Food, Agriculture, and the Environment Discussion Paper 34

Agricultural Researchand Poverty Reduction

Peter Hazell and Lawrence Haddad

International Food Policy Research Institute2033 K Street, N.W.

Washington, D.C. 20006 U.S.A.August 2001

INTERNATIONAL FOOD POLICY RESEARCH INSTITUTE

ON INTERNATIONAL AGRICULTURAL RESEARCHTECHNICAL ADVISORY GROUP (TAC), CONSULTATIVE GROUP

Copyright © 2001 International Food PolicyResearch Institute

All rights reserved. Sections of this report may bereproduced without the express permission of butwith acknowledgment to the International FoodPolicy 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

Contents

iii

Tables

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

Figures

1. Number of people living on less than $1 a day, 1987 and 1998 4

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

iv

International agricultural research has contributed enormously to increasing world food supplies to theircurrent state of plenty. Yet poverty remains a major problem and the challenge for agricultural researchnow lies in developing strategies that more explicitly address the needs of the poor. This paper, basedon the study commissioned by the Technical Advisory Committee (TAC) of the CGIAR system, addressesthis issue.

Based on an analysis of the links between agricultural research and poverty alleviation in differenttypes of countries and rural regions, Peter Hazell and Lawrence Haddad identify six key priorities for apro-poor agricultural research agenda: (1) increasing production of staple foods in countries where foodprice 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-poten-tial areas; (3) helping smallholder farms across the board diversify into higher value products, includinglivestock products, especially in countries with rapidly growing domestic markets for such productsand/or access to suitable export markets; (4) increasing employment and income-earning opportunitiesfor landless and near-landless workers in labor surplus regions; (5) developing more nutritious and saferfoods to enhance the diets of poor people; and (6) undertaking agricultural research in ways that aremore empowering to the poor.

Hazell and Haddad discuss strategies for achieving each of these goals with the least trade-off innational agricultural growth. In short, Hazell and Haddad suggest strategies to target agriculturalresearch 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 foragricultural research has nevertheless declined, especially in developing countries. Yet more is beingasked of agricultural research and extension systems. If new technologies and research paradigms areto be developed that specifically address the needs of the poor, then funding for international and nation-al agricultural research must be increased.

Per Pinstrup-AndersenDirector GeneralIFPRI

Emil JavierChairTechnical Advisory Committee/CGIAR

Foreword

v

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

Acknowledgments

vi

Until recently, researchers and policymakers sawpoverty alleviation as a subsidiary goal of agri-cultural research. The primary goal was toincrease food supplies through cost-reducing tech-nological changes that would lead to lower foodprices. Research focused on increasing the yieldsof important food staples in irrigated and high-potential rainfed areas, where researchers per-ceived the productivity returns to agriculturalresearch to be highest. This strategy was enor-mously successful and led to real benefits for thepoor as well as for societies in general. But it wasnot sufficient to eliminate rural poverty, whichabounds even in countries that now have nationalfood surpluses.

Plentiful global food supplies and decliningtrade barriers have created the opportunity todevelop explicitly pro-poor research strategies forthe public sector. The opportunity is enlarged to bythe private sector’s increasing role in addressingmainstream productivity challenges. Private-sectorinvolvement seems likely to grow as the evolutionof biotechnology and intellectual property regimesredefine the traditional public-goods nature ofmuch agricultural research. Although food sup-plies must double over the next 20 years, increas-ing productivity alone will not be enough. Thepublic sector will also need to focus sharply on thechanging nature of market failures that it has tra-ditionally sought to correct and on the growingimportance of countries’ other social and environ-mental goals. This shift will require the public sec-tor to invest in ways that offset the private sector’s

chronic underinvestment in research for poor farm-ers and regions and for containment of environ-mental problems.

Against this backdrop, what can public agri-cultural research systems that serve developingcountries do to increase the poverty-reducingimpact of their investments? This paper addressesthat question generally in the Introduction. Chapter2 examines the changing nature and extent ofpoverty, and its implications for agriculturalresearch. It notes the evolution in thinking aboutpoverty beyond measures of physiological depri-vation (an inability to meet basic material needs) toincorporate measures of social deprivation (pooraccess to the components of power such as deci-sionmaking processes, information, and authority).Chapter 2 also shows the shifts in poverty ratesand numbers from a spatial point of view.

