Agricultural research and poverty reduction

Material Information

Agricultural research and poverty reduction
Series Title:
Food, agriculture, and the environment discussion paper
Hazell, Peter
Haddad, Lawrence James
International Food Policy Research Institute
Place of Publication:
Washington D.C
International Food Policy Research Institute
Publication Date:
Physical Description:
vi, 41 p. : ill. ; 28 cm.


Subjects / Keywords:
Agriculture -- Research ( lcsh )
Poverty ( lcsh )
Food supply ( lcsh )
bibliography ( marcgt )
non-fiction ( marcgt )


Includes bibliographical references (p. 37-41).
General Note:
On cover: Technical Advisory Committee of the Consultative Group on International Agricultural Research.
General Note:
"20 20 vision"--Cover.
General Note:
"This paper has been prepared for the IFPRI 2020 conference on "Sustainable Food Security For all By 2020," Bonn, Germany, September 4-6, 2001 ..."--P. 2 of cover.
Statement of Responsibility:
Peter Hazell and Lawrence Haddad.

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
The University of Florida George A. Smathers Libraries respect the intellectual property rights of others and do not claim any copyright interest in this item. This item may be protected by copyright but is made available here under a claim of fair use (17 U.S.C. §107) for non-profit research and educational purposes. Users of this work have responsibility for determining copyright status prior to reusing, publishing or reproducing this item for purposes other than what is allowed by fair use or other copyright exemptions. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder. The Smathers Libraries would like to learn more about this item and invite individuals or organizations to contact Digital Services ( with any additional information they can provide.
Resource Identifier:
49314162 ( OCLC )
0896296393 ( ISBN )

Full Text
Food, Agriculture, and the Environment Discussion Paper 34
Agricultural Research and Poverty Reduction
Peter Hazell and Lawrence Haddad
International Food Policy Research Institute 2033 K Street, N.W.
Washington, D.C. 20006 U.S.A.
September 2001

Copyright @ 2001 International Food Policy Research Institute
All rights reserved. Sections of this report may be reproduced without the express permission of but with acknowledgment to the international Food Policy Research Institute.
ISBN 0-89629-639-3

Foreword v
Acknowledgments vi
1. Introduction 1
2. The Nature and Extent of Poverty 3
3. How Agricultural Research Can Help the Poor 9
4. Targeting Agricultural Research to Benefit the Poor 18
5. Strategies for Pro-Poor Agricultural Research 26
6. The Role of Public Research and Extension Systems 35
References 37

1. Trends in child malnutrition in developing countries, by region, 1970-95 6 2. Distribution of land types by region 6
3. Distribution of malnourished children by agroecological zone, 1990 7
4. Comparison of low- and middle-income countries 21
5. Priorities for agricultural research to reduce national poverty by type of adopting region 24
1. Number of people living on less than $1 a day, 1987 and 19984
2. Those living with HIV/AIDS and those newly infected: Asia and Sub-Saharan Africa 5 3. Food price trends in Bangladesh, 1973-96 15
4. Relationships between GNP per capita and population growth, agriculture's share
in GNP, and urbanization in low- and middle-income countries 22
5. Links between property rights, collective action, and technology adoption 28

International agricultural research has contributed enormously to increasing world food supplies to their current state of plenty. Yet poverty remains a major problem and the challenge for agricultural research now lies in developing strategies that more explicitly address the needs of the poor. This paper, based on the study commissioned by the Technical Advisory Committee (TAC) of the CGIAR system, addresses this issue.
Based on an analysis of the links between agricultural research and poverty alleviation in different types of countries and rural regions, Peter Hazell and Lawrence Haddad identify six key priorities for a pro-poor agricultural research agenda: (1) increasing production of staple foods in countries where food price effects are still important and/or that have a comparative advantage in growing these crops; (2) increasing agricultural productivity in many less-favored lands, especially heavily populated low-potential areas; (3) helping smallholder farms across the board diversify into higher value products, including livestock products, especially in countries with rapidly growing domestic markets for such products and/or access to suitable export markets; (4) increasing employment and income-earning opportunities for landlessand near-landless workers in labor surplus regions; (5) developing more nutritious and safer foods to enhance the diets of poor people; and (6) undertaking agricultural research in ways that are more empowering to the poor.
Hazell and Haddad discuss strategies for achieving each of these goals with the least trade-off in national agricultural growth. In short, Hazell and Haddad suggest strategies to target agricultural research on poor peoples' problems in ways that are "win-win" for growth and poverty reduction.
Although the rates of return to public agricultural research are known to be high, public funding for agricultural research has nevertheless declined, especially in developing countries. Yet more is being asked of agricultural research and extension systems. If new technologies and research paradigms are to be developed that specifically address the needs of the poor, then funding for international and national agricultural research must be increased.
Per Pinstrup-Andersen
Director General
Emil Javier
Technical Advisory Committee/CGIAR

The authors wish to thank Derek Byerlee, Dana Dalrymple, Alain de Janvry, Michael Lipton, and members of the Technical Advisory Committee of the Consultative Group on International Agricultural Research (CGIAR) for their valuable comments received on an earlier draft.

1. Introduction*
Until recently, researchers and policymakers saw chronic underinvestment in research for poor farmpoverty alleviation as a subsidiary goal of agri- jers and regions and for containment of environ-I cultural research. The primary goal was to mental problems. increase food supplies through cost-reducing tech- Against this backdrop, what can public agrinological changes that would lead to lower food cultural research systems that serve developing prices. Research focused on increasing the yields countries do to increase the poverty-reducing of important food staples in irrigated and high- impact of their investments? This paper addresses potential rainfed areas, where researchers per- that question generally in the Introduction. Chapter ceived the productivity returns to agricultural 2 examines the changing nature and extent of research to be highest. This strategy was enor- poverty, and its implications for agricultural mously successful and led to real benefits for the research. It notes the evolution in thinking about poor as well as for societies in general. But it was poverty beyond measures of physiological deprinot sufficient to eliminate rural poverty, which vation (an inability to meet basic material needs) to abounds even in countries that now have national incorporate measures of social deprivation (poor food surpluses. access to the components of power such as deciPlentiful global food supplies and declining sionmaking processes, information, and authority). trade barriers have created the opportunity to Chapter 2 also shows the shifts in poverty rates develop explicitly pro-poor research strategies for and numbers from a spatial point of view. the public sector. The opportunity is enlarged to by Chapter 3 reviews how agricultural research the private sector's increasing role in addressing can reduce poverty not only through traditional mainstream productivity challenges. Private-sector pathways (such as own-form productivity increasinvolvement seems likely to grow as the evolution es, greater employment, general equilibrium of biotechnology and intellectual property regimes effects, and the lowering of food prices), but also redefine the traditional public-goods nature of through newer pathways (such as community much agricultural research. Although food sup- empowerment through collective action for natuplies must double over the next 20 years, increas- ral-resource management). Chapters 2 and 3 suging productivity alone will not be enough. The gest the need to tightly target agricultural public sector will also need to focus sharply on the research. changing nature of market failures that it has tra- Chapter 4 outlines one possible typology to ditionally sought to correct and on the growing help do so. The typology was developed for importance of countries' other social and environ- potential technology-adopting regions within Mental goals. This shift will require the public sec- countries and is based on two country criteria tar to invest in ways that offset the private sector's (low, versus middle-income country and liberal*This paper is based on the TAC-commissioned study, which is still under review by TAG. Consequently, the paper represents a work that is in progress.

ized versus unliberalized market and trade poli- key to long-term poverty reduction. The six priority cies) and four region-specific characteristics (high areas for pro-poor research are: (1) maintaining versus low agricultural potential; good versus poor growth in staple food production, particularly in infrastructure, service provision, and market countries where food-price effects are still strong; access; low wages and abundant labor supply (2) intensifying less-favored lands; (3) helping versus high wages and scarce labor supply; and smallholders to diversify into higher value prodwhether the region's poor are empowered or dis- ucts; (4) increasing employment and income empowered in terms of access to land, other opportunities for landless workers; (5) increasing resources, and public services), the access of the poor (especially poor women) to
Chapter 5 describes the six pro-poor research foods rich in crucial micronutrients; and (6) underpriorities that emerge from the typology and con- taking agricultural research in ways that are more siders research strategies for achieving them, empowering to the poor. emphasizing strategies to minimize trade-offs Chapter 6 concludes with recommendations against sectorwide agricultural growth, which is for the public sector's future research agenda.

2. The Nature and Extent of Poverty
DeiigPoveirty basis for international comparisons and trends
Defininganalysis. Income and consumption measures are
A consensus is emerging around the view that usually preferred, but nutrition indicators based on poverty consists of two interacting deprivations- the height and weight of young children are often physiological and social. Physiological depriva- used as simpler measures of physiological deprition describes an inability to meet or achieve vation. basic material and physiological needs and can I Social deprivation typically is assessed at the be measured either as a lack of income, which individual or community level; it refers to an limits access to food and to education, health, absence of elements that are empowering--autonhousing, water, and sanitation services, or by the amy, time, information, dignity, and self-esteem. failure to achieve desired outcomes, such as a Lack of empowerment is reflected in exclusion high-quality diet rich in micronutrients, health sta- from important dlecisiont-making processes, even tus, educational attainment, and the quality of when the outcomes are of considerable imporhealth, water, and sanitation services received. tance to the poor-for example, decisions about Diet quality is crucial to individual well-being, par- public investments in the local community, manticularly for girls, women, and infants. agement of common properties, and priorities for
Although income and achievements correlate, agricultural research and extension. Physiological they do not do so perfectly. For example, house- and social deprivations interact in a vicious/virtuholds with relatively high-income consumption lev- ous cycle, with increased empowerment leading els often contain individuals who do not get to greater income-earning ability, which leads to enough of the right type of food to stave off greater power, and so on. This more comprehenhunger, anemia, goiter, or death. This can occur sive social definition of poverty provides richer because of lack of information (for example, insights into the process of becoming poor, stayhousehold leaders do not realize the value of girls, ing poor, or becoming less poor. It also provides education or do not know that their daughters' an opportunity for input from the poor themselves, diets are low in iron), preferences (there may be a in terms of the dimensions of deprivation identified deliberate bias toward education for boys at the and the way in which severity is assessed. This expense of girls), or an inability to use income to input can be gained through participatory purchase inputs required for well-being (for exam- research methods. ple, sanitation services that cannot be purchased Because measures of social deprivation are in the market). in their infancy and are based largely on qualiMeasures of physiological deprivation are rel- tative analysis, quantitative indicators that are atively easy to quantify at the household level in consistent over time and comparable across ways that are consistent over time and across regions and countries are not available to space. For this reason, usable data series have describe the extent and location of poverty been constructed for most countries to form the around the world, which this report attempts to

Figure 1-Number of people living on less than $1 a day, 1987 and 1998
Millions of people
550 522
101987 ......
El 1998 .......
350 291
303 291
250 213 217.
150 114
64 78
50 2II 9 2
East Asia China Europe and LAC MENA South Asia Sub-Saharan
(excluding Central Asia Africa
Source: World Bank 2000.
Note: [AC = Latin America and the Caribbean; MENA = Middle East and North Africa.
do. But where qualitative assessments have been ity across space and time. These rates also are undertaken, they provide a richer understanding reliable lead indicators of poverty because malnuof the nature of poverty and of poor peoples' trition at a young age leads to the early onset of livelihood strategies. Such information can be poor physical and cognitive productivity and to crucial to guide decisions about the design and higher rates of noncommunicable diseases such targeting of agricultural research that meshes as diabetes and heart disease later in life with the livelihood strategies of the poor. Public (ACC/SCN 2000). agricultural research systems may need to invest Globally, about 90 percent of the developing in collecting this kind of information in important world's poor live in either Asia or Sub-Saharan target regions. Africa (Figure 1 ). Asia dominates, with two-thirds
of the total poor, who are concentrated in South Asia (43 percent). Less than 1 percent of the poor Poverty Trends live in the Middle East and North Africa (MENA),
and about 7 percent live in Latin America and the Poverty continues to be a major problem in many Caribbean ([AC). parts of the developing world. Approximately 1 .2 The total number of poor has changed little billion rural people live in poverty/(defined as liv- since 1987, but the regional distribution has ing on less than $1 per day), and about 160 mil- chngdwth poverty declining in China, East lion preschool children are malnourished (World Asia, and MENA but increasing in South Asia and Bank 2000; Pinstrup-Andersen, Pandya-Lorch, Sub-Saharan Africa. Poverty is growing the fastest and Rosegrant 1997). Infant malnutrition rates reli- in Sub-Saharan Africa, where it is aggravated by a ably measure deprivation, given their comparabil- higher incidence of HIV/AIDS infection (Figure 2).

