Development in China: Participatory Management Practice under Bank-Supported Projects and Beyond. Social Develepment Note 83, World Bank, Washington. Liu, C. & Zhang, S. (2002). Drying up of the yellow river: its impacts and counter-measures. Mitigation and Adaptation Strategies for Global Change, 7(3) 203-214. Lohmar, B., Wang, J., Rozelle, S., Huang, J. & Dawe, D. (2003). China's Agricultural Water Policy Reforms: Increasing Investment, Resolving Conflicts, and Revising Incentives. Agriculture Information Bulletin 782, U,S, Department of Agriculture, Washington D.C. Meinzen-Dick, R. (1997). Farmer participation in irrigation – 20 years of experience and lessons for the future. Irrigation and Drainage Systems, 11(2) 103-118. Molden, D. J., Sakthivadivel, R., Perry, C. J. & Fraiture, C.D. (1998). Indicators for Comparing Performance of Irrigated Agricultural Systems. Reserach Report 20, IWMI, Colombo. Mollinga, P. P. (2001). Water and politics: levels, rational choice and South Indian canal irrigation. Futures, 33(8-9) 733-752.

34 Moore, M. (1989). The fruits and fallacies of neoliberalism: The case of irrigation policy. World Development, 17(11) 1733-1750. Nickum, J. E. (1998). Is China Living on the Water Margin?. The China Quarterly, 156() 880-898. Nickum, J. E. (2005). Uphill Flow of Reform in China’s Irrigation Districts. In Chennat Gopalakrishnan, Cecilia Tortajada & Asit K. Biswas (Eds.), Water Institutions: Policies, Performance and Prospects,Springer, Berlin Heidelberg. Saleth, R. & Dinar, A. (1999a). Water challenge and institutional response, a cross-country perspective. Working Paper 2045, World Bank, Washington. Saleth, R. & Dinar, A. (1999b). Evaluating water institutions and water sector performance. World Bank Technical Paper 447, World Bank, Washington. Saleth, R. & Dinar, A. (2004). The institutional economics of water : a cross-country analysis of institutions and performance. Edward Elgar, Cheltenham, UK. Seckler, D., Molden, D. & Sakthivadive, R. (2003). The Concept of Efficiency in Water Resources Management and Policy. In J. W. Kijne, R. Barker & D. Molden (Eds.), Water Productivity in Agriculture: Limits and Opportunities for Improvement,CABI Publishing, Wallingford, UK. Shalizi, Z. (2006). Addressing China's growing water shortages and associated social and environmental consequences. Policy Research Working Paper 3895, World Bank, Washington. Small, L. E. & Svendsen, M. (1990). A framework for assessing irrigation performance. Irrigation and Drainage Systems, 4(4) 283-312. Subramanian, A. (1997). User organizations for sustainable water services. World Bank Technical Paper 354, World Bank, Washington. Turral, H. (1995). Devolution of management in public irrigation systems: Cost shedding, empowerment and performance. A review. ODI Working Paper 80, ODI, UK. Van Koppen, B., Parthasarathy, R. & Safiliou, C. (2002). Poverty Dimensions of Irrigation Management Transfer in Large-Scale Canal Irrigation in Andra Pradesh and Gujarat, India. Research Report 61, IWMI, Colombo. Varis, O. & Vakkilainen, P. (2001). China's 8 challenges to water resources management in the first quarter of the 21st Century. Geomorphology, 41(2-3) 93-104. Vermillion, D. (1997). Impacts of Irrigation Management Transfer: A Review of the Evidence. Research Report 11, IIMI, Colombo. Vermillion, D. (2004). Collective action and property rights for sustainable development: Irrigation, Collective Action, and Property Rights. In R. Meinzen-Dick & M. Di Gregorio (Eds.), Collective Action and Property Rights for Sustainable Development,IFPRI, Washington. Wang, J., Xu, Z., Huang, J. & Rozelle, S. (2005). Incentives in water management reform: assessing the effect on water use, production, and poverty in the Yellow River Basin. Environment and development economics, 10(6) 769-799. Williamson, O. E. (1981). The modern corporation: origins, evolution, attributes. Journal of Economic Literature, 19(4) 1537-1568. World Bank (2006), “China: Water quality management – policy and institutional considerations.” Washington, D.C.: World Bank. Yang, H. & Zehnder, A. (2001). China's regional water scarcity and implications for grain supply and trade. Environment and Planning A, 33(1) 79-96. Yang, H., Zhang, X. & Zehnder, A.J.B. (2003). Water scarcity, pricing mechanism and institutional reform in northern China irrigated agriculture. Agricultural Water Management, 61(2) 143- 161. Young, R. (1986). Why Are There so Few Transactions among Water Users?. American Journal of Agricultural Economics, 68(5) 1143-1151. Zhang, J. (2007). Barriers to water markets in the Heihe River basin in northwest China. Agricultural Water Management, 87(1) 32-40. Zhang, J., Zhang, F., Zhang, L. & Wang, W. (2009). Transaction Costs in Water Markets in the Heihe River Basin in Northwest China. International Journal of Water Resources Development, 25(1) 95 - 105.

35