Chapter 3 reviews how agricultural researchcan reduce poverty not only through traditionalpathways (such as own-farm productivity increas-es, greater employment, general equilibriumeffects, and the lowering of food prices), but alsothrough newer pathways (such as communityempowerment through collective action for natu-ral-resource management). Chapters 2 and 3 sug-gest the need to tightly target agriculturalresearch.

Chapter 4 outlines one possible typology tohelp do so. The typology was developed forpotential technology-adopting regions withincountries and is based on two country criteria(low- versus middle-income country and liberal-

1. Introduction*

1

*This paper is based on the TAC-commissioned study, which is still under review by TAC. Consequently, the paper rep-resents a work that is in progress.

ized versus unliberalized market and trade poli-cies) and four region-specific characteristics (highversus low agricultural potential; good versus poorinfrastructure, service provision, and marketaccess; low wages and abundant labor supplyversus high wages and scarce labor supply; andwhether the region’s poor are empowered or dis-empowered in terms of access to land, otherresources, and public services).

Chapter 5 describes the six pro-poor researchpriorities that emerge from the typology and con-siders research strategies for achieving them,emphasizing strategies to minimize trade-offsagainst sectorwide agricultural growth, which is

key to long-term poverty reduction. The six priorityareas for pro-poor research are: (1) maintaininggrowth in staple food production, particularly incountries where food-price effects are still strong;(2) intensifying less-favored lands; (3) helpingsmallholders to diversify into higher value prod-ucts; (4) increasing employment and incomeopportunities for landless workers; (5) increasingthe access of the poor (especially poor women) tofoods rich in crucial micronutrients; and (6) under-taking agricultural research in ways that are moreempowering to the poor.

Chapter 6 concludes with recommendationsfor the public sector’s future research agenda.

2

Defining Poverty

A consensus is emerging around the view thatpoverty consists of two interacting deprivations—physiological and social. Physiological depriva-tion describes an inability to meet or achievebasic material and physiological needs and canbe measured either as a lack of income, whichlimits access to food and to education, health,housing, water, and sanitation services, or by thefailure to achieve desired outcomes, such as ahigh-quality diet rich in micronutrients, health sta-tus, educational attainment, and the quality ofhealth, water, and sanitation services received.Diet quality is crucial to individual well-being, par-ticularly for girls, women, and infants.

Although income and achievements correlate,they do not do so perfectly. For example, house-holds with relatively high-income consumption lev-els often contain individuals who do not getenough of the right type of food to stave offhunger, anemia, goiter, or death. This can occurbecause of lack of information (for example,household leaders do not realize the value of girls’education or do not know that their daughters’diets are low in iron), preferences (there may be adeliberate bias toward education for boys at theexpense of girls), or an inability to use income topurchase inputs required for well-being (for exam-ple, sanitation services that cannot be purchasedin the market).

Measures of physiological deprivation are rel-atively easy to quantify at the household level inways that are consistent over time and acrossspace. For this reason, usable data series havebeen constructed for most countries to form the

basis for international comparisons and trendsanalysis. Income and consumption measures areusually preferred, but nutrition indicators based onthe height and weight of young children are oftenused as simpler measures of physiological depri-vation.

Social deprivation typically is assessed at theindividual or community level; it refers to anabsence of elements that are empowering—auton-omy, time, information, dignity, and self-esteem.Lack of empowerment is reflected in exclusionfrom important decision-making processes, evenwhen the outcomes are of considerable impor-tance to the poor—for example, decisions aboutpublic investments in the local community, man-agement of common properties, and priorities foragricultural research and extension. Physiologicaland social deprivations interact in a vicious/virtu-ous cycle, with increased empowerment leadingto greater income-earning ability, which leads togreater power, and so on. This more comprehen-sive social definition of poverty provides richerinsights into the process of becoming poor, stay-ing poor, or becoming less poor. It also providesan opportunity for input from the poor themselves,in terms of the dimensions of deprivation identifiedand the way in which severity is assessed. Thisinput can be gained through participatoryresearch methods.

