4. Social impacts

4.1 Poverty impact pathways Agricultural research could also benefit the poor in less direct ways. Growth in adopt- The primary goal of agricultural research ing regions could create employment op- during the GR era was to increase food pro- portunities for migrant workers from other duction. Historically, this led to a focus on less dynamic regions. It could also stimulate foodgrains in high-potential areas where the growth in the rural and urban nonfarm quickest and highest returns to R&D could be economy with benefits for a wide range of expected. This strategy was extremely suc- rural and urban poor people. Research cessful in achieving its primary goal in South could lead to lower food prices for all types Asia. Additionally, it helped cut poverty in of poor people. It could also improve their the region during the 1970s and 1980s – access to foods that are high in nutrients from 59.1% of the population in 1975 to and crucial to their well-being – particularly 43.1% in the early 1990s (Rosegrant and that of poor women. Hazell, 2000). But it did not eliminate poverty or malnutrition, and today, despite However, agricultural research could also the fact that most South Asian countries now work against the poor. Some technologies have plentiful national food supplies, are more suited to larger farms, and some poverty is still a major problem. About 450 input-intensive technologies that are in million South Asians currently live below the principle scale-neutral may nevertheless US$1 per day poverty line (about the same as favor large farms because of their better in 1975), and 80% of these are rural and access to irrigation water, fertilizers, seeds, obtain at least part of their livelihood from and credit. Some technologies (e.g., mecha- agriculture and allied activities (World Bank, nization and herbicides) could displace 2007; Ahmed et al., 2008). The agricultural labor, leading to lower earnings for research systems have responded to this agricultural workers. By favoring some problem by targeting more of their research regions or farmers over others, technology towards the problems of small-scale farmers could harm non-adopting farmers by and the rural poor in the hopes of enhancing lowering their product prices even though its poverty-reducing impacts. only the adopting farmers benefit from cost reductions. Given the complex causes underlying poverty and the diversity of livelihoods Given that many of the rural poor are si- found among poor people, the relationship multaneously farmers, paid agricultural between agricultural research and poverty workers, net buyers of food, and earn alleviation is necessarily complex. There are nonfarm sources of income, the impacts of a number of pathways through which technological change on their poverty improved technologies could potentially status could be indeterminate, with house- benefit the poor (Hazell and Haddad, holds experiencing gains in some dimen- 2001). Within adopting regions, research sions and losses in others. For example, the could help poor farmers directly through same household might gain from reduced increased own-farm production, providing food prices and from higher nonfarm wage more food and nutrients for their own con- earnings, but lose from lower farm gate sumption and increasing the output of prices and agricultural wages. Measuring marketed products for greater farm net benefits to the poor requires a full income. Small-scale farmers and landless household income analysis of direct and laborers could gain additional agricultural indirect impacts, as well as consideration of employment opportunities and higher the impacts on poor households that are wages within adopting regions. Research not engaged in agriculture and/or who live could also empower the poor by increasing outside adopting regions. Much of the con- their access to decision-making processes, troversy that exists in the literature about enhancing their capacity for collective how R&D impacts on the poor has arisen action, and reducing their vulnerability to because too many studies have taken only a economic shocks via asset accumulation. partial view of the problem.

An Assessment of the Impact of Agricultural Research in South Asia since the — 

SC_south_asia_final.indd 17 1/8/08 15:28:22 There is a large literature on the impacts of gaining proportionally more income than agricultural research on the poor in South larger farmers, resulting in a net improve- Asia but hardly any impact studies exist ment in the distribution of village income that quantify the research costs of reducing (e.g., Hazell and Ramasamy, 1991; poverty. Many studies focus on assessing Maheshwari, 1998; Thapa et al., 1992). changes in income distribution or poverty in areas where new technologies have Freebairn (1995) reviewed 307 published been adopted, but only a few attempt to studies on the GR and performed a meta- link changes in inequity or poverty to analysis. Nearly all the studies that he research expenditures. More recently, reviewed focused on changes in inequality measures of poverty have also been and income distribution rather than on expanded to include broader and less absolute poverty, the latter emerging as a quantifiable social impacts, such as empow- more important issue in the 1990s. Free- erment and changes in social capital. One bairn found that 40% of the studies he consequence is that if we focus only on reviewed reported that income became quantitative studies that evaluate the more concentrated within adopting impact of research investments, this section regions, 12% reported that it remained un- of the paper would be very short indeed changed or improved, and 48% offered no and would not do justice to the large conclusion. He also found there were more amount of research that has been done on favorable outcomes in the literature on poverty issues or to the large number of Asia than elsewhere, and that within the studies that shed useful light on how Asian literature, Asian authors gave more improved technologies can benefit the favorable conclusions than non-Asian poor at farm and community levels. Those authors. Later studies did not report more who invest in agricultural research need to favorable outcomes than earlier studies, know that relevant work has been under- thereby casting some doubt on the proposi- taken with proven poverty-reduction tion that small-scale farmers adopted, impacts in the field, even if we do not yet albeit later than large-scale farmers. have much quantitative evidence to show However, it should be noted that Free- which types of research give the best bairn’s analysis did not include repeat poverty impact per dollar invested. studies undertaken at the same sites over time, such as Hazell and Ramasamy (1991) and Jewitt and Baker (2007), both of whom 4.2 Evidence on impacts within found favorable longer-term impacts on in- adopting regions equality. Freebairn (1995) also found that micro-based case studies reported the most The initial experience with the GR in Asia favorable outcomes, while macro-based stimulated a huge number of studies on essays reported the worst outcomes. how technological change affects poor farmers and landless workers within Walker (2000) argued that reducing in- adopting regions. A number of village and equality is not the same thing as reducing household studies conducted soon after the poverty, and that it may be much more dif- release of GR technologies raised concern ficult to achieve through agricultural R&D. that large farms were the main beneficia- More recent studies focusing directly on ries of the technology and poor farmers poverty confirm that improved technolo- were either unaffected or made worse-off. gies do impact favorably on many small- More recent evidence shows mixed scale farmers, but the gains for the smallest outcomes. Small-scale farmers did lag farms and landless agricultural workers can behind large farmers in adopting GR tech- be too small to raise them above poverty nologies, yet many of them eventually did thresholds (Hossain et al., 2007; Mendola, so. Many of these small-farm adopters 2007). However, the poor can benefit in benefited from increased production, other ways too. Hossain et al. (2007) found greater employment opportunities, and that in Bangladesh the spread of high- higher wages in the agricultural and yielding variety (HYV) rice helped reduce nonfarm sectors (Lipton with Longhurst, the vulnerability of the poor by stabilizing 1989). In some cases, small-scale farmers employment earnings, reducing food and landless laborers actually ended up prices and their seasonal fluctuations, and

18 — An Assessment of the Impact of Agricultural Research in South Asia since the Green Revolution