Figure 2-Those living with HIV/AIDS and those newly infected: Asia and Sub-Saharan Africa, 1999
Millions of people
25 23.3
20 -0 Adults and children living with HIV/AIDS
0 Adults and children newly infected with HIV/AIDS
3.8 6
1.3 0.53 01
Sub-Saharan Africa South and East Asia
Southeast Asia and Pacific
The incidence and distribution of malnour- same pattern as the total poor. A significant perished (underweight) children of preschool age centage of the rural poor live in less-tavored areas shows a similar pattern (Table 1). Most of the ihui ore challenged by difculJagroclimatic condeveloping world's malnourished children live in ditions, such as poor soil, low and unstable rainSouth Asia or Sub-Saharan Africa (together fal, steep slopes, and short growing seasons accounting for 70 percent), again with the and/or inadequate infrastructure and support majority concentrated in South Asia (51 per- services (roads, irrigation, markets, research and cent). About 49 percent of all preschool children extension, credit, schools, and health centers). in South Asia are malnourished, compared with 'Kccording to a report by the Technical Advisory 31 percent in Sub-Saharan Africa and 23 per- Committee (TAC) of the Consultative Group on cent in East Asia (including China). Trends in the International Agricultural Research (CGIAR/TAC number of malnourished preschool children also 2000), "favored" agricultural lands account for show similar patterns to monetary measures of only 10.7 percent of the agricultural area in the poverty. While the number of underweight chil- developing world and for 8.5 and 16.6 percent, dren in South Asia has slowly declined, the num- respectively, in Sub-Saharan Africa and Asia ber in Sub-Saharan Africa increased from 25.7 (Table 2). Moreover, only about one third of the million in 1990 to 31.4 million in 1995. In the rural population live in favored lands (Table 2). past five years, the situation in Eastern and The vast majority of rural people live outside these Southern Africa has continued to deteriorate, favored lands, which in the TAC study comprise spurred in part by the HIV/AIDS epidemic "marginal" lands (24 percent of the total agricul(ACC/SCN 2000). tural area), sparsely populated arid lands (26 perOf the total number of poor people living cent), and forest and woodlands (40 percent). below $1 per day, 75 percent, or 0.9 billion, live Some of these less-favored lands have good agriand work in rural areas (IFAD 2001). The rural cultural potential, particularly where roads and poor are distributed across regions in much the irrigation are available, so it cannot be concluded

Table 1-Trends in child malnutrition in developing countries, by region, 1970-95
Region 1970 1975 1980 1985 1990 1995 1970 to 1995
Percent of children malnourished Percent Percentage points
South Asia 72.3 67.7 63.7 61.1 53.4 49.3 -23.0
Sub-Saharan Africa 35.0 31.4 28.9 29.9 28.8 31.1 -3.9
East Asia 39.5 33.3 30.0 26.5 23.5 22.9 -16.6
Near East and North Africa 20.7 19.8 17.2 15.1 n.a. 14.6 -6.1
Latin America and the Caribbean 21.0 17.0 12.2 10.6 11.4 9.5 -11.5
All regions 46.5 41.6 37.8 36.1 32.3 31.0 -15.5
Number of children malnourished Millions
South Asia 92.2 90.6 89.9 100.1 95.4 86.0 -6.2
Sub-Saharan Africa 18.5 18.5 19.9 24.1 25.7 31.4 +12.9
East Asia 77.6 45.1 43.3 42.8 42.5 38.2 -39.4
Near East and North Africa 5.9 5.2 5.0 5.0 n.a. 6.3 +0.4
Latin America and the Caribbean 9.5 8.2 6.2 5.7 6.2 5.2 -4.3
All regions 203.8 167.6 164.3 177.7 176.7 167.1 -36.7
Source: Smith and Haddad 2000.
Notes: A child under five (0-59 months) is considered malnourished if the child falls below an anthropometric cut-off of -2 standard deviations below the median weight-for-age Z-score of the National Center for Health Statistics/World Health Organization international reference. n.a. is not available.
that the remaining two-thirds of the rural popula- child malnutrition. Sharma et al. (1996) have tion all live in low-potential areas. mapped the incidence of malnutrition among
Although reliable poverty data by land type preschoolers by the agroecological zones defined do not exist for most countries, more precise data by CGIAR (Table 3). Their analysis shows that the do exist for India and China. A recent IFPRI study incidence of child malnutrition is highest in warm, reports that for India in 1993, 42 percent of the semi-arid tropical and subtropical areas (zones 1 rural poor lived in low-potential rainfed areas, and 5) and that 43 percent of all malnourished while 16 percent lived in irrigated areas and 42 preschool children live in these areas. Food propercent lived in high-potential rainfed areas (Fan duction per hectare is also relatively low in these and Hazell 2000). A similar share of China's rural zones despite average or above-average levels of poor lives in low-potential areas (Fan et al. 2000). irrigation.
Additional insights into the distribution of rural The rural poor are predominantly smallholder poverty by land type are available from data on farmers and landless agricultural workers. For the
Table 2-Distribution of land types by region
Land type (% of total land) Rural population
Sparsely populated Forest and living in
Region Favored Marginal arid lands woodland favored lands (%)
Sub-Saharan Africa 8.5 23.1 24.6 43.7 27.0
Asia 16.6 30.0 18.5 34.6 37.0
Latin America and Caribbean 9.6 20.3 8.1 61.9 34.0
Near East and North Africa 7.8 22.6 65.8 3.9 24.0
Total (105 countries) 10.7 24.0 25.9 39.4 35.0
Saur-e: CGIAR/TAC 2000.

Table 3-Distribution of malnourished children by agroecological zone, 1990
Malnourished children
Food production/
% of hectare of % of arable
Agreecological zone Millions total children arable land (TGE) land irrigated
1. Warm, semi-arid tropics 47.9 49.0 0.98 17.2
2. Warm, sub-humid tropics 20.6 36.4 1.03 9.3
3. Warm, humid tropics 38.0 37.0 1.92 18.2
4. Cool tropics 8.1 26.0 1.50 8.3
5. Warm, semi-arid subtropics (summer rainfall) 31.7 44.0 1.07 40.0
6. Warm, sub-humid subtropics (summer rainfall) 7.4 38.0 1.44 26.8
7. Warm/cool humid subtropics (summer rainfall) 10.4 19.0 1.41 21.3
8. Cool subtropics (summer rainfall) 10.6 23.0 1.28 14.6
9. Cool subtropics (winter rainfall) 8.2 17.4 0.78 23.2
Total 183.4 33.7 1.20 20.0
Source: Sharma et al. 1996 (Table 2).
Note: TGE = total grain equivalent.
developing countries as a whole, about half the Asian countries seem to be following the Japanese rural popuation ived in smallholder farm house- experience, retaining large numbers of smallholdholds mi Y and one-quarter lived in landless er farms that are becoming part-time enterprises.
labor households (Jazairy et al. 1992). The ratio Diversification is more likely to be associated with of landless to smallholder farmers was much lower greater impoveisThment when increasing land in Sub-Saharan Africa (0.15) than in Latin scarcity occurs in conjunction wih slow agriculturAmerica (0.82) or Asia (0.53). '1 growth, stagnant national and regional
Smallholder farms are getting smaller and economies, and falling wages (Hazell and More numerous in most parts of the developing Reardon 1998). Such situations are not uncomworldHozell, Jagger, and IKnox 2UO}. This con- mon in many of the poorer countries in South Asia
trasts sharply with changes in the industrialized and Sub-Saharan Africa.
countries where farms are getting larger and there Although there are many explanations for curhas been an exodus of small-scale farmers from rent levels and trends in poverty around the agriculture, world, agricultural growth is an important conSmallholder households are also diversifying tributing factor. Its influence has been most studtheirF'elihood strategies and increasing their led for India. Prior to the Green Revolution in the share of nonfarm income (Reardon et al. 1998; late 1960s, the incidence of rural poverty in India Carney 1998). Smallholders and landless workers fluctuated widely. Researchers obtained different typically earn more than half their total household results on the relationship between poverty and income from nonagricultural sources. Such diversi- agricultural growth, depending on the period fication could reflect worsening impoverishment they chose for their analysis (Bardhan 1973; and desperation as land becomes increasingly Ahluwalia 1978; Gaiha 1989; Ghose 1989; scarce, or it could reflect increasing prosperity, as Griffin and Ghose 1979; Saith 1981). But after rural workers are attracted to higher-paying non- the Green Revolution began in the mid-1960s, farm jobs. The Southeast Asian experience has the incidence of rural poverty began a definite been largely of the latter variety, with rapid downward trend (From about two-thirds to onegrowth in rural nonfarm employment and income third of the rural population by the early 990s).
as a result of dynamic national economies A greater consensus began to emerge in the liter(Rosegrant and Hazell 2000). Many Southeast ature on the poverty-reducing impact of agricul-

tural growth (Ghose 1989; Fan, Hazell, and incomes across the board, it has significantly conThorat 1999; Datt and Ravallion 1997). Rural tributed to reducing the number of people living poverty also declined dramatically in China after below the poverty line. In contrast, in Sub-Saharan policy reforms launched rapid increases in agri- Africa, where poverty is increasing and food insecultural growth and in the purchasing power of curity is deteriorating, agricultural growth has rural households (Fan, Zhang, and Zhang 2000). been very disappointing, often struggling just to Rosegrant and Hazell (2000) report similar keep pace with population growth. broad relationships across much of Asia during It would be dangerous to conclude too much the Green Revolution era. Whereas 60 percent of from these patterns of association without more Asians lived in poverty in 1975, this ratio had micro-based evidence on cause-and-effect relafallen to less than one in three by 1995, and the tionships. Nor should it be concluded that agritotal number of poor declined from about 1.2 bil- cultural growth necessarily reduces poverty. But lion to about 0.8 billion despite a 1 billion the data do provide an optimistic backdrop that increase in the total population. justifies more careful analysis of how technologiAgricultural growth has not always improved cally driven agricultural growth can benefit poor income distribution, but by raising per capita people.

3. How Agricultural Research Can Help the Poor
Agricultural research that leads to improved tech- These benefits do not necessarily materialize nologies can benefit the poor in a number of for the poor. Many condlitioning factors determine ways: who benefits from technological change. Nor do
the benefits of research all necessarily work in the
I 1. Research can help poor farmers directly same direction. For example, while many of the
through increased own-farm production, pro- poor may benefit from less costly food and greater viding more food and nutrients for their own opportunities for nonfarm income, production and consumption and increasing the output of mar- employment benefits in the adopting regions may keted products for greater farm income, be disappointing or even perverse. Net outcomes, both for individual poor people and for entire
2. Small farmers and landless laborers can gain poor populations, can be difficult to determine
greater agricultural employment opportunities beforehand.
and higher wages within the adopting
On-Farm Productivity Impacts
3. The poor can have opportunities to migrate to
other agricultural regions. Poor farmers will obtain own-farm benefits from
new technologies only if they adopt them. This
4. Growth in the rural and urban nonfarm econ- means that the new technologies must be approomy induced by more rapid agricultural priate and profitable For farming conditions, and growth can benefit a wide range of rural and that poor farmers must have access to the knowlurban poor people. edge and inputs necessary to adopt the technology. In principle, improved crop varieties are
5. Research can lead to lower food prices for all scale-neutral and can be adopted by farms of all
consumers, whether from rural or urban areas. sizes, but the same is not always true of other technologies or of complementary inputs like irri6.Research can lead to greater physical and gation and machines, and access to fertilizers and ~ I economic access to crops that are high in credit. If institutions that provide these services
nutrients and crucial to the well-being of the and inputs are biased in favor of large farms, the poor-prticularly poor women. poor may not be able to adopt new technologies.
To invest in new technologies whose returns
7. Research can empower the poor by increasing occur over a number of years (for example,
their access to decisionmaking processes, improved tree crops or better soil-management enhancing their capacity for collective action techniques) and to obtain credit to finance such and reducing their vulnerability to economic long-term investments, poor farmers need secure shocks via asset accumulation, ownership or tenancy rights. Insecure rights to

land may increase poor farmers' vulnerability to their better access to irrigation water, Fertilizers, eviction should larger farmers and landlords want seeds, and credit, owners of large farms were to expand their own cropped area as the result of the main adopters of the new technologies, and more profitable technologies. smallholders were either unaffected or harmed
Under risky agroclimatic conditions, poor because the Green Revolution resulted in lower farmers may be reluctant to adopt profitable product prices, higher input prices, and owners' new technologies because they require input efforts to increase rents or force tenants off the investments that could be lost in an unfavorable land. Critics also argued that the Green year. On the other hand, larger farmers are Revolution encouraged unnecessary mechanizamore likely to assume such risks because they tion, with a resulting reduction in rural wages have larger reserves and better access to credit and employment. The net result, critics said, was and insurance. an increase in inequality of income and asset disFarmers who adopt new technologies often tribution and a worsening of absolute poverty succeed in lowering their production costs per (see Griffin 1974; Frankel 1976; Farmer 1977; unit of output (though not usually per hectare), ILO 1977; Pearse 1980). and therefore can better compete in the market. Although a number of village and household Moreover, if the technology is widely adopted studies conducted soon after the release of and market prices fall as a result, the decline in Green Revolution technologies lent some support unit cost may be essential for maintaining form to early critics, more recent evidence shows income. In this case, farmers who do not adopt mixed outcomes (Blyn 1983; Pinstrup-Andersen the technology will be disadvantaged not only and Hazell 1985; Lipton and Longhurst 1989; by stagnant production but also by declining Hazell and Ramasamy 1991; David and Otsuka prices and tighter profit margins. This profit 1994). Although small farmers did lag behind squeeze can be detrimental to nonadopters with- large farmers in adopting Green Revolution techin technology-adopting regions and to farmers nologies, many of them eventually did so. Many who live in regions that are inappropriate for the of these small-farm adopters benefited from new technology, increased production, greater employment
Even when poor formers do benefit from sig- opportunities, and higher wages in the agriculnificant productivity gains, these benefits are not tural and nonfarm sectors. Moreover, most small always shared equitably among household mem- farmers were able to hold onto their land and bers. In many societies, men and women have captured significant total production increases responsibility for growing different crops. from their holdings (Westley 1986; Hazell and Therefore, which crops benefit from technological Ramasamy 1991; Rosegrant and Hazell 2000). change determines who controls the increased In some cases, small farmers and landless laborproduction within the household. Technological ers actually ended up gaining proportionally change for women's food crops may translate into more income than larger farmers, with a net better nutrition and well-being for women and chil- improvement in the distribution of village income dren than technological change for men's cash (Hazell and Ramasamy, 1991). crops (Haddad, Hoddinott, and Alderman 1997). This is not to say that the Green Revolution
The initial experience with the Green was equitable everywhere (Freebairn 1995). But Revolution in Asia stimulated a large body of the conditions under which it and similar yieldempirical literature on how agricultural techno- enhancing technologies are likely to have equilogical change affects poor farmers (see Kerr table on-form benefits are now reasonably well and Kolavalli 1999 for a recent review). Critics understood. These include: (1) a scale-neutral techof the Green Revolution argued that because of nology package that can be profitably adopted