Because measures of social deprivation arein their infancy and are based largely on quali-tative analysis, quantitative indicators that areconsistent over time and comparable acrossregions and countries are not available todescribe the extent and location of povertyaround the world, which this report attempts to

2. The Nature and Extent of Poverty

3

do. But where qualitative assessments have beenundertaken, they provide a richer understandingof the nature of poverty and of poor peoples’livelihood strategies. Such information can becrucial to guide decisions about the design andtargeting of agricultural research that mesheswith the livelihood strategies of the poor. Publicagricultural research systems may need to investin collecting this kind of information in importanttarget regions.

Poverty Trends

Poverty continues to be a major problem in manyparts of the developing world. Approximately 1.2billion rural people live in poverty (defined as liv-ing on less than $1 per day), and about 160 mil-lion preschool children are malnourished (WorldBank 2000; Pinstrup-Andersen, Pandya-Lorch,and Rosegrant 1997). Infant malnutrition rates reli-ably measure deprivation, given their comparabil-

ity across space and time. These rates also arereliable lead indicators of poverty because malnu-trition at a young age leads to the early onset ofpoor physical and cognitive productivity and tohigher rates of noncommunicable diseases suchas diabetes and heart disease later in life(ACC/SCN 2000).

Globally, about 90 percent of the developingworld’s poor live in either Asia or Sub-SaharanAfrica (Figure 1). Asia dominates, with two-thirdsof the total poor, who are concentrated in SouthAsia (43 percent). Less than 1 percent of the poorlive in the Middle East and North Africa (MENA),and about 7 percent live in Latin America and theCaribbean (LAC).

The total number of poor has changed littlesince 1987, but the regional distribution haschanged, with poverty declining in China, EastAsia, and MENA but increasing in South Asia andSub-Saharan Africa. Poverty is growing the fastestin Sub-Saharan Africa, where it is aggravated by ahigher incidence of HIV/AIDS infection (Figure 2).

4

Millions of people

550

450

350

250

150

50

East Asia(excluding

China)

China Europe andCentral Asia

LAC MENA South Asia Sub-SaharanAfrica

114

65

303

213

124

64 78

9 6

474

522

217

291

� 1987� 1998

Figure 1—Number of people living on less than $1 a day, 1987 and 1998

Source: World Bank 2000.Note: LAC = Latin America and the Caribbean; MENA = Middle East and North Africa.

The incidence and distribution of malnour-ished (underweight) children of preschool ageshows a similar pattern (Table 1). Most of thedeveloping world’s malnourished children live inSouth Asia or Sub-Saharan Africa (togetheraccounting for 70 percent), again with themajority concentrated in South Asia (51 per-cent). About 49 percent of all preschool childrenin South Asia are malnourished, compared with31 percent in Sub-Saharan Africa and 23 per-cent in East Asia (including China). Trends in thenumber of malnourished preschool children alsoshow similar patterns to monetary measures ofpoverty. While the number of underweight chil-dren in South Asia has slowly declined, the num-ber in Sub-Saharan Africa increased from 25.7million in 1990 to 31.4 million in 1995. In thepast five years, the situation in Eastern andSouthern Africa has continued to deteriorate,spurred in part by the HIV/AIDS epidemic(ACC/SCN 2000).

Of the total number of poor people livingbelow $1 per day, 75 percent, or 0.9 billion, liveand work in rural areas (IFAD 2001). The ruralpoor are distributed across regions in much the

same pattern as the total poor. A significant per-centage of the rural poor live in less-favored areasthat are challenged by difficult agroclimatic con-ditions, such as poor soil, low and unstable rain-fall, steep slopes, and short growing seasonsand/or inadequate infrastructure and supportservices (roads, irrigation, markets, research andextension, credit, schools, and health centers).According to a report by the Technical AdvisoryCommittee (TAC) of the Consultative Group onInternational Agricultural Research (CGIAR/TAC2000), “favored” agricultural lands account foronly 10.7 percent of the agricultural area in thedeveloping world and for 8.5 and 16.6 percent,respectively, in Sub-Saharan Africa and Asia(Table 2). Moreover, only about one third of therural population live in favored lands (Table 2).The vast majority of rural people live outside thesefavored lands, which in the TAC study comprise“marginal” lands (24 percent of the total agricul-tural area), sparsely populated arid lands (26 per-cent), and forest and woodlands (40 percent).Some of these less-favored lands have good agri-cultural potential, particularly where roads andirrigation are available, so it cannot be concluded