SC_south_asia_final.indd 18 1/8/08 15:28:22 enhancing their ability to cope with natural 4.3 Evidence on economy and disasters. In India, Bantilan and Padmaja sector-wide impacts (2008) found that the spread of ICRISAT’s groundnut improvement technology based There is a large econometric literature that on a RBF concept helped increase social net- uses cross-country or time-series data to working and collective action within estimate the relationship between agricul- adopting villages, and this proved especially tural productivity growth and poverty. These helpful to poor farmers and women in studies generally found high poverty reduc- accessing farm inputs, credit, and farm im- tion elasticities for agricultural productivity plements, and in sharing knowledge. Use of growth. Thirtle et al. (2003) estimated that participatory research methods in the selec- each 1% increase in crop productivity tion of improved rice varieties in Uttar reduces the number of poor people by Pradesh, India has been shown to empower 0.48% in Asia. For India, Ravallion and Datt women as decision-makers in their farming (1996) estimated that a 1% increase in agri- and family roles, as well as leading to cultural value added per hectare leads to a greater adoption of improved varieties 0.4% reduction in poverty in the short run (Paris et al., 2008). and 1.9% in the long run, the latter arising through the indirect effects of lower food The lessons from many past studies may prices and higher wages. Fan et al. (2000b) have less relevance today because of the estimated that each 1% increase in agricul- changing nature of the livelihoods of the tural production in India reduces the rural poor in south Asia. With rapid growth number of rural poor by 0.24%. For South in nonfarm opportunities in much of South Asia, these poverty elasticities are still much Asia as well as shrinking farm sizes, farming higher for agriculture than for other sectors and agricultural employment have become of the economy (World Bank, 2007; Hasan less important in the livelihood strategies of and Quibria, 2004). the rural poor (Nargis and Hossain, 2006; Kajisa and Palanichamy, 2006; Lanjouw and There is some evidence that the poverty Shariff, 2004). Within this new context, elasticity of agricultural growth may be di- many poor people with limited access to minishing because the rural poor are land gain more from nonfarm opportuni- becoming less dependent on agriculture. In ties than from productivity gains or wage , for example, agricultural growth earnings in farming, though investments in was associated with rapid reductions in education and access to capital are often rural poverty in the 1970s and 1980s, but crucial for accessing such opportunities the incidence of rural poverty hardly (World Bank, 2007; Nargis and Hossain, changed in the 1990s despite continuing 2006; Kajisa and Palanichamy, 2006; agricultural growth (Dorosh et al., 2003). Krishna, 2005). This is not to say that Dorosh, et al. (2003) show that this is partly publicly funded agricultural research because a growing share of the rural poor cannot still usefully be targeted to the households (46% by 2001–2002) had problems of poor part-time farmers. Hazell become disengaged from agriculture; even and Haddad (2001) identified several op- small farm households and landless agricul- portunities, including increasing the pro- tural worker households received about ductivity of food staples to free up land and half their income from nonfarm sources. labor for other activities, improving the nutrient content of staples, developing new Some of the studies reviewed in Section 3 technologies for small-scale home garden- that quantified the productivity impacts of ing of micronutrient-rich food, and using public investments in agricultural R&D also participatory research methods to enhance assessed the impacts on poverty reduction the relevance of improved technologies for and provide comparisons with other types poor farmers. But questions arise about the of public investment. Fan et al. (1999) efficacy of these kinds of interventions and found that agricultural R&D investments in whether they are cost effective in reducing India have not only given the highest pro- poverty compared to alternative types of ductivity returns in recent decades, but interventions. Answering these questions have also lifted more people out of poverty should be a priority for future impact per unit of expenditure than most other studies. types of public investment (Table 5).

An Assessment of the Impact of Agricultural Research in South Asia since the Green Revolution — 9

SC_south_asia_final.indd 19 1/8/08 15:28:23 Investments in agricultural R&D and rural Haddad, 2001). A question arises as to roads dominate all others in terms of the whether the power of these indirect size of their impacts, and can be considered benefits has diminished over time with the best win–win strategies for achieving market liberalization and greater diversifi- growth and poverty alleviation in India. cation of South Asian economies. Also, if unit production costs are not falling as in Fan et al. (2007) have used an econometric the past (as reflected in stagnating TFP model to estimate the impact of rice growth) this will also constrain future food research in India on poverty reduction, in- price reductions. This is an issue that cluding providing a breakout of an warrants further study. estimate of IRRI’s contribution. They found that about 5 million rural poor people have been lifted out of poverty each year as a 4.4 Evidence on inter-regional result of rice improvement research in disparities India. Using plausible attribution rules, they estimated that IRRI’s research contribution Agricultural development in South Asia has accounts for significant shares of these not benefited all regions equally; some of the annual reductions in the number of rural poorest regions that depend on rainfed agri- poor. In 1991, IRRI was attributed with culture were slow in benefiting from the GR raising 2.73 million rural poor people out of (Prahladachar, 1983). The widening income poverty, but because of the lag structures in gaps that resulted have been buffered to their model, the contribution declined over some extent by inter-regional migration. In time to only 0.56 million rural poor in 1999. India, the GR led to the seasonal migration of They calculated that the number of persons over a million agricultural workers each year lifted out of poverty for each US$1 million from the eastern states to and spent by IRRI declined from 59,040 in 1991 (Oberai and Singh, 1980; Westley, to 15,490 persons in 1999. This corresponds 1986). These numbers were tempered in later to an increase in the cost of raising each years as the GR technology eventually spilt person out of poverty from US$0.046 per over into eastern India in conjunction with day in 1991 to US$0.177 per day in 1999. the spread of tube wells. In a study of the impact of the GR in a sample of Asian Fan (2007) also estimated the impact of villages, David and Otsuka (1994) asked agricultural research on urban poverty in whether regional labor markets were able to India. He estimated that in 1970, accumu- spread the benefits between adopting and lated agricultural research investments non-adopting villages and found that lifted 1.2 million urban poor out of poverty, seasonal migration did go some way to fulfill- and this annual reduction increased to 1.7 ing that role. But while migration can buffer million by 1995. These numbers correspond widening income differentials between to between 2 and 2.5% of the remaining regions, it is rarely sufficient to avoid them. urban poor each year. On a cost basis, 196 In India, for example, regional inequalities urban poor were lifted out of poverty in widened during the GR era (Galwani et al., 1970 for each million rupees spent, and this 2007), and the incidence of poverty remains had declined to 72 urban poor per million high in many LFAs (Fan and Hazell, 2000). rupees by 1995. Since the same investment on research also lifted many rural poor out of poverty (see above), there is a double 4.5 Evidence on nutrition impacts dividend that makes research investments especially attractive for reducing poverty. Agricultural research has been very success- ful in increasing the supply of food and Lower food prices and growth linkages to reducing prices of food staples in South Asia. the nonfarm economy played a large role in Making food staples more available and less most of the results cited above, and these costly has proved an important way through benefit the urban as well as the rural poor. which poor people benefited from techno- These indirect impacts have sometimes logical change in agriculture (Rosegrant and proved more powerful and positive than Hazell 2000; Fan et al., 1999; Fan, 2007). the direct impacts of R&D on the poor Several micro-level studies from the GR era within adopting regions (Hazell and in South Asia found that higher yields

20 — An Assessment of the Impact of Agricultural Research in South Asia since the Green Revolution