on farms of all sizes; (2) an equitable distribution increases. This may lead to greater agricultural of land with secure ownership or tenancy rights; employment, but lower living standards for work(3) efficient input, credit, and product markets so ers. The decline in living standards can be particthat farms of all sizes have access to needed mod- ularly sharp if rapid population growth coincides emn farm inputs and information and are able to with adoption of agricultural technologies with receive similar prices for their products; and (4) low employment elasticities. policies that do not discriminate against small
farms and landless laborers (for instance, no subsidies on mechanization and no scale-biases in Impact on Inter-regional agricultural research and extension). These condli- Migration tions are not easy to meet. Typically, government
must make a concerted effort to ensure that small Technological change in agriculture is usually sitefarmers have fair access to land, knowledge, and specific and does not benefit all regions equally. modern inputs. The Green Revolution was initially concentrated in
irrigated regions and only later spread to more
favorable rainfed areas. Technological change,
Agricultural Employment therefore, can contribute to widening disparities
and Wages between regions. Worse, if the technology leads
to lower production costs per unit of output in the
Many yield-enhancing technologies increase total adopting regions, producer prices may fall, leavon-farm employment, particularly if they expand ing non-adopting regions with lower prices as well the gross cropped area, for example, by growing as stagnant yields, so that their incomes could more crops per year with irrigation and short-sea- actually decline. But inter-regional migration acts son crop varieties. This can lead to less seasonal to buffer these gaps and provides an efficient way fluctuation in employment earnings. But whether of spreading the benefits to poorer regions with allI this translates into higher wage earnings for the more limited agricultural growth potential. poor depends in large part on the elasticity of the In Indio, the Green Revolution led to the sealabor supply. If labor is abundant in the adopting sonal migration of more than a million agriculturregion, then additional employment will have little al workers each year from the eastern states to effect on wages, workers will gain, and farmers Pun jab and Haryana (Westley 1986). Moreover, will have limited incentive to invest in labor-dis- in a study of the impact of the Green Revolution in placing machines. But if the labor supply is inelas- a sample of Asian villages, David and Otsuka tic, wages will rise sharply and labor-displacing (1994) found that seasonal migration played an machines may become attractive. Initial mecha- important role in spreading the benefits between nization may be targeted on labor-intensive tasks technology-adopting and -nonadlopting regions. like plowing and threshing, but once farmers But even though migration can buffer widening invest in tractors the incremental costs of mecha- income differentials between regions, it is rarl nizing other tasks may become quite low, causing sufficient to avoid it. In India, poverty in many lowmore widespread displacement of labor. potential rainfall areas has changed little even Mechanization may also occur prematurely if gov- while irrigated and high-potential rainfall areas erment policies such as cheap credit for large have progressed (Fan and Hazell 2000). farms make mechanization less costly. Regional inequalities have also worsened in
Population growth increases the supply elas- China in recent years (Knight and Song 1992; ticity of labor and therefore acts to dampen wage ZAng and Fan 2001).

Impact on the Nonfarm Nonfarm income is also important to the poor
Economy as a means to help stabilize household income in
drought years (Reardon et al. 1998). In a study of
Agricultural growth generates important rounds of several villages in the semi-arid tropics of India, for income and employment growth within the non- example, Walker and Ryan (1990) found that farm economy. These are driven by (1) increased nonagricultural self-employment and labor market demands for additional farm inputs, investment earnings became increasingly important sources of goods, and marketing services (demands that income during the 1980s, increasing mean income often increase per hectare with technological and dampening household income variability. change); (2) increased rural household demands Numerous studies have shown that agricultural for consumer goods and services as farm and growth generates important income and employwage incomes rise; and (3) generation of farm ment multipliers within the surrounding nonfarm savings, foreign exchange earnings, and lower economy. The multipliers are particularly large in prices for foods and other primary commodities. Asia, with $0.5 to $1.0 of additional income creThis paper distinguishes between the rural ated in the local nonfarm economy for each dollar nonfarm economy, including its market towns, and of additional income created in agriculture (Bell, the broader urban-based national economy. The Hazell, and Slade 1982; Hazell and Ramasamy rural nonfarm economy has more transparent links 1991; Haggblade and Hazell 1989). The multiplito agriculture and is especially important to the ers are about half as large in Africa and Latin rural poor. However, there are also powerful links America (Haggblade and Hazell 1989). between agricultural and national economic The multipliers are predominantly driven by growth that benefit society at large, including the increased rural household demands for consumer rural and urban poor. goods and services as farm incomes rise. Small,
informal, and labor-intensive rural nonfarm firms
The Rural Nonfarm Economy supply many of these goods and services. This
The rural nonfarm economy is of special impor- leads to high nonfarm employment elasticities tance to the rural poor and to the urban poor liv- within rural regions. Often, each 1 percent ing in rural towns. Landless and near-landless increase in agricultural output is associated with a households everywhere depend on nonfarm earn- 1 percent increase in rural nonfarm employment ings. Those with less than half a hectare earn (Gibb 1974; Hazell and Haggblade 1991). betwee~n30 and 90 ento r income m The strength of the regional growth linkages is
n sources (Haggblade, Hazell, and Brown higher in labor-abundant regions and increases
1989; Rosegrant and Hazel 2000). Nonfarm with regional development and per capita shares are strongly and negatively related to farm incomes. Irrigated regions dominated by mediumsize. Low-investment manufacturing and servic- sized farms and modern input-intensive farming es-including weaving, pottery, gathering, food systems generate the largest multipliers. The multipreparation and processing, domestic and per- pliers are smaller in rainfed farming systems and sonal services, and unskilled nonfarm wage in regions dominated by very small farms or large labor--account for a greater share of income for estates (Haggblade and Hazell 1989). In India, the rural poor than for the wealthy (Hazell and for example, the multipliers are largest in Green Haggblade 1993). The reverse is true of trans- Revolution states like Punjab and Haryana and port, commerce, and such manufacturing activities smallest in less developed states like Bihar and as milling and metal fabrication, which require Madhya Pradesh (Hazell and Haggblade 1991). sizable investments. As regions develop and labor markets tighten,

employment impacts become less important. The *supplies basic foods, raw materials for agroinlower multipliers in Africa are attributable to low dlustry, and exports, and frees up foreign per capita incomes (which are mostly spent on exchange for the importation of strategic food), poor infrastructure, and farming technolo- industrial and capital goods. gies that require few purchased inputs
(Haggblade and Hazell 1989). In Latin America, *releases labor and capital (in the form of rural higher per capita rural incomes ought to lead to savings and taxes) to the nonfarm sector. larger multipliers, but often fail to do so because Generates purchasing power among the rural the distribution of income is highly inequitable and population for nonfood consumer goods and the richer households have much stronger con- services, supporting growth in services and sumption linkages to cities than the rural nonfarm trade. economy.
Rural income multipliers and employment elas- *provides a nascent market for an emerging ticities of the sizes observed in Asia mean that manufacturing sector. technological change in agriculture has the potential to generate significant new nonfarm income- *reduces poverty by increasing labor productivearning opportunities for the poor. These may ity and employment and by lowering food arise in the form of greater nonfarm employment, prices for all. higher wages, and opportunities for the poor to As the transformation of an economy start or expand nonfarm businesses of their own. advances, agriculture's share in national income Increasing competition for labor between agricul- falls and its importance for national economic ture and the local nonfarm economy can also con- growth diminishes. The nonagricultural sector tribute to higher agricultural wages for the poor. becomes the primary engine of growth and is no
The benefits of growth in the rural nonfarm longer as dependent on resource Rlows from agrieconomy are concentrated in towns more than vil- culture or on agriculture's demand linkages. lages, affecting an important segment of the non- However, agriculture's share of total employment farm poor residing in towns. Distribution of falls more slowly than its share of national income, benefits between rural farm areas and towns with the inevitable result that agricultural labor depends on the state of infrastructure connecting productivity, and hence per capita farm incomes, the two, on population density, on government lag behind the nonagricultural sector. The problem policies, and on average per capita income levels is then to absorb workers out of agriculture at a (Haggblade, Hazell, and Brown 1989). sufficiently rapid rate to stop their average productivity (and hence their incomes) from lagging
The Nationall Economy too far behind the levels achieved in the nonagriAgricultural growth has broad general equilibrium cultural sector. Few countries have been able to
(GE) impacts on the national economy that vary manage this transition successfully. Either rural according to stage of economic development. In poverty has persisted until late in the development poorer countries, agriculture accounts for the lion's process or governments have engaged in expenshare of national income, employment, and sive farm-income support policies. export earnings. Under these conditions, even a Empirical studies confirm the importance of modest growth rate for agriculture can have sig- these GE effects in developing countries. In India, nificant leverage on the national economy. Rapid the fact that the nonfarm share of total national agricultural growth contributes to the economic employment did not change for over a century transformation of a country in a number of impor- until the full force of the Green Revolution was tant ways. It: underway in the 1970s provides strong circum-

stantial evidence of the importance of agricultural because of recent rounds of market liberalization growth as a motor for India's nonfarm economy. policies. But many poor countries still face high Rangarajan (1982) confirmed this, estimating that transport costs because of poor infrastructure, a 1 percent increase in the agricultural growth remoteness from world markets, or inefficient marrate stimulated a 0.5 percent rise in the growth keting institutions. Hence, domestic prices are still rate of industrial output and a 0.7 percent rise in responsive to local supply even after market liberthe growth rate of national income. alization. In many landlocked African countries,
Computable general equilibrium (CGE) mod- for example, domestic prices still fall sharply when eling studies show that these GE effects are domestic food production increases suddenly. stronger in economies that have more liberalized Some traditional food crops are not traded in trade and that invest in adequate levels of rural world markets, and therefore yam, millet, taro, infrastructure and service provision (Robinson, and teff prices continue to be endogenously deterRoe, and Yelden 1998). Modeling results also mined within the countries that grow them. confirm that agriculture-led growth strategies are Food-price benefits may be enhanced if techmore beneficial for overall economic growth than nological change leads to lower production costs industry-led strategies, particularly in agrarian per unit of output. Farmers can then maintain or economies (Adelman 1984; De Franco and increase profits even selling at lower prices. But
Godoy 1993). whether consumers benefit from these lower costs
depends on the food marketing and distribution
system being sufficiently competitive so that cost
Impact on Food Prices savings at the farm gate are passed up through
and Diet Quality the marketing chain. In some cases, the cost savings are simply captured as additional profits in
Technological change contributes to increases in the marketing chain. the aggregate output of affected commodities and Technological changes that smooth seasonal often lower unit costs. This has proved to be one food supplies, such as irrigation and short-season of the most important ways through which poor rice varieties, can help smooth seasonal price varipeople have benefited from technological change ation. This can be of considerable benefit to the in agriculture (Scobie and Posado 1978; poor. The rural poor may become more foodRosegrant and Hazell 2000; Fan, Hazell, and secure from increased local production by reducThorat 1999). ing the need to purchase food from outside the
If the demand for these products is downward region. Locally grown produce is cheaper sloping (that is, export opportunities are con- because there is no need to cover high transstrained by trade policy or by high transport portation costs. costs), the output price will fall. The more elastic Food-price declines amount to an increase in the supply relative to demand, the greater the real income for net food-purchasing households. price decline will be (Alston, Norton, and Pardey Real-income increases can be used to increase 1995). Lower food prices benefit rural and urban consumption of important staples and to purchase poor alike. Because food accounts for a large more diverse, nutritionally rich diets. What is the share of their total expenditures, the poor gain best way to improve the nutrient content of the proportionally more than the nonpoor from a poor people's diets through agricultural research? decline in food prices (Pinstrup-Andersen and By concentrating on increasing incomes through Hazell 1985). These price effects may be muted in productivity-enhancing investments in staple foods open economies with low transport costs. More or by concentrating on decreasing the relative countries now fall into this category than before prices of micronutrient-rich foods?