5

Millions of people

25

20

15

10

5

0

23.3

3.86

1.3 0.53 0.12

East Asiaand Pacific

South andSoutheast Asia

Sub-Saharan Africa

� Adults and children living with HIV/AIDS� Adults and children newly infected with HIV/AIDS

Figure 2—Those living with HIV/AIDS and those newly infected:

Asia and Sub-Saharan Africa, 1999

Source: www.unaids.org/hivaidsinfo/index.html

that the remaining two-thirds of the rural popula-tion all live in low-potential areas.

Although reliable poverty data by land typedo not exist for most countries, more precise datado exist for India and China. A recent IFPRI studyreports that for India in 1993, 42 percent of therural poor lived in low-potential rainfed areas,while 16 percent lived in irrigated areas and 42percent lived in high-potential rainfed areas (Fanand Hazell 2000). A similar share of China’s ruralpoor lives in low-potential areas (Fan et al. 2000).

Additional insights into the distribution of ruralpoverty by land type are available from data on

child malnutrition. Sharma et al. (1996) havemapped the incidence of malnutrition amongpreschoolers by the agroecological zones definedby CGIAR (Table 3). Their analysis shows that theincidence of child malnutrition is highest in warm,semi-arid tropical and subtropical areas (zones 1and 5) and that 43 percent of all malnourishedpreschool children live in these areas. Food pro-duction per hectare is also relatively low in thesezones despite average or above-average levels ofirrigation.

The rural poor are predominantly smallholderfarmers and landless agricultural workers. For the

6

Table 1—Trends in child malnutrition in developing countries, by region, 1970–95

Change, Region 1970 1975 1980 1985 1990 1995 1970 to 1995Percent of children malnourished Percent Percentage points

South Asia 72.3 67.7 63.7 61.1 53.4 49.3 -23.0Sub-Saharan Africa 35.0 31.4 28.9 29.9 28.8 31.1 -3.9East Asia 39.5 33.3 30.0 26.5 23.5 22.9 -16.6Near East and North Africa 20.7 19.8 17.2 15.1 n.a. 14.6 -6.1Latin America and the Caribbean 21.0 17.0 12.2 10.6 11.4 9.5 -11.5All regions 46.5 41.6 37.8 36.1 32.3 31.0 -15.5

Number of children malnourished MillionsSouth Asia 92.2 90.6 89.9 100.1 95.4 86.0 -6.2Sub-Saharan Africa 18.5 18.5 19.9 24.1 25.7 31.4 +12.9East Asia 77.6 45.1 43.3 42.8 42.5 38.2 -39.4Near East and North Africa 5.9 5.2 5.0 5.0 n.a. 6.3 +0.4Latin America and the Caribbean 9.5 8.2 6.2 5.7 6.2 5.2 -4.3All 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 stan-dard deviations below the median weight-for-age Z-score of the National Center for Health Statistics/World Health Organizationinternational reference. n.a. is not available.

Table 2—Distribution of land types by region

Land type (% of total land) Rural populationSparsely populated Forest and living in

Region Favored Marginal arid lands woodland favored lands (%)Sub-Saharan Africa 8.5 23.1 24.6 43.7 27.0Asia 16.6 30.0 18.5 34.6 37.0Latin America and Caribbean 9.6 20.3 8.1 61.9 34.0Near East and North Africa 7.8 22.6 65.8 3.9 24.0Total (105 countries) 10.7 24.0 25.9 39.4 35.0Source: CGIAR/TAC 2000.

developing countries as a whole, about half therural population lived in smallholder farm house-holds in 1988, and one-quarter lived in landlesslabor households (Jazairy et al. 1992). The ratioof landless to smallholder farmers was much lowerin Sub-Saharan Africa (0.15) than in LatinAmerica (0.82) or Asia (0.53).