SC_south_asia_final.indd 20 1/8/08 15:28:24 typically led to greater calorie and protein amounts on milk, meat, and other nutrient- intake among rural households within rich foods, the decline in the share of adopting regions. For example, Pinstrup- staples is also apparent among the poorest Andersen and Jaramillo (1991) found that 25% (J.V. Meenakshi, personal communica- the spread of HYV rice in North Arcot tion). However, since deficiencies in iron district, South India, led to substantial in- and the B vitamins are common among the creases over a 10-year period in the energy poor, the increases in micronutrient-rich and protein consumption of farmers and foods must not always have been high landless workers. Their analysis showed enough to offset the decline from cereals. that, after controlling for changes in Other micronutrient deficiencies exist (e.g., nonfarm sources of income and food prices, vitamins C and D), but these are not related about one-third of the calorie increase to reductions in cereal consumption. could be attributed to increased rice pro- duction. Ryan and Asokan (1977) also found Agricultural research has been directed at complementary net increases in protein and the problem of enhancing the nutritional calorie availability as a result of GR wheat quality of the diets of the poor. The main in the six major producing states of India, research strategies are: despite some reduction in the area of pulses grown. n Improvements in the productivity of fruits, vegetables, livestock, and fish, More aggregate analysis of the impacts of both in home gardens and ponds for on- rising incomes on diets and nutrient intake farm consumption and more generally to has proved more complex, particularly as increase the marketed supplies of these concern has shifted from calorie and protein nutrient-rich foods deficiencies to micronutrients and broader n Promotion of food-crop biodiversity, nutritional well-being. Food price declines especially traditional crops and cultivars are, in general, good for households that that are rich in nutrients purchase more food than they sell, as this n Biofortification of major food staples. amounts to an increase in their real income. Real income increases can be used to Ali and Hau’s (2001) assessment of AVRDC’s increase consumption of important staples program in Bangladesh showed significant and to purchase more diverse and nutrition- improvements in nutrition among partici- ally rich diets. However, a study of Bangla- pating farm families, as well as increased desh showed that a downward trend in the supplies and lower-priced vegetables in the price of rice between 1973–1975 and 1994– market. However, they also found that 1996 was accompanied by upward trends in while home gardens can increase incomes the real prices of others foods that are richer as well as improve nutritional intake, they in micronutrients, making these less acces- are not sufficient to improve nutrition to sible to the poor (Bouis, 2000). Similar desired levels and there is still need for nu- patterns were observed in India during the tritional education. After reviewing 30 agri- 1970s and 1980s when farmers diverted land cultural interventions (including six from away from pulses to wheat and rice, leading South Asia) to improve nutrition among to sharp increases in the price of pulses and participating families, Berti et al. (2004) also a drop in their per capita consumption conclude that interventions need to be (Kennedy and Bouis, 1993; Kataki, 2002). complemented by investments in nutrition education and health services and targeted Since then, there have been substantial in ways that empower women with addi- changes in food intake patterns in rural tional spending power. India. In particular, the share of cereals in total food expenditure has declined, while Biofortification research is relatively new that of milk, meat, vegetables, and fruits and, although the CGIAR and its national has increased. Per capita consumption of partners are working together on some cereals has also fallen in absolute terms aspects of this under the aegis of the (Nasurudeen et al., 2006). It is significant Harvest Plus Challenge Program (Bouis et that these substitutions occurred both al., 2000), it is rather early to measure any among the rich and the poor; not only do impacts, although one ex ante study has the top 25% spend relatively greater been completed (Meenakshi et al., 2007).

An Assessment of the Impact of Agricultural Research in South Asia since the Green Revolution — 21

SC_south_asia_final.indd 21 1/8/08 15:28:25 5. Environmental impacts

5.1 Environmental impact pathways cropped area, helping to save forest and other fragile lands from agricultural conver- Agricultural growth can impact on the envi- sion (Nelson and Maredia, 1999). But inten- ronment in many ways and it is helpful to sification often brings its own environmen- distinguish between the problems associ- tal problems. These include water ated with intensive irrigated and high-po- contamination with nitrates and phos- tential rainfed areas, where agricultural phates from fertilizers and manures, pesti- growth is largely of the land-intensification cide poisoning of people and wildlife, un- (yield-increasing) type, and the problems of sustainable extraction of irrigation water less-favored or less-developed areas, where from rivers and groundwater, and loss of agricultural growth is often of the expan- biodiversity within agriculture and at land- sionary (land-increasing) type. It should be scape levels (Santikarn Kaosa-Ard et al., noted, however, that the problems of the 2000; Pingali and Rosegrant, 2001). Intensi- two types can sometimes overlap. The fication pathways are associated with the drivers of change and the appropriate GR and arise mostly in irrigated and high- research and policy responses are quite dif- potential rainfed areas. ferent in these two environments (Hazell and Wood, 2008). Just how serious are the environmental problems associated with agriculture, and In LFAs, crop area expansion is often are they likely to undermine future produc- realized by reductions in the length of tion and South Asia’s ability to feed itself? fallows and by encroachment into forests Measuring environmental impacts of and fragile lands (e.g., steep hillsides and research and technological change is diffi- watershed protection areas), resulting in cult and as a result good empirical evidence land erosion, declining soil fertility, and loss is fragmentary, often subjective, and some- of biodiversity. Expansionary pathways in times in direct contradiction with the South Asia are typically associated with overall trends in agricultural productivity. areas of poor infrastructure and market The available evidence tells a mixed story. access, poverty, and population pressure. Some good news is that despite continued Agricultural intensification in high-poten- agricultural growth, the total forest area in tial areas helps avoid the kinds of problems South Asia has changed little since 1990 that prevail in many LFAs. By increasing (Table 11). Declines in , Pakistan and yields, it reduces pressure to expand the Sri Lanka have been offset by forest expan-

Table 11. Change in extent of forest and other wooded land (‘000 ha)

Forest Other wooded land Country 1990 2000 2005 1990 2000 2005

Bangladesh 882 884 871 44 53 58 Bhutan 3035 3141 3195 566 609 611 India 63,939 67,554 67,701 5894 4732 4110 Nepal 4817 3900 3636 1180 1753 1897 Pakistan 2527 2116 1902 1191 1323 1389 Sri Lanka 2350 2082 1933 0 0 0 Total 77,580 79,677 79,238 8875 8470 8065

Source: FAO (2005)

22 — An Assessment of the Impact of Agricultural Research in South Asia since the Green Revolution

SC_south_asia_final.indd 22 1/8/08 15:28:27 sion in India. There has, however, been a road construction, and urban and industrial 10% decline in the total area of other encroachment. Even where agriculture is woodland, including a 30% reduction in responsible, we need to separate out the India, which may be a better indicator of land degradation due to agricultural exten- the competition between tree cover and sification versus agricultural intensification. agricultural expansion, particularly in LFAs. It is also hard to reconcile some of these es- timates with the continuing growth in Less encouraging are several international average yields and land productivity across land-assessment exercises that have South Asia. While there are reports of reported widespread degradation of most hotspot areas where degradation is ad- types of agricultural land in South Asia. The versely affecting both the productivity and Global Land Assessment of Degradation sustainability of land, there must be large (GLASOD) mapping exercise of Oldeman et areas where agricultural productivity is not al. (1991) found that 43% of South Asia’s adversely affected and where the problems agricultural land was degraded to some are overstated. Some of the problem areas degree. Young (1993) subsequently revis- are intensively farmed irrigated areas, but ited these estimates using additional many are rainfed farming areas that, espe- national data, claiming the problem was cially in the Himalayas and semi-arid areas, actually more severe and that nearly three- are farmed more extensively. quarters of the agricultural land area was degraded to some extent, and that 40% More detailed data are available about the was moderately or severely degraded impact of irrigation on the waterlogging (Table 12). Degradation associated with ir- and salinization of irrigated land: rigation accounts for 23% of the total degraded area and for 25% of the moder- n About 4.2 million hectares of irrigated ately or severely degraded area. For India, lands (26% total) are affected by salini- Sehgal and Abrol (1994) estimated that zation in Pakistan (Ghassemi et al., 64% of the land area is degraded to some 1995). Chakravorty (1998) claims that extent, with 54% moderately to severely one-third of the irrigated area in degraded. Pakistan is subject to waterlogging and 14% is saline. Salinity retards plant Although these data provide a useful growth – he also claims agricultural out- warning, they do not tell us much about put is lower than it would otherwise be, the causes. Agriculture is only one contrib- by about 25%. uting factor; others include geological pro- n Dogra (1986) estimates that in India cesses (especially in the Himalayas), mining, nearly 4.5 million hectares of irrigated