Figure 3-Food price trends in Bangladesh, 1973-96
Index (1973-75 = 1.0)
2.5 "'... Tomatoes
2.5- / ....
2.0- Pumpkin
1.5 Chicken
Rice paddy
.. ... . . . ...... .... ..... .. ... .....--.-.. ... ... ... . .............. .................. ..... ... .
------- ----- ........... . . . .
0.5I I I I I I I
1973-75 1976-78 1979-81 1982-84 1985-87 1988-90 1991-93 1994-96
Source: Bouis 2000.
Figure 3 shows a downward trend in the price the rapidly changing consensus in the nutrition of rice in Bangladesh. However, it also shows community as to the limiting factors in the diet (from upward trends in the real prices of other foods that protein to calories and micronutrients) made the are richer in micronutrients. This may reflect under- quality-protein maize (QPM) experience somewhat investment in technologies for the production of demoralizing for the plant-breeding community nutrient-rich Foods such as Fruits, vegetables, and (Tripp 1990). nonruminant livestock. Areas that are more remote Despite this recent history, a new generation tend to have less access to perishable foods via of plant-breeding efforts is underway (see Graham the marketplace, so investment in these crops may and Welch 1996 for a good summary). The focus well deliver the highest return in terms of micronu- this time is not on protein, but on micronutrients. trients delivered per dollar of research resources There are three broad goals: (1) increase the spent. In areas with good market access, the ques- micronutrient concentration in the crop, (2) tion is more complicated, since access to micronu- decrease the concentration of absorption trient-rich foods is dependent on food-price inhibitors such as phytic acid, and (3) increase the elasticities, the extent to which women control of concentration of promoter compounds (for iron household income, and the quality of information and zinc in particular) such as sulphur-containing and education programs related to diet. amino acids (Ruel and Bouis 1998). The two
broad technologies are traditional breeding (looking for naturally occurring genetic variation in
Enhancing the Nutrients in micronutrient content) and biotechnology (genetic
Staple Food Crops modification of foods and the creation of new
Agricultural research that enhances the nutrient The breeding approaches face many chalquality of foods poor people eat can directly lenges. Can high nutrient-density cultivars be improve their diets. Breeding maize for higher found with (1) little or no yield trade-off so that quality proteins is an early example. Unfortunately, farmers will be interested in adopting them, (2) lit-

tie impact on consumer acceptance (storage, of poverty and vulnerability. Access to assets is cooking, appearance, and taste), and (3) no neg- important, but so is the ability to use them in comative impact on bioavailability (for the strategies bination to create secure livelihoods. Assets can that increase micronutrient density)? These chal- be seen as a base of power, enabling people "to lenges are similar to those faced by other food- act and to reproduce, challenge, or change the based interventions, but with the added baggage rules that govern the control, use, and transformaof the QPM experience. tion of resources" (Bebbington 1999).
Compared to the traditional breeding Recognizing that the lack of capacity is a comapproach, the biotechnology work is at a much ponent of poverty, development practitioners are earlier stage. But it is yielding promising results. focusing on empowerment. Kerr and Kolavalli The Swiss Federal Institute of Technology's Institute (1999) define empowerment as "a development for Plant Sciences has demonstrated some success strategy that seeks to bring about change through in introducing genes that increase iron and vita- modifications in the power structure, changing the min A concentrations in rice (the so-called "gold- social order in which the poor live." They distinen" rice). The Swiss team plans to collaborate with guish between two types of empowerment: motithe International Rice Research Institute (IRRI) to vational, which involves enhancing peoples' test the health and environmental consequences of abilities thereby contributing to greater confidence the technology and to evaluate the acceptability of and self-reliance, and relational, which implies the rice to farmers in terms of yield impacts. changing power structures and gaining access to political decisionmaking. The objectives of
Impact on the Vulnerability, Assets, empowerment are increased self-reliance and selfand Empowerment of the Poor determination-essential tools for breaking out of
Poverty is more than a lack of sufficient income or the poverty trap as well as pursuing a prosperous, food to meet basic material needs. It is also a state fulfilling life. of social deprivation, involving vulnerability and Empowerment tools can offer users of local lack of participation in decisionmaking and in resources better access to and control over the civil, social, and cultural life. It places a wide local environment as well as associated producarray of limitations on the capacity of the poor to tion and conservation technologies. Some of the substantially improve their lives. The lack of capac- most fundamental empowerment tools include ity or power is itself a fundamental characteristic property rights (which strengthen the asset base), of being poor (Carney 1998). local collective action and organizations (such as
The assets that individuals, households, and cooperatives and microcredit groups), and politicommunities control are critical for their capacity col organization and advocacy. But agricultural to cope with vulnerability and to establish secure technology can also contribute to empowerment, livelihoods. In many developing countries, the particularly if the poor participate in technology poor are highly dependent on natural resources development. Who decides research priorities within their local environments. Poverty therefore and technology development approaches are two can be exacerbated by not having access to those important aspects of how research systems operresources. The cycle is self-perpetuating when ate. In most countries, resource-poor farmers in poor people have no access to technologies and unfavorable regions are passive recipients of techinputs that enable better use of resources, or when nologies; they have no control over the priorities the poor do not participate in the design and eval- of the research systems that serve them. The uation of those technologies. In addition, lack of research process ignores farmers' knowledge and access to financial and human capital, social net- experience even though they may offer insights works, and political power compound conditions that could help develop effective technologies for

unfavorable areas. Such systems may perpetuate strategies and the dynamics of change is therefore 4J
a sense of helplessness among resource-poor fundamental for assessing the impact of technofarmers who wait in vain for effective technologi- logical change at the household level.
cal solutions to come from outside. Assessing how technological change affects
Participatory research and dissemination the poor in aggregate is even more complex and
strategies offer an alternate approach based on uncertain. Not only is there complexity in assessempowerment, building farmers' own capability ing the net impacts on different types of poor to innovate and giving them greater influence over households, but the various direct and indirect decisions in agricultural research. Participatory impacts also affect different types of poor houseresearch involving women farmers can also con- holds in different ways. Some poor households tribute to their empowerment. Participatory may be net gainers while others are net losers. It research and dissemination approaches where is possible, For example, that the indirect growth farmers communicate results within and between and food-price benefits are strong and generally communities can enhance the asset base of pea- favorable for the urban poor, but that they are pie and communities by building social capital weaker in rural areas and perhaps are even offset and organizational capacity. by adverse direct impacts because of inequitable
land distribution or poor service support for small
farmers. Whether there is a net gain or loss to the
Net Impacts on the Poor poor in aggregate will then depend on how many
of them are rural and the relative size of the gains
W IPoor people have complex livelihood strategies. and losses that different groups experience.
The rural poor, for example, are often part farm- It is also necessary to distinguish between ers, part laborers, and part nonfarmers-and long- and short-term impacts on the poor. Direct always consumers. As such, they may gain or lose benefits within technology-adopting regions are in different dimensions at the same time, so that appealing because they can almost immediately the net impact of technological change on poor and transparently affect the rural poor. Indirect households can remain ambiguous. A poor farmer growth and food-price benefits take longer to might be able to gain from increased on-farm pro- materialize, and their links to agricultural technolduction as a technology adopter, but may lose or ogy are less apparent. However, the indirect bengain from increased agricultural wages or efits can be much more important in the long term reduced food prices depending on whether he or for reducing both urban and rural poverty. In Asia, she is a net buyer or seller of labor or food. A for example, although the Green Revolution had small, nonfarm, business entrepreneur might gain mixed direct impacts on the rural poor, its contrifrom cheaper food, but business profits might fall butions to productivity enhancement were a major or rise depending on whether or not hired labor factor in reducing food prices and in launching costs rise faster than sales. There is also a tempo- the rapid economic growth of the region. This ral dimension to these outcomes, with some costs growth led to significant increases in per capital and benefits being realized before others have incomes (especially in Southeast Asia) and a had time to work through (for example, GE decline in the number of people living in poverty impacts take longer to materialize than on-farm (Asian Development Bank 2000; Rosegrant and ~ > ~benefits). Understanding household livelihood HazelI 2000).

4. Targeting Agricultural Research
to Benefit the Poor
Chapter 3 highlights the important role that a num- National Characteristics ber of national and local characteristics play in 1. Market liberalization policies and nationconditioning the ways in which agricultural tech- a! rural infrastructure development. Countries that nologies affect the poor and the difficulty of gen- pursue liberal trade policies and liberalize their eralizing about when technological change will domestic markets typically grow faster than other benefit the poor. To target pro-poor agricultural countries and this contributes to greater long-term research more effectively, the types of research reductions in poverty (World Bank 2000). These undertaken need to be tailored to specific country growth effects can be even stronger in agrarianand region conditions. A key question is whether based economies (for example, much of Africa one can go beyond pure site specificity and iden- and South Asia) when trade liberalization is comtify sufficient commonalities across sites to con- plemented by increased investment in rural infrastruct a typology that can serve as a filtering structure (Robinson, Roe, and Yelden 1998). device for selecting appropriate types of agricul- National investments in agricultural research and tural research for different socioeconomic con- rural infrastructure also contribute to agricultural texts. and rural nonfarm economic growth and to rural
poverty reduction in their own right, even when
markets are not widely liberalized (Fan, Hazell,
A Typology and Thorot 1999; Fan, Zhong, and Zhang 2000).
Agricultural research can be expected to conThis chapter develops a typology of agricultural tribute more to growth and therefore to long-term regionss based on agroclimatic and socioecono-m-- poverty reduction in countries with liberalized ic factors that ondition the size and distribution markets and good levels of rural infrastructure. By of the benefits that might be obtained from tech- connecting farmers to larger markets, both domesnological change. Though such a typology tic and international, and enhancing economic
would seem essential for helping to target pro- growth, farmers may benefit from greater opporpoor agricultural research, its construction _does tunities to diversify into higher-value products. not seem to have been attempted before. This These are typically employment-intensive and can attempt is largely exploratory (even speculative) benefit smallholder farms and landless workers. but it is hoped that it lays a useful foundation for Moreover, the prices farmers receive for staples future work. To construct the typology, the more ore less likely to fall as production increases, important conditioning factors of the size of the enabling food-surplus farmers to capture larger direct and indirect benefits identified in the previ- shares of the productivity increases arising from ous chapter are used as classification criteria, technological change. The downside to this is that with differentiation between national and local or consumers and food-deficit farmers are less likely region-specific factors. to benefit from price reductions for staple foods as

production increases, though they may benefit poor will be smaller if the poor are predominantly From more plentiful supplies of higher value and urban-based. In these cases, indirect effects such more nutritious foods, such as vegetables, fruits, as food price changes and growth of the nonfarm and livestock products. Farmers living in remote economy will be much more important, especially regions also face potential hardships if they are in countries that are not well integrated with globunable to compete in more liberalized markets al food markets. It is then appropriate to give because of higher transport and marketing costs. greater attention to pure productivity gains in Smallholder farms everywhere may also be domestic agriculture (with an eye to the relative unable to compete in markets for higher-value prices of foods that are key for the eradication of products if they are unable to organize and mar- diet-quality problems of the urban poor) rather ket their smaller volumes of output competitively than worrying about their social distribution withwith large farms, in the rural sector. The opposite holds when the
2. National per capita income level. As poor are predominantly engaged in agriculture. countries grow and national per capita incomes 4. Population growth. Population growth rise, agriculture's share in national income dimin- adds to the rural labor supply in agrarian counishes. This has a number of important implications tries and, other things being equal, acts to reduce for growth-poverty relationships. First, because wages. Rapid population growth can lead to staghouseholds allocate increasing shares of their nant living standards and can mask gains from income to nonfoods as they get richer, rising per technological change. Population growth typically capita incomes have a powerful impact on both declines as per capita incomes rise, but where this the level and composition of demand for foods is not true, then agricultural growth should continand nonfoods. Rural nonfarm income multipliers ue to be labor intensive until late in the developtend to be larger with higher per capita incomes, ment process. a situation that benefits the rural poor, but broader general equilibrium effects on the national Regional or Local Characteristics economy get smaller as the agricultural sector 1. Agroclimatic conditions. The agroclimatic becomes relatively less important. Second, the conditions of a region determine whether it has structure of poverty also changes with increases in high or low potential for agriculture. The use of average per capita incomes, with clearer demar- modern inputs and the potential to achieve higher cation of hard-core poor poverty groups that do levels of output per hectare and per worker are not easily benefit from agricultural and national typically greater in high potential areas (HPAs), economic growth. These poverty groups tend to creating greater growth opportunities for reducing concentrate in urban areas and less-favored or local poverty. But even many low-potential areas marginal regions. Third, since agriculture's (LPAs) can achieve high levels of output per employment share typically diminishes at a slower hectare if they have good infrastructure and rate than its income share as per capita incomes access to markets (through livestock and tree crop rise, labor productivity lags behind that of other production, for example, rather than staple foods). sectors. This can lead to worsening inequality Unfortunately, many IPAs have also been neglectbetween farm and nonfarm incomes, though not ed through past patterns of public investment and necessarily to worsening poverty. Technologies have weak infrastructure and market access, and policies to increase average labor productivi- which greatly restricts their ability to improve agrity therefore become more important as countries cultural productivity and reduce poverty. Because develop. the development of the local non farm economy
3. The share of 1poor who are urban. The also hinges crucially on agricultural growth to gendirect benefits of agricultural technology for the erate markets for most of its products, there are

typically fewer opportunities for farmers and land- 4. Infrastructure and the provision of key less workers to diversify into productive nonfarm agricultural services. These are essential for ensuractivities in LPAs. In short, the rural nonfarm econ- ing that poor farmers can adopt new technologies omy tends to be much weaker in LPAs and and take advantage of new income-earning
stronger in HPAs, reinforcing the initial inequities opportunities such as diversification into higher in agroclimatic conditions (Hazell and Reardon, value products. They also affect the size of the 1998). nonfarm income multipliers within an adopting
2. Labor market situation. Agricultural wage region. Regions with poor infrastructure and servearnings for the poor tend to be greater when ice support are not only likely to be poorer, but agriculture has a high employment elasticity. In poor farmers living within those regions are less South Asia, employment elasticities of around likely to be able to adopt improved technologies. 0.7-0.8 were observed during the peak of the 5. Local institutions and empowerment. Green Revolution (that is, each 1 percent increase Cultural and institutional factors that work against in agricultural output led to a 0.7-0.8 percent the poor are not easily overcome by economic increase in agricultural employment), but employ- changes alone. These factors can include racial, ment elasticities are probably about half that level tribal, and religious discrimination, gender biases, today. The rural nonfarm economy also has a high exclusive property rights arrangements, local employment elasticity (often around 1 .0), so that power structures that favor the rich, and poor the combination of an employmentelastic agricul- delivery of public services to needy groups and ture and a vibrant rural nonfarm economy can less developed regions. In regions where such lead to strong growth in demand for rural labor, biases are strong, technological change in agriwhich can be very beneficial far the poor. On the culture may sometimes work against the interests other hand, an abundant or elastic supply of labor of the poor. Improved governance arrangements (such as occurs with rapid population growth but and greater empowerment of the poor may then slow agricultural growth) keeps wages down, need to be a prerequisite for successful pro-poor thereby diluting the benefits of growth in employ- agricultural growth. ment opportunities for the poor. Promotion of The list of criteria is too long for a practical labor-intensive crop and livestock technologies are typology, and it is necessary to cluster some of particularly important in labor-abundant regions, these criteria to obtain a more manageable numas is the need to avoid policies and investments ber. Fortunately, some important national characthat lead to premature mechanization of forming. teristics are strongly correlated. The World Bank
3. Land distribution and the incidence of classifies countries by per capita income level, landlessness. An unequal distribution of land and two groups of countries are important for this reduces the percentage of the rural poor who can study: low-income countries (LICs) that had 1999 gain from own-farm productivity increases and per capita incomes below $755 and middletypically contributes to excessive mechanization income countries (MICs) that had 1999 per capiand low employment elasticities in agriculture. It to incomes of between $756 and $9,265. As also leads to weaker demand far local nonfood Table 4 shows, this classification reveals that LiCs goods and services, and hence to smaller rural not only have lower per capita incomes than MICs income and employment multipliers. Technological ($410 versus $2,000) but they also have smaller change rarely contributes to rural poverty reduc- urban population shares, higher population tion in technology-adopting regions when the dis- growth rates, and larger agricultural sector shares tribution of land is badly skewed, as amply in national output. The income classification of a demonstrated by some experiences with the country therefore is used as a proxy far these three Green Revolution. variables. But Figure 4 provides a cautionary