Smallholder farms are getting smaller andmore numerous in most parts of the developingworld (Hazell, Jagger, and Knox 2000). This con-trasts sharply with changes in the industrializedcountries where farms are getting larger and therehas been an exodus of small-scale farmers fromagriculture.

Smallholder households are also diversifyingtheir livelihood strategies and increasing theirshare of nonfarm income (Reardon et al. 1998;Carney 1998). Smallholders and landless workerstypically earn more than half their total householdincome from nonagricultural sources. Such diversi-fication could reflect worsening impoverishmentand desperation as land becomes increasinglyscarce, or it could reflect increasing prosperity, asrural workers are attracted to higher-paying non-farm jobs. The Southeast Asian experience hasbeen largely of the latter variety, with rapidgrowth in rural nonfarm employment and incomeas a result of dynamic national economies(Rosegrant and Hazell 2000). Many Southeast

Asian countries seem to be following the Japaneseexperience, retaining large numbers of smallhold-er farms that are becoming part-time enterprises.Diversification is more likely to be associated withgreater impoverishment when increasing landscarcity occurs in conjunction with slow agricultur-al growth, stagnant national and regionaleconomies, and falling wages (Hazell andReardon 1998). Such situations are not uncom-mon in many of the poorer countries in South Asiaand Sub-Saharan Africa.

Although there are many explanations for cur-rent levels and trends in poverty around theworld, agricultural growth is an important con-tributing factor. Its influence has been most stud-ied for India. Prior to the Green Revolution in thelate 1960s, the incidence of rural poverty in Indiafluctuated widely. Researchers obtained differentresults on the relationship between poverty andagricultural growth, depending on the periodthey chose for their analysis (Bardhan 1973;Ahluwalia 1978; Gaiha 1989; Ghose 1989;Griffin and Ghose 1979; Saith 1981). But afterthe Green Revolution began in the mid-1960s,the incidence of rural poverty began a definitedownward trend (from about two-thirds to one-third of the rural population by the early1990s).A greater consensus began to emerge in the liter-ature on the poverty-reducing impact of agricul-

7

Table 3—Distribution of malnourished children by agroecological zone, 1990

Malnourished childrenFood production/

% of hectare of % of arableAgroecological zone Millions total children arable land (TGE) land irrigated1. Warm, semi-arid tropics 47.9 49.0 0.98 17.22. Warm, sub-humid tropics 20.6 36.4 1.03 9.33. Warm, humid tropics 38.0 37.0 1.92 18.24. Cool tropics 8.1 26.0 1.50 8.35. Warm, semi-arid subtropics (summer rainfall) 31.7 44.0 1.07 40.06. Warm, sub-humid subtropics (summer rainfall) 7.4 38.0 1.44 26.87. Warm/cool humid subtropics (summer rainfall) 10.4 19.0 1.41 21.38. Cool subtropics (summer rainfall) 10.6 23.0 1.28 14.69. Cool subtropics (winter rainfall) 8.2 17.4 0.78 23.2

Total 183.4 33.7 1.20 20.0Source: Sharma et al. 1996 (Table 2).Note: TGE = total grain equivalent.

tural growth (Ghose 1989; Fan, Hazell, andThorat 1999; Datt and Ravallion 1997). Ruralpoverty also declined dramatically in China afterpolicy reforms launched rapid increases in agri-cultural growth and in the purchasing power ofrural households (Fan, Zhang, and Zhang 2000).Rosegrant and Hazell (2000) report similarbroad relationships across much of Asia duringthe Green Revolution era. Whereas 60 percent ofAsians lived in poverty in 1975, this ratio hadfallen to less than one in three by 1995, and thetotal number of poor declined from about 1.2 bil-lion to about 0.8 billion despite a 1 billionincrease in the total population.

Agricultural growth has not always improvedincome distribution, but by raising per capita

incomes across the board, it has significantly con-tributed to reducing the number of people livingbelow the poverty line. In contrast, in Sub-SaharanAfrica, where poverty is increasing and food inse-curity is deteriorating, agricultural growth hasbeen very disappointing, often struggling just tokeep pace with population growth.