Table 12. Extent of degradation of agricultural land in South Asiaa

% total that is % total that moderately or Type of degradation is degraded severely degraded

Water erosion 25 15 Wind erosion 18 13.9 Soil fertility decline 13 1.3 Waterlogging 2 1.5 Salinization 9 6.5 Lowering of water table 6 2.4 Total 73 40.6

a. Includes Afghanistan, Bangladesh, Bhutan, India, Iran, Nepal, Pakistan, and Sri Lanka. Source: Young (1993) as summarized by Scherr (1999)

An Assessment of the Impact of Agricultural Research in South Asia since the Green Revolution — 23

SC_south_asia_final.indd 23 1/8/08 15:28:28 land are affected by salinization and a 5.2 The R&D response further 6 million hectares by waterlog- ging; India had about 57 million hectares A growing awareness of these environmen- of net irrigated land in the late 1990s. tal problems has led to significant changes Umali (1993), quoted in Maredia and in agricultural R&D in South Asia since the Pingali (2001, p. 13) claims that 7 million early GR years. It has led to the entry of en- hectares of arable land has been aban- vironmentally oriented NGOs, some of doned in India because of excessive salt. whom have contested the GR approach and n In a random sample of 110 farmers from undertaken research and extension activi- four villages in , Joshi and ties of their own to broaden the spectrum Jha (1991) found a 50% decline in crop of technologies and farming practices avail- yields over 8 years due to salinization and able to farmers. The national and interna- waterlogging in irrigation systems. tional R&D systems have also invested heavily in NRM research and technologies Even more disconcerting for irrigated agri- and management practices for improving culture is the threat from the growing water, pest, and soil fertility management. scarcity of fresh water in much of South Asia. Many countries are approaching the One of the outcomes of greater NGO in- point where they can no longer afford to volvement has been a lively debate about allocate two-thirds or more of their fresh competing farming paradigms, and ‘alter- water supplies to agriculture (Comprehen- native’ farming has been offered as a more sive Assessment Secretariat, 2006). Most of sustainable and environmentally friendly the major river systems in South Asia are alternative to the modern-input based already fully exploited, and the massive ex- approach associated with the GR. The alter- pansion of tubewell irrigation in Bangla- native farming approach includes extremes desh, India, and Pakistan has led to serious that eschew use of any modern inputs as a overdrawing of groundwater and falling matter of principle (e.g., ), water tables. but also includes more eclectic whole- farming systems approaches such as low On the Indian subcontinent, groundwater external input farming (Tripp, 2006) and withdrawals have surged from less than 20 ecoagriculture (McNeely and Scherr, 2003). cubic kilometres to more than 250 cubic ki- Pretty (2008) provides a useful review of lometres per year since the 1950s (Shah et these approaches. al., 2003). More than a fifth of groundwa- ter aquifers are overexploited in the While the alternative farming literature Punjab, Haryana, Rajasthan, and Tamil provides many successful examples of agri- Nadu, and groundwater levels are falling cultural intensification, most of these have (World Bank, 2007; Postel, 1993). Even as arisen in rainfed farming systems that current water supplies are stretched, the largely missed out on the GR. We shall demands for industry, urban household use, review several of these experiences in and environmental purposes are growing Section 5.4 on LFAs. But by ‘sleight of ag- (Comprehensive gregation’, proponents of alternative agri- Assessment Secretariat, 2006; Rosegrant culture frequently mix these kinds of suc- and Hazell, 2000). It would seem that either cesses with much more modest results farmers must learn to use irrigation water obtained in GR areas, giving the impression more sparingly and more sustainably, or the that productivity levels can be increased irrigated area will have to contract. significantly across the board by switching to alternative farming approaches. In fact, Finally, as discussed in Section 3, there is most alternative farming approaches growing evidence from long-term crop cannot match the high productivity levels trials and declining TFP of the adverse achieved by modern farming methods in impact of environmental stress on crop GR areas. Pretty et al. (2007) in a revisit of yields in some GR areas. This may be the Pretty et al. (2003) examined yield claims result of the formation of hard pans in the for 286 projects sub-soil, soil toxicity buildups – especially iron – and micronutrient deficiencies –  Sometimes also called ‘sustainable’ or ‘ecological’ especially zinc (Pingali et al., 1997). farming.

24 — An Assessment of the Impact of Agricultural Research in South Asia since the Green Revolution

SC_south_asia_final.indd 24 1/8/08 15:28:29 disaggregated into eight farming systems that span the spectrum from zero use of categories developed by Dixon et al. (2001) modern inputs to high but precision- and showed that the more sizeable gains managed use. nearly all arose within rainfed farming systems. Moreover, the gains reported for Organic farming rice and wheat yields, the main GR crops, Despite widespread publicity to the contrary, were modest, sometimes even negative. organic farming seems to have little to offer farmers in GR areas who wish to continue to Despite significant R&D investments in envi- grow cereals. A recent study (Halberg et al., ronmentally oriented research of both par- 2006, p. 40) concludes: “In high-yielding adigms, there are very few impact studies regions with near to economic optimal of the value of that work. As with poverty inputs of fertilizers and pesticides, the yields impact assessment, the state of the art in of organic farming are between 15 and 35% assessing environmental impacts in ways lower than present yields when comparing that can be quantified in social cost–benefit single crops, and possibly at the low end calculations is still poorly developed. This is (35%) when including crop failures and the partly because of difficulties in measuring need for green manure in crop rotations.” environmental changes over the time spans This statement draws heavily on results from and levels of scale required, and also temperate countries, and crop losses could because of difficulties in assigning be even higher in tropical countries because economic values to changes, even when of greater problems with pest and disease they can be measured (Freeman et al., control. The same study concludes that 2005). The few impact studies that exist organic farming has more to offer farmers in either report changes in selected physical less-intensively farmed areas, such as many indicators or rely on farmers’ perceptions LFAs, or farmers who can benefit from price of change in resource or environmental premiums for organically produced foods. condition. However, these are sufficient to Zundel and Kilcher (2007) report somewhat demonstrate that relevant work has been lower yield losses for organic farming in undertaken with proven productivity and temperate and irrigated areas, but do not environmental impacts in the field, even allow for crop failures and diversion of land though we do not yet have calculations of to produce green manure and other organic the rates of return to those investments to matter. show which types of institutions or research give the best returns. Badgley et al. (2007) reviewed a large number of published studies comparing In reviewing these developments and their organic and conventional crops. Although impacts, we continue with the useful dis- they claim organically grown grains in de- tinction between intensively farmed GR veloping countries have an average yield areas and extensively farmed LFAs. advantage of 57%, the more detailed results in their Table A1 tell a more nuanced story. Organically grown rice 5.3 Evidence on impact in Green under irrigated conditions in South Asian Revolution areas countries showed little if any yield gain. The best organic farming yield gains for Only a few GR critics argue for a drastic South Asia were obtained on upland rice reversal from GR to traditional technologies and for maize and sorghum grown under of the kinds that dominated South Asia before the GR (e.g., Shiva, 1991; Nellitha-  Since organic agriculture involves greater generation of nam et al., 1998). Such authors claim that plant nutrients and organic matter within the landscape yield growth rates were already high through crop rotations, fallows, green manures, and before the GR, but ignore the fact that this integration of livestock into cropping systems, each was largely the result of the spread of irri- hectare of cropland harvested must be supported by additional land dedicated to these other needs. While gation and fertilizers prior to the introduc- it might well be possible to obtain comparable yields tion of HYVs (Evenson et al., 1999). More for some crops at the plot level, farm level productivity generally, R&D has contributed to a broad can be considerably lower for organic farming. Yet few range of technologies for improving soil, studies of yield gains with organic farming seem to make this basic correction, leading to results that are inevitably water and pest management in GR areas biased in their favour.