Table 4-Comparison of low- and middle-income countries
Indicator Low-inicome countries Middle-income countries
GNP per capita (1999 US$) 410 2,000
Urban population r/% total in 1999) 31 50
Annual population growth rate (%/), 1990-99 2.0 1.2
Agriculture's share in GDP I% in 1999) 27 10
Population density (people/square kilometer In 1999) 73 40
Arable land per capita (hectares in 1995-97) 0.19 0.23
Source: World Bank 2000.
reminder that there is still considerable variation in 0 region has low wages and abundant labor these relationships within the two country groups supply versus high wages and scarce labor and that one must be careful in practice not to supply; and assume too much for a particular country situation.
It is also reasonable to merge concern for the region has favorable versus unfavorable level of national investment in rural areas and empowerment of the poor. local or regional development of infrastructure and agricultural services and to define this variable at the regional level. Finally, a strong and Implications for Agricultural positive association between an inequitable distri- Research Priorities bution of land and a high incidence of landlessness with institutional and cultural biases against Table 5 shows how the typology can be used to the poor is hypothesized. Taken together, these set priorities for agricultural research that will be factors can be considered an index of empower- most appropriate For reducing poverty in different ment of the poor at the regional level, defined in types of regions and country situations. Two types terms of their access to land and other resources, of priorities are captured in this summary: (1) interand to public services, regional allocation of research resources to
With these changes, a regional typology con achieve the largest reduction in poverty at the be defined based on two country characteristics national level and (2) the kinds of research that and four local characteristics: are most appropriate within individual types of
regions. Reading across a row (corresponding to region located in a low- versus middle-income a type of country) in Table 5, the shaded cells
country; depict the types of regions that should receive
highest research priority within such countries. The region located in a country with liberalized digital codes in the cells depict priority research
versus unliberalized market and trade policies; activities within each type of region. Of course, not all countries have all the types of regions region has high versus low agroclimatic poten- shown in the table, in which case there are fewer
tial for agricultural growth; interregional choices to be made. The unshaded
cells in the table depict regions where research is region has good vs. poor rural infrastructure, more likely to lead to smaller net reductions in
service provision, and market access; national poverty than if the same resources were

Figure 4-Relationships between GNP per capita and population growth, agriculture's share in GNP, and urbanization in low- and middle-income countries
Population growth %) 50
3.0 o
2.0 ,
1.54 ,
1 .0 i '
0.0 ................................. .... .. ... ..... ........ .......
1000 2000 3000 4000 5000 6000 7000 8000 9000 GNP per capita (US$) Agricultural share in GNP (%) 50.070
a to
500 a a
10.0 a
5.0 V.0 ............. ............ .......... .. ......... .. ... ... .... ...............
1000 2000 3000 4000 5000 6000 7000 8000 9000 GNP per capita (US$) % of population in urban areas 100.0.
80. .
A A 70.0
60.0 "
50.0. A 'b.
30.0f 20.0 A
0.0 1 ..................................................................................................................
1000 2000 3000 4000 5000 6000 7000 8000 9000 GNP per capita (US$) Source: World Bank 2000.

allocated to shaded regions. But research in these pie. But where national staple food surpluses have regions may still benefit important poverty groups. already been achieved, or where infrastructure
Although empowerment of the poor was iden- and markets are weak, production increases will tified as a classification criterion in constructing need to be spread more broadly across regions the typology, the cells in Table 5 are not disag- (including some [PAs) to achieve widespread food gregated by this criterion. This is because the security. Because national demands for highertypes of research considered in the table are value crop and livestock products are limited at defined at a high level of generality, and the low per capita income levels, significant research appropriate choice at this level is less one of on these products may be relevant only where selecting the type of research or technology need- there are major export opportunities. These are ed to empower poor people than of choosing the more likely to occur in countries that have liberal way in which agricultural research and extension market and trade policies. However, attempts to is conducted. Agricultural research can contribute improve the nutrient quality of food staples may to the empowerment of the poor if conducted in have high payoffs for improving the nutritional participatory ways, but on its own it has limited well-being of the poor, especially in regions with capacity beyond labor market impacts to directly inadequate infrastructure and market access. overcome the problems of seriously disempow- Research priorities for different types of ered people who are denied access to land and regions within LICs need to be adjusted to their other key resources and public services, key characteristics if they are to be most helpful to
In these cases, new technologies may bring the local poor. In labor surplus regions, emphasis significant changes for the poor only if comple- should be placed on the development of staple mentary social and political changes are under- food technologies that are labor-intensive and are taken. Appropriate actions might need to include attractive to smallholder forms. In labor-scare land redistribution programs, rental market regions, technologies should enhance labor proreforms, microfinance to increase the assets of the ductivity, and this may require focus on some poor, and improved delivery of public services to forms of mechanization. Technologies that require the poor. In some cases, social action may be intensive use of fertilizers, purchased seeds, and required to organize poor people for greater polit- pesticides will be more relevant for regions that ical voice in local decisions that affect them, or for have high agricultural potential and good infrabetter access to and management of common structure and market access, while improved natuproperty like communal grazing and woodlot ral resource management and low external-input areas. This often requires direct interventions by farming technologies will be more relevant for grassroots organizations such as NGOs or the LPAs and/or regions with weak infrastructure and reform of local government. Agricultural research inadequate market access. Poor farmers are least systems have little capacity or mandate to under- likely to be able to afford modern inputs in LPAs, take this kind of work, but they may need to devel- a pattern that is reinforced when they are disemop new kinds of partnerships with other types of powered in their access to land, other resources, development agencies if their technologies are to and public services. benefit the disempowered poor. In MICs, high priority should be given to
In LICs, the importance of staple foods in rural increasing agricultural productivity through diverlivelihoods and the national diet requires that high sification into higher-value crops and livestock in national priority be given to agricultural research most kinds of regions with reasonable infrastructhat increases staple food production in high- ture and market access. Staple foods are still potential areas. This will create larger marketed important, but except in MICs that have not libersurpluses and benefit large numbers of poor peo- alized their trade and markets, staple food

Table 5-Priorities for agricultural research to reduce national poverty by type of adopting region Regional characteristics
Good infrastructure Poor infrastructure
Surplus labor Scarce labor Surplus labor Scarce labor
Low High Low High Low High Low High
Country setting potential potential potential potential potential potential potential potential
Middle-income country Markets liberalized 1 ,2,35 2 5, ,. 1,4,6 4 ,6, 8 1: 3, 5, 7 3,5,8 1,4,6,7 4,6,8
Markets not liberalized 1, 2, 3,5 1,2,3, 58 1,4,6 1,4,6, 8 1 J, 3,5, 7, 9 1, 3,5 5' 1,4,6,7,9 1,4,6,8
Low-income country Markets liberalized 3,5 1 3,8 2,4,5,8 1 24,5, 6, 8: 1,3,5,7,9 1, 3,5,7 9 1,4,5,7,9 1,4,5,7,9
Markets not liberalized 1,3,5,9 1,31, 5,8, 9 1,4,5,8,9 1, 4, 5, 8, 9 1,3,5,7,9 1,315,7,9 1,4,5,7,9 1,4,5,7,9
Priorities for agricultural research:
1. Staple food production
2. High-value crops, trees and livestock
3. Employment intensive growth 4. Increased labor productivity
5. Smallholder farms
6. Medium and large farms
7. Low external-input farming 8. High external-input farming
9. Nutritional content of food staples

research should receive greatest priority in LPAs LPAs that have poor infrastructure and market that have poor infrastructure and their own food access. security problems, and in HPAs that have contin- National labor markets typically become uing comparative advantage in growing these tighter and wages rise as MICs develop, and farm crops. Smallholders continue to deserve priority incomes generally grow more slowly than nonin all labor-surplus areas, but research priorities farm incomes. Maintaining reasonable parity in labor-scarce regions should be consistent with between farm and nonfarm incomes requires that the need to increase holding sizes, particularly in research shift toward the development of techregions that have good infrastructure and vibrant nologies that increase labor productivity. Less local nonfarm economies and that are well con- favored areas also tend to get left behind and nected to urban areas. Labor-intensive technolo- require targeted research and other public investgies remain important in labor-surplus regions, ments to reduce poverty and food insecurity but diversification into higher-value crops and among their populations. Investments that are winlivestock can also add significantly to local win for growth and poverty reduction are needed employment. High external-input forming is more to avoid significant tradeoffs against the interests widely applicable in MICs, except perhaps in of the poor outside targeted LPAs.

5. Strategies for Pro-Poor Agricultural Research
Chapter 4 identified several key topics for a pro- well-integrated rural areas, increases physical poor agricultural research agenda. These include: access in remote rural areas, or increases the nutrient content of food staple crops via tradiI. Increasing production of s in coun- tional or transgenic technologies. This is espetries where food price effects are still important cially important for LICs where poor diet and/or that have comparative advantage in quality generates micronutrient deficiencies growing these crops. This includes most LICs, that impair the health of current and future genbut also many MICs that have not liberalized erations.
their trade and markets.
6. Undertaking agricultural research in ways that
2. Increasing agricultural productivity in many are more empowering to the poor.
less-favored lands. Special attention is needed
for heavily populated LPAs, but also some If these objectives could be achieved without HPAs that are constrained by poor infrastruc- having to trade off much agricultural growth at the ture and market access. Appropriate technolo- sector level, this would lead to larger indirect bengies for these areas will often have to be efits in the nonfarm economy that would contribute based on low use of external inputs, to further rounds of longer-term poverty reduction.
In short, agricultural research must target poor
3. Helping smallholder forms in all kinds of areas peoples' problems in ways that are win-win for
to diversify into higher value products, includ- growth and poverty reduction. The following is a ing livestock products, especially in countries discussion of appropriate research strategies for
with rapidly growing domestic markets for achieving each of these objectives.
such products (most MICs) and/or access to
suitable export markets.
Research Strategies for Staple
4. Increasing employment and income-earning Food Production
opportunities for landless and near-landless
workers in labor-surplus regions. This is espe- Irrigated and high-potential rainfed areas will cially important in LICs with growing popula- continue to be the major producers of staple tions and land scarcity. foods in most countries, and improved technologies and natural resource management (NRM)
5. Developing more nutritious and safer foods to practices for these areas will be critical for mainenhance the diets of poor people by investing taining ample and affordable food supplies.
in agricultural technology that reduces the Growth in staple food production will also be price of micronutrient-rich foods in urban and important in many less-favored lands as a way of

targeting poverty and food security (see below). Research Strategies for LessIn high-potential areas with abundant labor in the Favored Areas form of landless workers or small farms, technologies and NRM practices are needed that are While some types of commodity improvement employment intensive (Lipton with Longhurst work seem vital for less-favored areas-improving 1989; IFAD 2001). On the other hand, in high- drought tolerance, yield response to scarce plant potential areas that must compete for labor in nutrients, food nutrient content, pest and disease tightening labor markets, technologies that resistance, and livestock health and productivityincrease both labor and land productivity are there is a growing consensus that major producrequired. Technologies also need to reduce pro- tivity improvements will first come from improved auction costs per unit of output so that farmers natural resource management (NRM) practices can better compete in world markets (including and technologies. In areas of poor soils and harsh competing with imports) and consumers can ben- climate, NRM can lead the way in improving soil efit from cheaper foods and raw materials, depth, organic matter, fertility, and moisture conThe Green Revolution technologies achieved tent, and eventually higher yields from fertilizers many of these goals, but they have now largely and improved varieties. run their course. There is a need for more The poor infrastructure and market access that
upstream germplasm improvement work, includ- characterize many less-favored areas make the ing biotechnology. Yet at the same time, use of high levels of external inputs uneconomic,
germplasm improvement needs to be integrated placing a premium on low external input (LEI) techwith better natural resource management prac- nologies. LEI technologies are typically labor intentices to contain environmental problems that are sive, both seasonally and in total, and this can now common in many intensively farmed areas. constrain their use. Fellows and green manures The more efficient use and better management of also keep land out of crop production, and comexternal inputs such as chemical fertilizer, irriga- posting and manuring compete with household tion water, and/or land-use diversification may energy use for scarce organic matter. The challead to higher productivity and reduced environ- lenge is to develop LEI technologies that boost mental degradation. Such knowledge-intensive both labor and land productivity. farming will require improved extension systems Although poorer LFAs need improved techand greater investment in farmer education. nologies for food crops for subsistence and local
Focusing on staple foods production can be needs, sustained increases in per capita income win-win for growth and poverty reduction when will hinge on diversification into higher-value agrithe research benefits regions that have a compar- cultural products and nonfarm activities. ative advantage in these crops. Countries that lack Successful examples include dairying in India, comparative advantage in foodgrains but that growing olives in North Africa, and raising hortinevertheless pursue self-sufficiency in their produc- cultural crops in parts of Central America (Hazell, tion may achieve greater national immunity from Jagger, and Knox 2000). the vagaries of world food markets, but at the cost A key lesson from many past research investof economic growth opportunities. Such opportu- ments in LFAs is the importance of social and instinity costs are smaller when the foodgrains are tutional constraints, particularly the effectiveness grown in LPAs rather than HPAs, and the trade-off of indigenous property rights systems and local may be worthwhile if this leads to significant gains capacity for organizing and sustaining collective in food security in those regions. action for managing natural resources. Figure 5