It would be dangerous to conclude too muchfrom these patterns of association without moremicro-based evidence on cause-and-effect rela-tionships. Nor should it be concluded that agri-cultural growth necessarily reduces poverty. Butthe data do provide an optimistic backdrop thatjustifies more careful analysis of how technologi-cally driven agricultural growth can benefit poorpeople.

8

Agricultural research that leads to improved tech-nologies can benefit the poor in a number ofways:

1. Research can help poor farmers directlythrough increased own-farm production, pro-viding more food and nutrients for their ownconsumption and increasing the output of mar-keted products for greater farm income.

2. Small farmers and landless laborers can gaingreater agricultural employment opportunitiesand higher wages within the adoptingregions.

3. The poor can have opportunities to migrate toother agricultural regions.

4. Growth in the rural and urban nonfarm econ-omy induced by more rapid agriculturalgrowth can benefit a wide range of rural andurban poor people.

5. Research can lead to lower food prices for allconsumers, whether from rural or urban areas.

6. Research can lead to greater physical andeconomic access to crops that are high innutrients and crucial to the well-being of thepoor—particularly poor women.

7. Research can empower the poor by increasingtheir access to decisionmaking processes,enhancing their capacity for collective actionand reducing their vulnerability to economicshocks via asset accumulation.

These benefits do not necessarily materializefor the poor. Many conditioning factors determinewho benefits from technological change. Nor dothe benefits of research all necessarily work in thesame direction. For example, while many of thepoor may benefit from less costly food and greateropportunities for nonfarm income, production andemployment benefits in the adopting regions maybe disappointing or even perverse. Net outcomes,both for individual poor people and for entirepoor populations, can be difficult to determinebeforehand.

On-Farm Productivity Impacts

Poor farmers will obtain own-farm benefits fromnew technologies only if they adopt them. Thismeans that the new technologies must be appro-priate and profitable for farming conditions, andthat poor farmers must have access to the knowl-edge and inputs necessary to adopt the technolo-gy. In principle, improved crop varieties arescale-neutral and can be adopted by farms of allsizes, but the same is not always true of othertechnologies or of complementary inputs like irri-gation and machines, and access to fertilizers andcredit. If institutions that provide these servicesand inputs are biased in favor of large farms, thepoor may not be able to adopt new technologies.

To invest in new technologies whose returnsoccur over a number of years (for example,improved tree crops or better soil-managementtechniques) and to obtain credit to finance suchlong-term investments, poor farmers need secureownership or tenancy rights. Insecure rights to

3. How Agricultural Research Can Help the Poor

9

land may increase poor farmers’ vulnerability toeviction should larger farmers and landlords wantto expand their own cropped area as the result ofmore profitable technologies.

Under risky agroclimatic conditions, poorfarmers may be reluctant to adopt profitablenew technologies because they require inputinvestments that could be lost in an unfavorableyear. On the other hand, larger farmers aremore likely to assume such risks because theyhave larger reserves and better access to creditand insurance.

Farmers who adopt new technologies oftensucceed in lowering their production costs perunit of output (though not usually per hectare),and therefore can better compete in the market.Moreover, if the technology is widely adoptedand market prices fall as a result, the decline inunit cost may be essential for maintaining farmincome. In this case, farmers who do not adoptthe technology will be disadvantaged not onlyby stagnant production but also by decliningprices and tighter profit margins. This profitsqueeze can be detrimental to nonadopters with-in technology-adopting regions and to farmerswho live in regions that are inappropriate for thenew technology.

Even when poor farmers do benefit from sig-nificant productivity gains, these benefits are notalways shared equitably among household mem-bers. In many societies, men and women haveresponsibility for growing different crops.Therefore, which crops benefit from technologicalchange determines who controls the increasedproduction within the household. Technologicalchange for women’s food crops may translate intobetter nutrition and well-being for women and chil-dren than technological change for men’s cashcrops (Haddad, Hoddinott, and Alderman 1997).