An Assessment of the Impact of Agricultural Research in South Asia since the Green Revolution — 25

SC_south_asia_final.indd 25 1/8/08 15:28:30 rainfed conditions. These are areas where (Sheehy et al., 2005). Proponents argue that the conventionally grown crops usually there are strong synergies between the dif- receive limited nutrient inputs of any kind ferent management components of SRI that and hence produce low yields. lead to strong root growth and higher yields, although these synergies are not System of rice intensification (SRI) well understood (Mishra et al., 2006). SRI was developed in the early1980s by Henri de Laulanie, a French missionary Few of the yield claims have been verified priest in Madagascar, as another alternative under controlled experimental conditions. farming approach to the available GR rice Trials undertaken at IRRI found no signifi- technologies for small-scale farmers. It has cant yield differences between SRI and con- since been widely promoted by a number ventional GR practices (quoted in Namara of NGOs and the International Institute for et al., 2003). McDonald et al. (2006) Food, Agriculture and Development (IIFAD) analyzed 40 sets of field trial results at Cornell University (http://ciifad.cornell. reported in the literature (five from Mada- edu/sri). Not only was SRI initially devel- gascar and 35 from 11 Asian countries) oped outside the international and public- which compared SRI with ‘best manage- sector research systems, but if its claimed ment practices’ appropriate to each site. benefits proved true, it would render irrel- Apart from the five Madagascar studies, evant much of the research on intensive which consistently showed higher yields rice farming that has been conducted in with SRI, SRI led to an average yield loss of recent decades by the public and interna- 11% in the other 35 studies, with a range tional R&D systems. Not surprisingly, SRI has of -61% to 22%. attracted the attention of the scientific and donor communities and sparked a lively Yield gains appear to be better in farm debate and research agenda. adoption studies. Farmers in Ratnapura and Kurunegala Districts in Sri Lanka obtained The main components of SRI are: trans- 44% higher yields on average with SRI than planting of young seedlings (8–15 day-old with modern rice farming methods instead of 3–4 week-old plants) on small (Namara et al., 2003), and the average yield hills at much lower plant densities than gain was 32% for farmers in Purila District usual; water management that keeps the of (Sinha and Talati, 2007). soil moist rather than flooded; frequent However, in both studies SRI farmers weeding; and use of large amounts of showed considerable variation in the man- organic compost for fertilizer. agement methods they used, making it rather unclear as to what was being The claimed benefits include: high yields compared in the name of SRI. For example, even with traditional rice varieties; a signifi- many SRI farmers used inorganic fertilizer cant savings in seed; little or no artificial as well as compost, many grew modern as fertilizer required; natural pest and disease well as traditional rice varieties, and their control, eliminating the need for pesticides; weeding and water management practices reduced water use; and a flexible manage- varied considerably. ment that allows farmers to experiment and adapt the approach to their particular SRI has yet to be widely adopted in any one growing conditions. The approach is country, although it can be found on small claimed to be environmentally sustainable scales in many countries, including many and of particularly relevance for poorer parts of South Asia.10 Some of the reasons farmers who cannot afford modern inputs for poor uptake include: the difficulties of (Uphoff, 2003). controlling water with sufficient precision in many surface irrigation systems, the need Controversy has arisen because of claims of for large amounts of compost, and the high very high yields, sometimes exceeding the labor demands for transplanting, hand best experiment station yields for modern weeding11 and generating and distributing rice technologies, sometimes even without the use of fertilizer or modern varieties. 10 See http://ciifad.cornell.edu/sri These high yields defy current understand- 11 The combination of wide spacing and reduced flooding creates ideal conditions for weed growth and hence the ing of the physiology of rice plant growth need for frequent weeding.

26 — An Assessment of the Impact of Agricultural Research in South Asia since the Green Revolution

SC_south_asia_final.indd 26 1/8/08 15:28:31 compost. This is confirmed by available undertook focus group discussions with adoption studies. In Sri Lanka, adoption is adopting and nonadopting farmers at sites positively related to family size (availability in India and two sites in Southeast Asia. For of labor) and ownership of animals (avail- India, yields of adopting farmers were ability of manure) and is more common found to be 17% higher. Modest savings in among rainfed than irrigated rice farmers fertilizer use were largely offset by higher (Namara et al., 2003). Moser and Barrett labor costs, but profit per hectare was 48% (2003) obtained similar results in an higher. There was also a useful reduction in adoption study in Madagascar. Moser and emissions of nitrous oxide, a powerful Barrett (2003), Namara et al. (2003), and greenhouse gas. In an impact study in West Sinha and Talati (2007) all found that Bengal, India, Islam et al. (2007) found adopters only practice SRI on small parts of small (but not significant) increases in their rice area despite higher returns to yields, but 20% savings in nitrogen use and both land and labor, and they also found 50% savings in pesticide use, and economic high rates of disadoption. This again benefits of US$19–27 per hectare depend- suggests important constraints, possibly ing on the season. labor or suitability of available irrigation systems, as well as disappointing returns. The International Center for Soil Fertility and Agricultural Development (IFDC) has Improved nutrient management been pioneering urea deep placement More pragmatic approaches to intensive (UDP) technology in rice. This involves the farming seek to increase the efficiency of deep placement of urea in the form of su- fertilizer use rather than displace it, thereby pergranules or small briquettes into reducing production costs and environmen- puddled soil shortly after transplanting the tal problems. Fertilizer efficiency can be rice (Bowen et al., 2005). The method improved through more precise matching improves nitrogen-use efficiency by of nutrients to plant needs during the keeping most of the urea nitrogen in the growing season, and by switching to soil close to the plant roots and out of the improved fertilizers such as controlled- floodwater where it is susceptible to loss. release fertilizers and deep-placement On-farm trials in Bangladesh that compared technologies. UDP with standard urea broadcasting prac- tices showed 50–60% savings in urea use Site-specific nutrient management (SSNM) and yield increases of about one ton per was developed by IRRI and its partners as a hectare (Bowen et al., 2005). The briquettes way of reducing fertilizer use, raising yields, are also simple to make with small pressing and avoiding nitrate runoff and greenhouse machines, and can create additional local gas emissions (especially nitrous oxide) from employment. Adoption data are not avail- intensive rice paddies (Pampolino et al., able, but the approach appears to be 2007). Developed in the mid-1990s, SSNM is spreading in Bangladesh with the active a form of precision farming that aims to support of the government. apply nutrients at optimal rates and times – taking account of other sources of nutrients Low or zero tillage (ZT) in the field and the stage of plant growth – In response to the declining growth in pro- to achieve high rice yields and high efficien- ductivity of the rice–wheat farming system cy of nutrient use by the crop. Farmers apply in the IGP, ZT has been adapted and intro- nitrogen several times over the growing duced by the RWC, a partnership of CGIAR period and use leaf color charts to deter- centers and the NARS from Bangladesh, mine how much nitrogen to apply at differ- India, Nepal, and Pakistan. The technology ent stages. SSNM has been tested through involves the direct planting of wheat after on-farm trials in several Asian countries and rice without any land preparation. Rice crop IRRI has developed practical manuals and a residues from the previous season are left web site (www.irri.org/irrc/ssnm) to guide on the ground as mulch. The wheat seed is application. typically inserted together with small amounts of fertilizer into slits made with a Pampolino et al. (2007) provide an special tractor-drawn seed drill. The tech- economic assessment of SSNM compared nology has many claimed advantages over to farmers’ usual fertilizer practices. They conventional tillage in the rice–wheat