Figure 5-Links between property rights, collective action, and technology adoption
* Social forestry
* Watershed development
0 0 Integrated pest management
u Soil erosion
0 Farm trees
0 High-yielding varieties
Security of property rights
Source: Knox, Meinzen-Dick, and Hazell 1998.
(taken from Knox, Meinzen-Dick, and Hazell Integrated pest management, for example, 1998) plots increasingly secure property rights on requires that all farmers in an area work togeththe horizontal axis and increasing levels of collec- er; the technology is not effective if some farmers tive action on the vertical axis. Some of the most spray indiscriminately or if planting dates are not successful agricultural technologies lie close to the synchronized. However, the returns are relatively origin in this figure. For example, the benefits of quick, so secure property rights are less of an high-yielding cereal varieties (HYVs)-the lynch- issue. For these reasons, IPM appears in the pin of the Green Revolution--could be captured upper left corner of Figure 5. In contrast, planting within a single agricultural season and hence did trees on farms (agroforestry) is a long-term investnot require secure property rights. Even share- ment that requires secure property rights. But cropping tenants with single-season leases were since trees can be planted by individual farmers able to adopt these technologies. Moreover, since regardless of what their neighbors do, "farm individual farmers could adopt regardless of what trees" appears in the lower right-hand corner. But their neighbors decided to do, collective action many other technologies for improved NRM was not necessary. These features made adoption require both secure property rights and effective decisions relatively simple, and they help explain collective action and therefore appear in the why Green Revolution technologies spread so upper right-hand quadrant. Watershed developquickly and widely despite considerable diversity ment, for example, requires secure property rights in local socioeconomic conditions, because it involves long-term investments in check
But where research agendas must focus on dams, land contouring, and tree planting in water the sustainable use of natural resources, local catchment areas, and it can be successfully done institutional issues become much more prominent. only if the entire community living within the rele-

vant landscape is mobilized to support collective contribute more to agricultural growth per action. If these institutional conditions are not met, research dollar. Growing conditions are seen as the technology is not likely to be adopted and diverse in LFAs, and so improved technologies maintained, regardless of its profitability and sci- may not have widespread application (in contrast entific soundness. to Green Revolution technologies that spread over
Difficulties arise because institutions for prop- tens of millions of hectares of irrigated land). erty rights and collective action are rarely formal- Technologies are also perceived to be more diffiized in developing countries, and communities cult and perhaps more costly to develop. There is vary widely in their ability to organize and sustain undoubtedly some basis for these concerns, espesuch institutions on their own. Socioeconomic cially for commodity improvement research. On research has much to contribute toward our the other hand, recent evidence from India and understanding of local institutions and the condi- China shows that, dollar for dollar, agricultural tions under which they are likely to be effective, research investments in LFAs can outperform Without such knowledge, and a corresponding investments in HPAs in productivity growth and ability on the part of policymakers to intervene poverty reduction in some types of less-favored and to strengthen local institutions when neces- areas (Fan, Hazell, and Haque 2000; Fan, sary, many promising NRM technologies are not Zhang, and Zhang 2000). likely to be adopted. As should be expected, the returns to addiNRM research for less-favored lands should tional research investments vary by agroclimatic build on farmers' own indigenous knowledge and zone and are very low for both growth and poverpractical innovations. Some NGOs have been ty in some of the most marginal zones. But the successful in pursuing this agenda and in working range of conditions under which research investwith local communities to overcome social and ments have significant impact is impressive (Fan, institutional constraints. The more effective linking Hazell and Haque, 2000). One reason for these of formal research to these kinds of grassroots results is that India and China have already invest. development activities could lead to real improve- ed heavily in their irrigated and high-potential ments in the relevance and uptake of much NRM rainfed areas, and productivity growth has slowed research. But sustained growth in on-farm produc- in many of these regions. With diminishing returns tivity will also require improvements in crop in HPAs and relatively little research investment in germplasm and disease and pest control. Without many LFAs, it is not surprising that the latter should continuing improvements in knowledge and genet- now give higher returns to research investments on ic resources in less-favored areas, productivity will the margin. But care must be taken in extrapolatquickly stagnate again. The potential to breed ing these results to other countries that have investhigher-yielding varieties in many less-favored ed a great deal less in both HPAs and LFAs, for lands may be constrained by the low-yield poten- example, much of Sub-Saharan Africa. In these tial of the existing range of genetic material, cases there could well be some important tradeExisting material has often been selected by man offs between sectorwide growth and regional and nature for robustness under harsh and risky poverty reduction from targeting research growing conditions. Transgenic biotechnology resources on the problems of LFAs. may prove an essential tool for achieving the One way to reduce the cost of research in wider crosses that will be needed (IFAD 2000). LFAs is to focus on NRM problems that are comMany agricultural researchers are skeptical mon to a significant number of poor people-but about the efficiency of research in less-favored only problems that can be scaled up from benchareas, arguing that it is better to invest in HPAs mark sites. The scaling up need not mean that all because these give higher returns to research and sites have to be homogenous, just that improved

NRM practices can be easily adapted by local aping cash crops for export or expanding opporpeople and institutions to different site-specific cir- tunities for seasonal migration to cities. Other cumstances. The research also should involve rele- income-augmenting measures include creation of vant socioeconomic work to understand the social rural processing facilities to enable higher-valueand economic constraints on adopting improved added from agricultural output. NRM practices, and partnerships need to be Targeting small rather than large farms will
formed with institutions with the capacity to over- complement agricultural sector growth if small come these constraints at the grassroots level, farms are at least as economically efficient as large farms. An impressive body of empirical evidence confirms that land productivity is inversely
Research Strategies for related to farm size in many developing countries
Smallholder Farmers (see Heltberg 1998 far a recent review). One reason for this is that hired labor is less efficient and
As farms get smaller, it becomes increasingly more costly to manage than family labor, giving important to develop technologies that increase smaller farms a competitive advantage. Another their labor as well as their land productivity, reason may be the higher management intensity Otherwise, they will simply be working hard-er that is possible on smaller farms. But because and harder to achieve the same level of per capi- small farms use more labor-intensive technologies, ta income. their average labor productivity is also lower. As
In order to maintain their incomes, smallhold- countries develop and agriculture must compete er farms will need to diversify beyond traditional with higher-paying nonfarm employment, mechacrop production into profitable activities that are nization and more capital-intensive farming methcompatible with their diminishing land/labor ods become important to increase labor ratios. Countries with rising per capita incomes productivity. This requires that farms become bigoffer expanding opportunities to diversify into live- ger. There is little hard information about when stock and horticultural production to meet rapidly this crossover occurs, and it is likely to vary with growing domestic demands. These activities typi- the rate of rural population growth (which slows cally give high returns per unit of land and are the transition), with credit subsidies and other agrilabor-intensive, and hence well suited to small- cultural support policies (which often encourage farm conditions. Such diversification is already premature mechanization), and with opportunities happening in many countries, especially in Asia for income diversification into nonfarm activities in and Latin America. Rising per capita incomes are rural areas (which can have mixed effects also associated with expanding rural nonfarm because greater opportunities encourage contineconomies, which offer opportunities for small- ued part-time farming by small farmers but also holder farm families to diversify into nonfarm raise wage rates). Nor does a crossover necesemployment or into nonfarm businesses. In sarily occur for some employment-intensive types Southeast Asia, small farmers seem to be follow- of farming-for example, high-value horticultural ing the development pattern that occurred in crops and intensive livestock production. It seems Japan: becoming part-time farmers with regular likely that the crossover is important for foodgrain employment in nonfarm, often industrial, occupa- production by the time countries reach middletions (Rosegrant and Hazell 2000). income status, but small farms may retain their
The opportunities for income-enhancing diver- competitive edge indefinitely for some types of sification are much more constrained in countries horticultural and livestock forming. with low and stagnant per capita incomes. In Another concern is that, with increasing rural these cases, attention needs to be given to devel- populations but scarce land, small farms may

eventually become too small and fragmented to be reforms in many countries may discourage inapefficient. Carter and Wiebe (1990) have provided propriate investments in mechanization and help evidence from Kenya showing that profits per shift agriculture toward more employment-intensive hectare decline when farms get too small. It is patterns of growth, helping landless and near-landpremature to draw any definite conclusions about less farmers. Increased investments in improved this concern, and, even if true, it is likely that small natural resource management in less-favored areas farms first lose their competitive advantage for should also help create additional employment growing foodgrains rather than labor-intensive because many of these investments are labor intenlivestock and horticultural products. sive. In countries where rising per capital incomes
Even though small farms are generally the are spurring demand far livestock and horticultural more efficient producers, this advantage may be products, agricultural employment is likely to grow offset by problems in reaching markets. Because as farms diversify into these activities. small farms typically trade only in small quantities, The landless also can benefit from types of both inputs and outputs, they cannot command the livestock production where animals are kept in same price advantages as larger farms dealing in stalls or cages or fed from common property greater bulk. They are also less likely to have the grazing resources and purchased feeds (for same access to market information and contacts. example, poultry, rabbits, goats, and dairy Such disadvantages can be offset by coopera- cows). Rehabilitation and better management of tives, contract farming, and other marketing common property resources in general can also arrangements, but these often introduce addition- help the landless, since the use of these resources al costs and challenges. Resolving such marketing is often an important element in their livelihood problems--once the mandate of now-disbanded strateg ies. marketing parastatals-may be critical to achiev- Increasing employment and labor productivity ing greater complementarity between targeting through improvements in labor and capital marsmallholder farms and agricultural productivity kets is also good for economic efficiency and growth. offers win-win opportunities. Targeting research
on the own-agricultural activities of landless workers probably adds little to overall agricultural
Research Strategies for growth rates in the short term, but it could lead to
Landless Laborers longer-term growth by increasing worker productivity and the economic contributions of their cii
Landless laborers are difficult to reach through agri- dren (see next section). cultural growth except through the labor market
(both locally and through inter-regional migration)
and the rural nonfarm economy. In irrigated and Research Strategies for More high-potential areas, agriculture's employment elas- Nutritious Foods ticity has shrunk in the post-Green Revolution era in
many countries. In India, for example, the employ- Conventional research approaches can help ment elasticity has fallen from about 0.7 in the improve productivity and lower the price of non1970s to about half that level today. This means staple crops that are rich in micronutrients. This that a 1 percent increase in agricultural output potentially powerful indirect effect should be serigenerates a much smaller percentage of addition- ously considered in regions that experience high al employment than before. The recent removal of micronutrient deficiency and are poorly linked to credit subsidies and policy distortions as part of international or domestic markets. The following structural adjustment and market liberalization approaches should be considered.