An Assessment of the Impact of Agricultural Research in South Asia since the Green Revolution — 27

SC_south_asia_final.indd 27 1/8/08 15:28:32 system: it saves labor, fertilizer and energy; RWC partners in developing the technology minimizes planting delays between crops; for India’s IGP. Even with conservative as- conserves soil; reduces irrigation water sumptions about yield gains (6%) and cost needs; increases tolerance to drought; and savings (5%), the estimated benefit–cost reduces greenhouse gas emissions (Eren- ratio is 39 and the internal rate of return is stein et al., 2007; World Bank, 2007). But it 57%. With more optimistic assumptions often requires some use of herbicides for (yield gains and cost savings of 10%), the general weed control. A key ingredient for benefit–cost ratio increases to 68 and the its success has been the development of an internal rate of return to 66%. This analysis appropriate seed drill for local conditions in does not include any environmental the IGP. benefits.

In an assessment of the technology based on Improved water management a sample of farmers in Haryana, India and Improved water management in South the Punjab, Pakistan, Erenstein et al. (2007) Asian agriculture is essential for redressing found that ZT adoption has been rapid. In growing water scarcities, improving water Haryana, 34.5% of the sampled farmers had quality, and halting the degradation of adopted in 2003/04 and 19.4% in the additional irrigated land. This will require Punjab, even though diffusion of the tech- significant and complementary changes in nology only began around 2000. Adopting policies, institutions, and water manage- farmers used the technology on large shares ment technologies. Agricultural research of their total wheat areas. Adoption has has been conducted on all three aspects, been highest on larger farms with tractors. although little of this research has been The study found mixed results for yield gains subjected to impact analyses. and water savings (more significant in Haryana than the Punjab) but all farms Technical research has shown the potential made drastic savings in tractor and fuel to increase yields in irrigated farming with costs. There were no observed impacts on substantial savings in water use (e.g., the following rice crop. Although the tech- Mondal et al., 1993; Guerra et al., 1998). nology is attractive to farmers, the high per- Realizing these gains is easiest when centage of non-adopting farmers together farmers have direct control over their water with disadoption rates of 10–15% suggests supplies, as with tubewell irrigation or continuing constraints on its use. No one small-scale farmer-managed irrigation factor was clearly identified in the study, but schemes. For larger schemes, the best hope access to tractors and ZT seed drills is impor- lies in the devolution of water manage- tant, especially for smaller farms. Rental ment to local water user groups or associa- markets for these machines exist but may tions, an approach known as irrigation not offer farmers sufficient flexibility in the management transfer (IMT). timing of their operations, which is crucial if higher yields are to be obtained. Other ZT IMT began to be adopted in some South assessments from adoption studies, on-farm Asian countries during the late 1980s as a trials and focus group discussions confirm response to the disappointing performance the large savings in tractor and fuel costs, of many large-scale irrigation schemes. It and most show significant water savings and was hoped that IMT would increase the yield gains (Laxmi et al., 2007; Laxmi and accountability of water irrigation services to Mishra, 2007). farmers, encourage greater farmer input into the maintenance of irrigation systems, It is estimated that about 200,000 hectares improve cost recovery, and enable of wheat was planted under ZT in the improved control of water at local levels. Pakistan IGP in 2001/02 and 820,000 All this was expected to lead to higher hectares in the Indian IGP in 2003/04 (about water use efficiency, increased agricultural 8% of the total wheat area). The latter had productivity, better environmental doubled by 2004/05 (Laxmi et al., 2007). outcomes, and irrigation schemes that were Based on an estimated ceiling adoption more financially sustainable. rate of 33%, Laxmi et al. (2007) undertook an economic assessment of the likely Despite the promise, there was little hard returns to the research costs incurred by the evidence to show that IMT did in fact lead

28 — An Assessment of the Impact of Agricultural Research in South Asia since the Green Revolution

SC_south_asia_final.indd 28 1/8/08 15:28:33 to these realized benefits. IWMI therefore many of the first HYVs released had poor embarked on a set of studies in 1992 to resistance to some important pests. The monitor and evaluate the experience with problem was compounded by a shift to IMT and provide guidelines for its successful higher cropping intensities, monocropping, implementation in the future. The results high fertilizer use (which creates dense, from the Asian case studies proved disap- lush canopies in which pests can thrive), pointing. Sri Lanka, which began to imple- and the planting of large adjacent areas to ment IMT in 1988, is typical of the results similar varieties with a common susceptibil- obtained. Samad and Vermillion (1999) ity. Control was initially based on prophy- surveyed irrigation schemes that had been lactic chemical applications, driven by the transferred and some that had not, and calendar rather than incidence of pest within each there were schemes that were attack. This approach disrupted the natural rehabilitated and some that were not. The pest–predator balance and led to a resur- findings suggest only modest gains to gence of pest populations that required farmers or the sustainability of irrigation even more pesticide applications to control. schemes. Farmers in IMT areas did not incur Problems were compounded by the buildup additional water supply costs, but neither of pest resistance to the most commonly did they perceive any improvements in the used pesticides. As pesticide use increased, quality of water services they received from so did environmental and health problems. their irrigation agency. There were signifi- Rola and Pingali (1993) found that the cant gains in yields, land, and water pro- health costs of pesticide use in rice reached ductivity in some IMT areas, but the best the point where they more than offset the results were obtained in schemes that were economic benefits from pest control. both rehabilitated and transferred to producer organizations. Simply devolving As these problems began to emerge, re- management without also rehabilitating searchers gave greater attention to the de- the irrigation schemes achieved little. velopment of crop varieties that have good resistance to important pests and biological Following these mixed findings, IWMI and ecological pest control methods. This embarked on a follow-up program of led to the development of IPM, an research to identify best practice approach- approach that integrates pest-resistant vari- es from around the developing world. eties, natural control mechanisms, and the Within South Asia, IWMI subsequently judicious use of some pesticides. The CGIAR provided policy advice to the governments centers have been important sources of of Sri Lanka and Nepal in developing research on IPM, and IRRI has been espe- national IMT strategies, and engaged in cially important for IPM in rice in Asia action research in Pakistan and Sri Lanka to (Waibel, 1999). help improve implementation policies. This led to the development with the Food and Bangladesh has been in the forefront of Agriculture Organization of the United IPM since 1981, and the government, with Nations (FAO) of a handbook on best assistance from FAO, has aggressively practice (Vermillion and Sagardoy, 1999) promoted the approach through farmers’ and to a number of guidelines papers on training schools. Sabur and Molla (2001) specific implementation issues. undertook a farm survey in 1997/98 and found that IPM farmers used less than half A subsequent assessment of IWMI’s work on the amount of pesticides on rice than non- IMT is provided by Giordano et al. (2007). IPM farmers and had significantly higher They claim significant impact on water gross income per hectare. Similar results policies in Nepal and Sri Lanka and some were obtained by Susmita et al. (2007) and success in affecting the employment of by Rasul and Thapa (2003). Both studies improved techniques in Pakistan and Nepal. found that IPM farmers saved significantly They also report high demand for IWMI’s on costs (labor and pesticides). None of the guideline publications on IMT. studies report any significant productivity impact from use or IPM, so the main Integrated pest management (IPM) economic benefits arise from lower costs. Pest problems emerged as an important Farmers perceived fewer health problems problem during the early GR era because with IPM in all three studies, though

An Assessment of the Impact of Agricultural Research in South Asia since the Green Revolution — 29

SC_south_asia_final.indd 29 1/8/08 15:28:34 neither Susmita et al. (2007) or Rasul and soils, steep slopes, or other biophysical con- Thapa (2003) could find statistical differ- straints; as well as areas that may have ences between the perceptions of adopting higher development potential but that are and non-adopting farmers. None of the presently under-exploited due to poor in- studies provides any data on environmental frastructure and market access, low popula- impacts. tion density, or other socioeconomic con- straints. Conceptually they include all the There is no hard evidence to show that IPM shaded areas in Table 13. has been widely adopted among South Asian farmers. There are two difficult con- An attempt to operationalize this two-di- straints to overcome. One is farmer mensioned concept of LFAs suggests that training; IPM is knowledge-intensive, about one quarter of South Asia’s rural requiring farmers have the capability to population live in LFAs (World Bank, 2007). identify harmful and beneficial insects and the ability to flexibly manage their Much of the deforestation, woodland loss, response to pest attacks. Farmer field and land degradation (including soil schools have had some success in providing erosion and soil fertility loss) that has the required training (Waibel, 1999; Tripp occurred in South Asia arose in LFAs that et al., 2006; van den Berg and Jiggins, did not benefit much from the GR. This 2007). But this can be a slow and expensive degradation is often driven by insufficient way of training large numbers of farmers – agricultural intensification relative to popu- particularly if, as Tripp et al. (2006) found in lation growth. As more and more people Sri Lanka, knowledge-intensive methods seek to eke a living out of these areas, they like IPM do not easily spread from farmer expand cropping in unsustainable and to farmer. The other constraint is the need erosive ways and fail to replenish the soil for collective action among neighboring nutrients that they remove. While migra- farmers. IPM cannot be successfully under- tion and nonfarm development have im- taken at single plot or farm levels but must portant roles to play in reducing pressures be adopted at landscape levels. This is dif- on the natural resource base, more sustain- ficult to organize without effective commu- able forms of agricultural growth are nity or producer organizations. needed if the environmental problems in these areas are to be reversed.

5.4 Evidence on impact in less- LFAs also account for a significant propor- favored areas tion of the rural poor in South Asia. Precise estimation is difficult because poverty data Following Pender and Hazell (2000), LFAs are reported by administrative units rather are broadly defined in this paper to include than by agroecological areas or farming lands that have been neglected by humans systems. Fan and Hazell (2000) estimated as well as by nature. They include marginal that 41% of India’s rural poor (76 million lands that are of low agricultural potential people) lived in LFAs in 1993, and ICRISAT due to low and uncertain rainfall, poor estimates that 40% of India’s rural poor live

Table 13. Classification of favored and less-favored areas

Agricultural potential

Access to markets and Low infrastructure High (biophysical constraints)

High Favored areas Marginal areas (LFAs)

Low Remote areas (LFAs) Marginal and remote LFAs

Source: Pender and Hazell (2000)

30 — An Assessment of the Impact of Agricultural Research in South Asia since the Green Revolution

SC_south_asia_final.indd 30 1/8/08 15:28:35 in the semi-arid tropics and another 16% in improving NRM practices that conserve and arid areas and semi-arid temperate areas efficiently use scarce water, control erosion, (Rao et al., 2005). There is some controversy and restore soil fertility while using low about whether the incidence of poverty is amounts of external inputs. These kinds of higher among LFA populations than in ir- technology improvements can lead to sig- rigated and high-potential rainfed areas, nificant gains in productivity and stability but since estimates range from “no signifi- while reversing some types of resource deg- cant difference” (Kelley and Rao, 1995) to radation. Within this context, there has “higher concentrations of poor in LFAs” been considerable convergence between (Fan and Hazell, 2000), this controversy the objectives and approaches of different need not detract from the importance of farming paradigms for LFAs. agricultural research for LFAs. The analysis by Pretty et al. (2007) of yield An early and appropriate (at that time) bias claims for 286 sustainable agriculture during the GR era towards R&D spending projects from around the developing world on irrigated areas and best rainfed areas showed that the more sizeable gains nearly has changed. Pal and Byerlee (2006) found all arose within rainfed farming systems. no evidence of any underinvestment Some of the most successful projects for (relative to irrigated areas) in rainfed and these areas included improved crop variet- marginal lands by 1996–1998. At a com- ies, water harvesting, soil, and water con- modity level, Byerlee and Morris (1993) did servation at catchment or watershed levels, not find any bias for wheat research, but and use of organic residues for soil im- Pandey and Pal (2007) found a modest bias provement. For South Asia, yield gains of against LFAs in the allocation of research 63% are reported for highland mixed scientists for rice research. These studies cal- farming systems in India, Nepal, Pakistan, culate desired research shares on the basis and Sri Lanka, and 79% for rainfed mixed of congruency with agricultural or com- farming systems in India. modity outputs and not on the basis of poverty. An analysis based on poverty Of 293 yield ratios for organic versus might tell a different story, but that would modern crop production methods, first require resolving the controversy about reviewed by Badgley et al. (2007), only 10 where the poor are most concentrated, as have relevance to LFAs in South Asia. There well as an analysis of the relative merits of are five ratios for upland rice (ranging from the indirect (e.g., food and labor market) 1.23 in Pakistan to 3.4 in Nepal) and five for benefits from investing in each type of area sorghum and millets in India (ranging from (Renkow, 2000). An environmental perspec- 1.65 to 3.5). Organic farming in these loca- tive might also justify greater investment in tions requires mixed farming, soil and agricultural research in many LFAs12. water conservation, and use of organic residues for soil improvement. Most LFAs in South Asia are unsuitable for the kinds of intensive monocrop farming While there are grounds to be skeptical associated with the GR. A lack of irrigation about the high yield levels claimed in some potential, erratic and often deficient of these studies (Cassman, 2007), they are rainfall, poor soils, and, often, sloped land consistent with the fact that the existing make crops less responsive to fertilizers, farming systems are low-yielding, usually and the fragility of the resource base because of low rates of application of fertil- requires more integrated and mixed izers or organic matter and poor soil and farming approaches to avoid degradation. water management. In these circumstances, Economically, the remoteness of many LFAs many improved NRM practices that reverse from markets also makes modern inputs ex- land degradation, improve soil condition, pensive relative to the prices farmers and provide much-needed water and nutri- receive for their products. In this context, ents for crops can make a large difference, a lot of research has been targeted at whether motivated by alternative or modern agricultural philosophies. Even so, one recent study undertaken in a less-developed 12 An attempt to prioritize agricultural R&D on the basis of and hilly area of Himachal Pradesh, India, production, poverty, and environmental goals has been undertaken by Mruthyunjaya et al. (2003). found that while organically grown wheat