New Technology for Small-Scale International Potato Center (CIP) with sweet potaHome Gardening of Micronutrient- toes. A series of papers have been written that
Rich Food describe the technical and economic feasibility of
A recent review of food-based interventions to deriving and using sweet potato chips and flours combat vitamin A deficiency points to the potential in chapati and bread processing, as well as conof home gardening combined with promotional sumer acceptance (Hagenimana and Owori and education interventions. With regard to iron, 1997a and 1997b). The goal of this work is mainproduction and education interventions to ly to reduce the costs of production and the cost to increase the supply and intake of iron from plant the consumer. Sweet potatoes performed well in foods have not been as popular as those for vita- every area. The successful application of such min A. Experience with food-based approaches to techniques to orange- and yellow-flesh varieties increase production and/or consumption of heme that are richer in beta-carotene has directly helped or non-heme iron-rich foods is limited, but some improve nutrition in Kenya (Hagenimana and lessons are clear. In addition to the well-known Oyunga 1999). problems of bioavailability with iron from plant
sources, the experience with animal production Breeding Techniques to Improve the suggests trade-offs between increased income Nutrient Content of Staples from selling home-produced animal products and There are several ways to improve the nutrient increasing one's own consumption of these prod- availability of food staples: (1) increase the densiucts to improve dietary quality. Similar to home ty of micronutrients in the grain, (2) decrease the gardening interventions, a strong nutrition educa- density of factors that inhibit human absorption of tion component is critical to achieving improved micronutrients already in the grain, and (3) dietary diversity through animal production inter- increase the level of promoters of human absorpventions. tion. Breeding methods can be conventional or
Postharvest Technologies The CGIAR micronutrients project involving
Postharvest activities can affect nutrient availabili- the International Rice Research Institute (IRRI), the ty by increasing the general use of nutrient-rich International Center for Tropical Agriculture foods (for instance, beta-carotene-rich varieties of (CIAT), CIP, and IFPRI is beginning to produce sweet potatoes), increasing the nutrient density of some promising results (Bouis, Graham, and foods consumed by infants, and decreasing nutri- Welch 2000). First, two high-yielding, high-iron ent losses from the processing of widely available rices were identified among improved lines foods. Postharvest activities include storage, com- already being tested by IRRI and now being submercial processing, in-home processing, and jected to human feeding trials. Second, 24 selectpreparation. Food processing includes physical ed genotypes of beans from CIAT were found to processes (heat/cold treatment, mechanical sepa- have substantial variation in iron concentration ration such as milling, and reduction of water and a constant level of bioavailability. The develactivity), chemical processes (addition of acid, opment of low phytic acid mutants of maize and alkaline, oxidizing, and reducing agents), enzy- other cereals for use in food and feed is another matic processes (hydrolysis of proteins and inacti- approach to improving the micronutrient content vation of toxins), and biological processes of staples. An important advantage of low phytic (fermentation and germination) (WHO 1998). A acid mutants is that the bioavailability of a range good example of work that has the potential to of minerals may be improved. Yields of the best increase the general use of nutrient-rich foods via low phytic acid lines, first developed in the midprocessing is provided by the work of the 1990s, now range between 5 and 15 percent

below those of the highest-yielding commercial than low protein or lysine consumption in South varieties. Because of the benefits for nutrition, low Asia's semi-arid tropical areas. As a result, breedphytic acid crops may have higher total benefits to ing for high yield was the best way to help poor the poor, although the rate of adoption will remain people and there was no trade-off between effia drawback as long as yields are compromised ciency and poverty alleviation goals. (Raboy 2000). However, this is not so clearly the case in
Enhancing the nutrition status of the poor will breeding for greater micronutrient content, such as lead to additional productivity growth (both in iron and zinc, for regions where poor people sufagriculture and other activities) today and in future fer from serious deficiencies. Such breeding work generations. Undernutrition involves serious eco- runs the risk of drawing resources away from nomic costs. Both stunting and anemia are known other objectives such as strengthening tolerance to to lower productivity in physical labor. Adults who various stresses. Even so, the cost is likely to be are moderately stunted are 2 to 6 percent less pro- much lower (at least for iron and zinc) than for a ductive, and those who are severely stunted are 2 capsule program that has to be sustained on a to 9 percent less productive, than those of normal long-term basis (Ruel and Bouis 1998). stature (Horton 1999, using estimates in Haddad
and Bouis 1991 and Alderman et al. 1997). Iron
deficiency anemia is associated with a loss of pro- Research Methods that ductivity of 5 percent in light blue-collar work and Empower the Poor 17 percent in heavy manual labor (Ross and
Horton 1998). Participatory research, in which poor people play
Evidence suggests that childhood stunting and a role in setting the research agenda and carrying iron deficiency, as well as maternal iodine defi- it out, has potential to make agricultural research ciency, are also associated with lower cognitive more effective in empowering poor people. It can outcomes in children and hence lower adult pro- give them more influence over the research system ductivity in the next generation. The effects on to address their needs and provide them with the adult earnings and productivity is estimated at 10 skills needed to solve many of their own problems. percent for low-height-for-age (from an older study Participatory research is still new, and there by Selowsky and Taylor 1 973), 4 percent for has been little attempt to assess its impact on agrichildhood anemia (Ross and Horton 1998), and cultural productivity or poverty reduction. Initial 10 percent per child born to a mother with goiter attempts at participatory research have given (Ross 1997). These losses are conservatively 2 to greater priority to involving poor people in evalu3 percent of GDP in low-income countries. In eating new technologies (such as specific traits of South Asia, the estimated losses associated with improved varieties) than in setting priorities for iron deficiency alone are estimated to be $5 bil- research itself (such as which kinds of varieties to lion per year (Ross and Horton 1998). develop, and for what types of crops, regions,
Technologies that improve the nutritional con- and farmers). Involving farmers in problem diagtent of the foods that poor people eat need not nosis and field-testing of technologies can provide involve trade-offs with other desirable traits. useful information to researchers and result in Researchers at the International Crops Research more useful products for farmers (Farrington and Institute For the Semi-Arid Tropics (ICRISAT), for Martin 1988). Using participatory research to example, concluded some years ago that breed- promote empowerment as a direct goal requires ing for high protein and lysine content in food working with communities that are organized and grains was not a priority because calorie con- skilled in working together, solving problems, and sumption was a more severe problem for the poor resolving conflicts. Kerr and Kolavalli (1999)

argue that "participatory research should go both researchers' perceptions and the hand-in-hand with participatory community devel- incentives that guide them. Another conopment that can help improve access to credit and straint is that initial costs of participatory markets and can teach local people the skills they research can be high because travel and need to organize themselves, analyze and solve training budgets rise. In a project to develproblems as a group, and resolve conflicts." op pest control measures in Ghana,
Increasing pressures on research organiza- farmer participation increased project tions to improve their effectiveness in reaching the costs by 66 percent and accounted for 80 poor will lead to increasing collaboration with percent of researchers' time (Magrath et farmers. But Kerr and Kolavalli (1999, p. 137- al. 1997). However, the Rwanda case 138) identify several constraints: shows that the extra cost can lead to higher returns by reducing the time needed to
More participatory research requires mul- identify promising technologies. The cost tidisciplinary work that may be difficult to per variety released actually fell in the organize; this was a constraint in earlier Rwanda case (Sperling et al. 1993). Also, farming systems research (Farrington and developing the farmers' own capabilities Martin 1988). Some scientists are reluc- in developing improved pest management tant to learn from indigenous knowledge, systems, or conducting field trials, or even and economists shy away from participa. breeding can be a cost-ffective way to tory methods because they do not always adapt technology to local needs where
yield quantitative data. Scientific journals conditions are spatially diverse. Where are less receptive to research based on participation may mean the difference
participatory than traditional methods. As between success and failure in developing a result, major changes are still needed in technologies, there are no cost trade-offs.

6. The Role of Public Research
and Extension Systems
Public research and extension systems will have a ceilings had already been reached and were in! key role in implementing the pro-poor agricultural need of basic biological breakthroughs, and research strategy described in the previous chap- because crops were grown in monoculture syster. Because targeting the poor is not always win- tems under homogenous conditions with good win for growth and poverty alleviation, countries access to markets, the new technologies could be that can afford large research budgets, or enjoy widely adopted with little if any local adaptation. significant private-sector investment in productivity But as discussed in previous chapters, the enhancing research, will be best placed to under- challenges facing public research and extension take this agenda. Where the trade-offs are high, it systems today are rather different. In irrigated and is appropriate to consider alternative policies For high-potential areas, crop diversification and poverty alleviation. Technology is only one instru- improved environmental management have ment for helping the poor, and it is not always the become key challenges for small and large farms most effective one. Its role must be seen within the alike, and the problems facing less-favored lands broader context of rural development and grass- are much more diverse and location-p and roots development efforts. involve changes in complex natural resource manWhere productivity trade-offs exist but are not agement systems that have been developed over overwhelming, targeting research can sometimes generations to cope with uncertain rainfall and offer more cost-effective and long-term solutions to weather conditions, poor and often fragile soils, poverty than alternative interventions. For exam- and the high costs of external inputs, given poor pie, research to improve soil nutrient and water market access. Moreover, research needs to be management practices or to develop more undertaken in more participatory ways if it is to
drought-tolerant cereal varieties might lead to sub- 'become more effective in empowering the poor. stantial long-term savings in relief aid in many At the same time, public research systems also poor, drought-prone areas. In these cases, the need to become more engaged in biotechnology trade-off against agricultural productivity growth research to ensure that at least part of its potential may be less important than the cost savings is used to enhance agricultural growth and to help achieved in alternative relief programs, especially poor farmers. if government and donors can be persuaded to To meet these challenges, there is a need for allocate some of those savings to additional agri- a more client-oriented, probtem-sotvina approach cultural research investments. throughout the agricultural research system, an
If public research and extension systems are to approach not limited to a particular kind of techbe effective in undertaking targeted pro-poor nology or a particular type of agriculture or zone. research, they will also have to make some insti- This approach will often translate into a need for tutional adjustments. Their past successes lay in more on-farm research under conditions that are developing and spreading Green Revolution tech- difficult and diverse. Not all of the technological nologies in high-potential areas. Because yield challenges facing poor people will be solved by

more on-farm work; biotechnology conducted in a their farming conditions. Participatory research strict laboratory environment may be critical, for methods are proving to be a fruitful way of
example, in raising yield ceilings or for improving achieving this goal (Kerr et al. 2000).
drought tolerance. However, even biotechnology Because of the public-goods nature of much will be more effective if it addresses priorities agricultural research and extension, market forces based on a client-oriented, problem-solving alone will be insufficient to integrate the roles of approach that draws many of its insights from the public and private sectors and NGOs and interaction with farmers. make them fully accountable to their clients. New
J Institutional reforms are necessary to change institutional mechanisms will also be required.
incentive structures within public research and Competitive research grants offer one interesting extension systems so that scientists and extension approach, particularly if they are open to all releofficers are responsive to the needs of their clients. vant public, private, and NGO agencies, and if But to be effective, these changes will need to farmers (including poor farmers and women) are extend to all levels of management (Byerlee and represented in the decisionmaking process Alex 1998). The kinds of changes needed in pub- (Farrington and Martin 1998). Local research lic agricultural research and extension systems will committees composed of relevant stakeholders also require the forging of new partnerships show some promise as a device for setting between the public system and NGOs, private- research priorities for the public sector (Alsop
sector firms, and farmers. 1998).
Private seed companies and input suppliers Even as agricultural research and extension
are playing a larger role as many countries liber- systems are being asked to take on more diverse alize and privatize their agricultural input mar- and difficult challenges, their budgets are being kets. Many of these companies not only develop cut in many countries (Pardey and Beintema improved products of their own (including under- 2001). National agricultural research systems taking agricultural research), but also advise (NARS) in Africa have been particularly affected, farmers about the use of products they sell. and the availability of resources per scientist has NGOs have also become important actors in fallen sharply. On average, developing countries spreading natural resource management prac- spend about one half of 1 percent of their agritices regarding soil and water management, cultural GDP on public research, which is much watershed development, and social forestry. They less than the 2 percent averaged by the industrihave a particular advantage in helping commu- alized countries. If new technologies are to be nities take collective action to implement developed to address the poverty and environimproved natural resource management practices mental problems of developing countries, and to at the landscape level (Kerr et al. 2000). enable them to capture some of the potential benFarmers, including smallholder and women farm- efits of biotechnology, then there is an urgent need ers, also need to be actively inyaoyed-in-the to increase available funding and to implement
design and evaluation of research intenrid for needed institutional reforms.

ACC/SCN (United Nations Administrative Bell, C., P.B.R. Hazell and R. Slade. 1982. Project
Committee on Coordination/Sub-Committee on evaluation in regional perspective: A study of an Nutrition). 2000. Fourth report on the world irrigation project in northwest Malaysia.
nutrition situation. Geneva: ACC/SCN in col- Baltimore, MD.: Johns Hopkins University Press.
laboration with IFPRI. Blyn, G. 1983. The Green Revolution revisited.
Adelman, I. 1984. Beyond export-led growth. Economic Development and Cultural Change
World Development 12 (9): 937-949. 31 (4): 705-725.
Ahluwalia, M. S. 1978. Rural poverty and agricul- Bouis, H.E. 2000. Development of modem varieties of
tural performance in India. Journal of rice: Impacts on food security and poverty. Paper
Development Studies 14 (3): 298-323. prepared For presentation at the International Rice
Alderman, H., J. Behrman, V. Lavy, and R. Menon. Research Institute Conference, Los Banos, Laguna,
1997. Child nutrition, child health, and school Philippines, March 31 -April 3.
enrollment. Policy Research Working Paper Bouis, H., R. Graham, and R. Welch. 2000. The 1700. Washington, D.C.: World Bank. CGIAR Micronutrients Project: Justification and
Alsop, R. 1998. Coalitions and the organization of objectives. Food and Nutrition Bulletin 21 (4):
multiple-stakeholder action: A case study of 374-381.
agricultural research and extension in Byerlee, D., and G. E. Alex. 1998. Strengthening Rajasthan, India. Environment and Production national agricultural research systems: Policy Technology Division Discussion Paper No. 34. issues and good practice. Washington, D.C.:
Washington, D.C.: IFPRI. World Bank, Rural Development Department.
Alston, J., G. Norton, and P. Pardey. 1995. Science Carney, D., ed. 1998. Sustainable rural livelihoods:
under scarcity: Principles and practice for agri- What contribution can we make? London:
cultural research evaluation and priority setting. Department for International Development.
Ithaca, N.Y.: Cornell University Press. Carter, M. R., and K. D. Wiebe. 1990. Access and Asian Development Bank. 2000. RuralAsia: Beyond capital as impact on agrarian structure and prothe Green Revolution. Manila. ductivity in Kenya. American Journal of
Bardhan, P. K. 1973. On the incidence of poverty in Agricultural Economics 72 (5): 1146-1150.
rural India in the sixties. Economic and Political CGIAR (Consultative Group on International Weekly 4, 5, and 6 (February). Agricultural Research)/Technical Advisory
Bebbington, A. 1999. Capitals and capabilities: A Committee (TAC). 2000. CGIAR research priframework for analyzing peasant viability, rural orities for marginal lands. Washington, D.C.
livelihoods and poverty. World Development Conway, G. 1997. The Doubly Green Revolution.
27 (12): 2021-2044. London: Penguin Books.