An Assessment of the Impact of Agricultural Research in South Asia since the Green Revolution — 

SC_south_asia_final.indd 31 1/8/08 15:28:36 and maize were more profitable than their The measured impacts of some of the com- modern production counterparts, this was modity-improvement work reviewed in nearly all due to a price premium of about Section 3.2 have arisen in LFAs (e.g., maize, 100% (Thakur and Sharma, 2005). sorghum, and millet), although the cited studies do not separate out the impacts in Important lines of research in LFAs involving LFAs from GR areas. However, a few CGIAR centers in South Asia include crop im- examples illustrate the impacts of crop- provement, watershed development, and improvement research that was targeted to integrated soil nutrient management. the specific problems of poor people in LFAs. Crop improvement research Much plant breeding for LFAs has focused As mentioned earlier, Shiyani et al. (2002) on producing varieties that can withstand found that ICRISAT-improved chickpea vari- drought and poor soil conditions and that eties have been widely adopted in a poor have greater pest and disease resistance. tribal area in Gujarat, India, with favorable Such varieties can raise average yield impacts on yields, unit production costs, response and reduce yield instability. They and net returns per hectare. can also contribute to reductions in pesti- cide use and, by raising the productivity of ICRISAT’s package of improved groundnut food crops, help reduce the cropped area varieties grown in combination with needed by subsistence-oriented farmers. improved agronomy practices built around This can reduce the pressure on more an RBF concept (see Table 9 and earlier dis- fragile lands and free up some land and cussion) is another example of a commod- labor for other activities. Most of ICRISAT’s ity-improvement program that has paid off crop improvement research is directed at handsomely in an LFA – in this case the LFAs, and there are spill-in benefits to these semi-arid tropical areas of central India. The areas from the crop improvement work high internal rate of return of about 25% that IRRI (upland rice), CIMMYT (maize), reported by Joshi and Bantilan (1998) is and CIP (potatoes) undertake more broadly seemingly robust to within a percentage in Asia. point or two, even when corrected for possible positive and negative environmen- At an aggregate level, there is evidence tal outcomes that affect yield and produc- from India that crop improvement research tion costs (Bantilan et al., 2005). This is one is having favorable productivity and of the few available impact studies that poverty impacts in many LFAs (Fan and attempts to value environmental impacts Hazell, 2000). Based on an econometric within a benefit–cost analysis framework. analysis of time-series data for three differ- ent types of agricultural areas (irrigated, Watershed development high-potential rainfed, and low-potential There have been significant investments in rainfed), they found more favorable research on watershed development in marginal returns (measured as rupees of South Asia in recent decades. India began agricultural production per additional developing model operational research hectare planted to modern varieties) for projects in a number of representative wa- crop improvement research in low-potential tersheds in the mid-1970s, and these were rainfed areas than in either high-potential used to test and validate integrated water- rainfed areas or irrigated areas. Moreover, shed management approaches before they additional crop research investment in low- were scaled up in huge publicly funded potential rainfed areas lifts more people schemes across the country. By 1999/2000, out of poverty than in the other two types India had spent Rs. 35,915 million to of areas. Fan et al. (2000a) provide a more develop 37 million ha, or 22% of the nuanced set of results for 13 different types problem area (Babu and Dhyani, 2005), of rainfed zones in India. They found seven and by the late 1990s was spending about zones where the benefit–cost ratio for US$500 million each year on additional additional crop-improvement research is watershed development projects (Kerr et greater than five and where there are also al., 2000). The total had exceeded US$2 favorable poverty impacts. Neither of these billion by 1999/2000 (Joshi et al., 2004). studies assesses environmental impacts. ICRISAT and IWMI have both undertaken

32 — An Assessment of the Impact of Agricultural Research in South Asia since the Green Revolution

SC_south_asia_final.indd 32 1/8/08 15:28:37 research on watershed development and Kerr et al. (2000) surveyed 86 villages in related soil and water management issues Maharashtra and , some of and have been involved in watershed evalu- which were included in watershed projects ation work. and some not. Three types of projects were included: government (Ministry of There have been many evaluations of Agriculture)-run projects, NGO-run watershed development projects in India, projects, and projects that were run collab- though seemingly none on the returns to oratively between NGOs and state govern- research on watershed development. Joshi ment. The government projects largely et al. (2005b) undertook a meta-analysis of focused on technical improvements; NGO 311 evaluation studies spanning a large projects focused more on social organiza- number of types of projects and agrocli- tion; and the collaborative projects tried to matic conditions. They found that the draw on the strengths of both approaches. average benefit–cost ratio was 2.14 (with a Qualitative and quantitative data were range of 0.8–7.1), and the average internal both collected, including data on condi- rate of return was 22% (with a range of tions in the study villages before and after 1.4–94%). On average, the projects created the projects were implemented. additional employment of 181 days per hectare per year, increased the irrigated Overall, the participatory NGO projects area by 34% and the cropping intensity by performed better than their technocratic, 64%, and slowed soil losses by 0.82 tons government-run counterparts. However, per hectare per year. Among other things, participation combined with sound techni- the meta-analysis showed that the cal input performed best of all. For benefit–cost ratio was: example, while all projects reduced soil erosion on uncultivated lands in their n Higher in areas with annual rainfall of upper watersheds reasonably well, the between 700 and 1100 mm than in NGO and NGO/government collaborative areas with low (less than 700 mm) or projects had particularly good records in high (greater than 1100 mm) rainfall this regard. Greater NGO and community n 42% greater in macro-watersheds involvement also helped ensure that (greater than 1250 hectares) than in project investments were maintained over micro-watersheds time. Although definitive hydrological n Larger when state governments were data were not available, farmers in villages involved in the planning and execution in NGO and NGO/government projects fre- compared to purely central government quently perceived that the projects’ water- projects harvesting efforts increased the availability n Higher when there was active people’s of water for irrigation and their net participation. returns to rainfed farming were higher.

An Assessment of the Impact of Agricultural Research in South Asia since the Green Revolution — 

SC_south_asia_final.indd 33 1/8/08 15:28:38