Daft, G., and M. Ravallion. 1997. Macroeconomic research reports. World Development 23 (2):
crises and poverty monitoring: A case study for 265-279.
India. Review of Development Economics 1 (2): Gaiho, R. 1989. Poverty, agricultural production 135-152. and price fluctuations in rural India: A reformuDavid, C., and K. Otsuka, eds. 1994. Modern rice lation. Cambridge Journal of Economics 13 (2):
technology and income distribution in Asia. 333-352.
Boulder, Colo., U.S.A.: Lynne Reinner. Ghose, A. K. 1989. Rural poverty and relative De Franco, M., and R. Godoy. 1993. Potato-led prices in India. Cambridge Journal of
growth: The macroeconomic effects of techno- Economics 13 (2): 307-331.
logical innovations in Bolivian agriculture. Gibb, A. 1974. Agricultural modernization, nonJournal of Development Studies 29 (3): farm employment and low-level urbanization: A
561-587. case study of a Central Luzon sub-region. Ph.D.
Fan, S., and P. Hazell. 2000. Should developing thesis, University of Michigan, Ann Arbor.
countries invest more in less-favored lands? An Graham, R. D., and R. M. Welch. 1996. Breeding empirical analysis of rural India. Economic and for staple food crops with high micronutrient Political Weekly 35 (17): 1455-1464. density. Agricultural Strategies for
Fan, S., P. Hazell, and T. Haque. 2000. Targeting Micronutrients Working Paper 3. Washington,
public investments by agroecological zone to D.C.: IFPRI.
achieve growth and poverty alleviation goals in Griffin, K. 1974. The political economy of agrarian
rural India. Food Policy 25 (4): 411-428. change. London: Macmillan.
Fan S., P. Hazell, and S. Thorat. 1999. Linkages Griffin, K., and A. K. Ghose. 1979. Growth and
between government spending, growth and impoverishment in the rural areas of Asia.
poverty in rural India. Research Report 110. World Development 7 (4/5): 361-384.
Washington, D.C.: IFPRI. Haddad, L., and H. Bouis 1991. The impact of nutriFan, S., L. Zhang, and X. Zhang. 2000. Growth tional status on agricultural productivity: Wage
and poverty in rural China: The role of public evidence from the Philippines. Oxford Bulletin
investments. Environment and Production of Economics and Statistics 53 (1): 45-68.
Technology Division Discussion Paper No. 66. Haddad, L., J. Hoddino, and H. Alderman, eds.
Washington, D.C.: IFPRI. 1997. Intrahousehold resource allocation:
Farmer, B. H. 1977. Green Revolution? Technology Models, methods, and policy. Baltimore, Md.:
and change in rice growing areas of Tamil Johns Hopkins University Press for IFPRI.
Nadu and Sri Lanka. London: Macmillan.
Hagenimana, V., and C. Owori. 1997a. Sweet Farrington, J., and A. Martin. 1988. Farmer partici- potatoes in Chapatis processing: Feasibility and
pation in agricultural research: A review of con- acceptability in rural areas. Journal of Food
cepts and practices. Agricultural Administration Technology in Africa 2 (1) 4-8.
Unit Occasional Paper 9. London: Overseas Hagenimana, V., and C. Owen. 19971. Feasibility, Development Institute. acceptibility and production costs of sweet potaFrankel, F. R. 1976. India's Green Revolution: to baked products in Lira Municipality, Uganda.
Economic gains and political costs. Princeton, In Proceedings of the Fourth Triennial Congress NJ.: Princeton University Press. of the African Potato Association. Pretoria,
Freebairn, D. K. 1995. Did the Green Revolution South Africa: African Potato Association.
concentrate incomes? A quantitative study of

Hagenimana, V., and M. A. Oyunga. 1999. The Evidence from Pakistan. World Development
effects of women farmers' adoption of orange- 26 (10): 1807-1826.
fleshed sweet potatoes: Raising Vitamin A Horton, S. 1999. Opportunities for investments in intake in Kenya. Research Report Series No. 3. nutrition in low-income Asia. Asian
Washington, D.C.: International Center for Development Review 17 (1, 2): 246-273.
Research on Women. IFAD (International Fund for Agriculture). 2001.
Haggblade, S., and P. Hazell. 1989. Agricultural Rural poverty report 2001: The challenge of
technology and farm-nonfarm growth linkages, ending rural poverty. Oxford: Oxford University
Agricultural Economics 3 (4): 345-364. Press.
Haggblade, S., P. Hazell, and J. Brown. 1989. ILO (International Labor Organization). 1977.
Farm-nonfarm linkages in rural Sub-Soharan Poverty and landlessness in rural Asia. Geneva.
Africa. World Development 17 (8): Jazairy, I., M. Alamgir, and T. Panuccio. 1992. The
state of wodd rural poverty: An inquiry into its
Hazell, P., and S. Haggblade. 1991. Rural-urban causes and consequences. Rome: International
growth linkages in India. Indian Journal of Fund for Agricultural Development.
Agricultural Economics 46 (4): 515-529. Kerr, J., and S. Kolavalli. 1999. Impact of agriculHazell, P., and S. Haggblade. 1993. Farm-nonfarm tural research on poverty alleviation:
growth linkages and the welfare of the poor. In Conceptual framework with illustrations from Including the poor, ed. M. Lipton and J. van der the literature. Environment and Production Gaag. Washington, D.C.: World Bank. Technology Division Discussion Paper No. 56.
Hazell, P., and C. Ramasamy. 1991. The Green Washington, D.C.: IFPRI.
Revolution reconsidered: The impact of high- Kerr, J., G. Pangare, V. L. Pangare, and P. J.
yielding rice varieties in South India. Baltimore, George. 2000. Sustainable agriculture and natMd.: John Hopkins University Press for IFPRI. ural resource management in India's semi-arid Hazell, P., and T. Reardon. 1998. Interactions tropics. In Tradeoffs or synergies? Agricultural
among the rural nonfarm economy, poverty, intensification, economic development and the and the environment in resource-poor areas. environment, ed. D. R. Lee and C. B. Barrett.
Paper presented at an IFPRI/World Bank-spon- Wallingford, U.K.: CAB International.
sored workshop on strategies for stimulating Knight, J., and L Song. 1992. Income inequality in growth of the rural nonfarm economy in devel- rural China: Communities, households and hoping countries, May, Warrenton, Va., U.S.A. resource mobility. Oxford Applied Economics Hazell, P., P. Jagger, and A. Knox. 2000. Discussion Paper Series No. 150. Oxford:
Technology, natural resources management and University of Oxford Institute of Economics and
the poor. Background paper prepared for Statistics.
International Fund for Agricultural Knox, A., R. Meinzen-Dick, and P. Hazell. 1998.
Development's Rural Poverty Report 2000 and Property rights, collective action and technoloavailable in The Report of IFAD's Workshop on gies for natural resource management: A conRural Poverty, Rome, January 24-25. Rome: ceptual framework. CGIAR System-wide IFAD. Program on Collective Action and Property
Heltberg, R. 1998. Rural market imperfections and Rights (CAPRI) Working Paper No. 1.
the farm size-productivity relationship: Washington, D.C.: IFPRI.

Lipton, M., with R. Longhurst. 1989. New seeds and Raboy, V. 2000. Research on bioavailability: Low
poor people. Baltimore, Md.: Johns Hopkins phytic-acid grains. Food and Nutrition Bulletin
University Press. 21 (4): 423-427.
Magrath, P. 1997. Cost benefit analysis of client Rangarajan, C. 1982. Agricultural growth and
participation in agricultural research: A case industrial performance in India. Research
study from Ghana. ODI Agricultural Research Report 33. Washington, D.C.: IFPRI.
and Extension Network Paper No. 74. London: Reardon T., K. Stamoulis, M. E. Cruz, A. Balisacan, Overseas Development Institute. J. Berdugue, and K. Savadogo. 1998.
Mellor, J. W. 1976. The new economics of growth: Diversification of household incomes into nonA strategy for India and the developing world farm sources: Patterns, determinants and effects.
Ithaca, N.Y.: Cornell University Press. Paper presented at an IFPRI/World Bank-sponOwens, T., and J. Hoddinott. 1998. Investing in sored workshop on strategies for stimulating
development or investing in relief: Quantifying growth of the rural nonfarm economy in develthe poverty tradeoffs using Zimbabwe house- oping countries, May, Warrenton, Va., U.S.A.
hold panel data. Washington, D.C.: IFPRI. Robinson, S., T. L. Roe, and A. E. Yelden. 1998.
Photocopy. Macroeconomic policies and the rural nonfarm
Pardey, P., and N. Beintema. 2001. Losing ground: economy: A general equilibrium investigation
What's happened with agricultural research of alternative development strategies in an regarding less-developed countries? archetype model for Africa and South Asia.
Washington, D.C.: IFPRI. Photocopy. Paper presented at an IFPRI/World Bank-sponsored workshop on strategies for stimulating
Pearse, A. 1980. Seeds of plenty, seeds of want. growth of the rural nonfarm economy in develLondon: Oxford University Press. hoping countries, Warrenton, Va., U.S.A.
Pingali, P. L., and M. W. Rosegrant. 2001. Intensive Rosegrant, M., and P. Hazell. 2000. Transforming
food systems in Asia: Can the degradation be Rosegrant, M., and P. Hazell. 2000. Transforming food systems in Asia: Can the degradation be the rural Asia economy: The unfinished revolureversed? In Tradeoffs or synergies? Agricultural tion. Hong Kong: Oxford University Press for
intensification, economic development and the the Asian Development Bank.
environment, ed. D. R. Lee and C. B. Barrett.
Wallingford, U.K.: CAB International. Ross, J. S. 1997. Calculating the effects of malnutrition on economic productivity and survival.
Pingali, P. L., M. Hossain, and R. V. Gerpacio. PROFILES guidelines. Washington, D.C.:
1997. Asian rice bowls: The returning crisis? Academy for Educational Development.
Wallingford, U.K.: CAB International in association with the International Rice Research Ross, J. S., and S. Horton. 1998. Economic conseInstitute. quences of iron deficiency. Ottawa:
Micronutrient Initiative.
Pinstrup-Andersen, P., and P. Hazell. 1985. The Micronutrient Initiative.
impact of the Green Revolution and prospects Ruel, M., and H. Bouis. 1998. Plant breeding: A for the future. Food Reviews Intemrnational 1 (1): long-term strategy for the control of zinc defi1-25. ciency in vulnerable populations. American
Journal of Clinical Nutrition 68 (2s):
Pinstrup-Andersen, P., R. Pandya-Lorch, and M. 488s-494s.
Rosegrant. 1997. The world food situation:
Recent developments, emerging issues, and Saith, A. 1981. Production, prices and poverty in long-term prospects. 2020 Vision Food Policy rural India. Journal of Development Studies 19
Report. Washington, D.C.: IFPRI. (2): 196-214.

Scobie, G. M., and R. Posada. 1978. Technical selection in Rwanda. Experimental Agriculture
change and income distribution: The case of 29 (4): 509-519.
rice in Columbia. American Journal of Tripp, R. 1990. Does nutrition have a place in agriAgricultural Economics 60 (1): 85-92. cultural research? Food Policy 15 (6):
Selowsky, M., and L. Taylor. 1973. The economics 467-474.
of malnourished children: An example of disin- Walker, T., and J. Ryan. 1990. Village and housevestment in human capital. Economic hold economies in India's semi-arid tropics.
Development and Cultural Change 22 (1): Baltimore, Md.: Johns Hopkins University Press.
17-30. Welch, R., and R. Graham. 2000. A new paradigm
Sharma, M., M. Garcia, A. Qureshi, and L. Brown. for world agriculture: Productive, sustainable,
1996. Overcoming malnutrition: Is there an nutritious, healthful food systems. Food and
ecoregional dimension? Food, Agriculture, and Nutrition Bulletin 21 (4): 361-366.
the Environment Discussion Paper 10. Westley, J. R. 1986. Agriculture and equitable Washington, D.C.: lFPR. growth: The case of Punjab-Haryana. Westview
Smith, L. C., and L. Haddad. 2000. Overcoming Special Studies in Agriculture Science and
child malnutrition in developing countries: Past Policy. Boulder, Colo., U.S.A.: Westview Press.
achievements and future choices. Food, World Bank. 2000. World development report Agriculture, and the Environment Discussion 2000/2001: Attacking poverloy Washington,
Paper 30. Washington, D.C.: IFPRI. D.C.
Spencer, D. 1994. Infrastructure and technology World Health Organization (WHO). 1998.
constraints to agricultural development in the Complementary feeding of young children in humid and subhumid tropics of Africa. developing countries: A review of current scienEnvironment and Production Technology Geneva.
Division Discussion Paper No. 3. Washington,
D.C.: IFPRI. Zhang, X., and S. Fan. 2001. Public investment and
Sperling, L., M. Loevinsohn, and B. Ntabomvura. regional inequality in rural China. Environment
1993. Rethinking the farmer's role in plant and Production Technology Division Discussion
breeding: Local bean experts and on-station Paper No. 71. Washington, D.C.: IFPRI.
Peter Hazell is director of the Environment and Production Technology Division of the International Food Policy Research Institute (IFPRI). Lawrence Haddad is director of the Food Consumption and Nutrition Division of IFPRI.

with permission from the International Food Policy
Research Institute