The initial experience with the GreenRevolution in Asia stimulated a large body ofempirical literature on how agricultural techno-logical change affects poor farmers (see Kerrand Kolavalli 1999 for a recent review). Criticsof the Green Revolution argued that because of

their better access to irrigation water, fertilizers,seeds, and credit, owners of large farms werethe main adopters of the new technologies, andsmallholders were either unaffected or harmedbecause the Green Revolution resulted in lowerproduct prices, higher input prices, and owners’efforts to increase rents or force tenants off theland. Critics also argued that the GreenRevolution encouraged unnecessary mechaniza-tion, with a resulting reduction in rural wagesand employment. The net result, critics said, wasan increase in inequality of income and asset dis-tribution and a worsening of absolute poverty(see Griffin 1974; Frankel 1976; Farmer 1977;ILO 1977; Pearse 1980).

Although a number of village and householdstudies conducted soon after the release ofGreen Revolution technologies lent some supportto early critics, more recent evidence showsmixed outcomes (Blyn 1983; Pinstrup-Andersenand Hazell 1985; Lipton and Longhurst 1989;Hazell and Ramasamy 1991; David and Otsuka1994). Although small farmers did lag behindlarge farmers in adopting Green Revolution tech-nologies, many of them eventually did so. Manyof these small-farm adopters benefited fromincreased production, greater employmentopportunities, and higher wages in the agricul-tural and nonfarm sectors. Moreover, most smallfarmers were able to hold onto their land andcaptured significant total production increasesfrom their holdings (Westley 1986; Hazell andRamasamy 1991; Rosegrant and Hazell 2000).In some cases, small farmers and landless labor-ers actually ended up gaining proportionallymore income than larger farmers, with a netimprovement in the distribution of village income(Hazell and Ramasamy, 1991).

This is not to say that the Green Revolutionwas equitable everywhere (Freebairn 1995). Butthe conditions under which it and similar yield-enhancing technologies are likely to have equi-table on-farm benefits are now reasonably wellunderstood. These include: (1) a scale-neutral tech-nology package that can be profitably adopted

10

on farms of all sizes; (2) an equitable distributionof land with secure ownership or tenancy rights;(3) efficient input, credit, and product markets sothat farms of all sizes have access to needed mod-ern farm inputs and information and are able toreceive similar prices for their products; and (4)policies that do not discriminate against smallfarms and landless laborers (for instance, no sub-sidies on mechanization and no scale-biases inagricultural research and extension). These condi-tions are not easy to meet. Typically, governmentmust make a concerted effort to ensure that smallfarmers have fair access to land, knowledge, andmodern inputs.

Agricultural Employment and Wages

Many yield-enhancing technologies increase totalon-farm employment, particularly if they expandthe gross cropped area, for example, by growingmore crops per year with irrigation and short-sea-son crop varieties. This can lead to less seasonalfluctuation in employment earnings. But whetherall this translates into higher wage earnings for thepoor depends in large part on the elasticity of thelabor supply. If labor is abundant in the adoptingregion, then additional employment will have littleeffect on wages, workers will gain, and farmerswill have limited incentive to invest in labor-dis-placing machines. But if the labor supply is inelas-tic, wages will rise sharply and labor-displacingmachines may become attractive. Initial mecha-nization may be targeted on labor-intensive taskslike plowing and threshing, but once farmersinvest in tractors the incremental costs of mecha-nizing other tasks may become quite low, causingmore widespread displacement of labor.Mechanization may also occur prematurely if gov-ernment policies such as cheap credit for largefarms make mechanization less costly.

Population growth increases the supply elas-ticity of labor and therefore acts to dampen wage

increases. This may lead to greater agriculturalemployment, but lower living standards for work-ers. The decline in living standards can be partic-ularly sharp if rapid population growth coincideswith adoption of agricultural technologies withlow employment elasticities.

Impact on Inter-regionalMigration

Technological change in agriculture is usually site-specific and does not benefit all regions equally.The Green Revolution was initially concentrated inirrigated regions and only later spread to morefavorable rainfed areas. Technological change,therefore, can contribute to widening disparitiesbetween regions. Worse, if the technology leadsto lower production costs per unit of output in theadopting regions, producer prices may fall, leav-ing non-adopting regions with lower prices as wellas stagnant yields, so that their incomes couldactually decline. But inter-regional migration actsto buffer these gaps and provides an efficient wayof spreading the benefits to poorer regions withmore limited agricultural growth potential.

In India, the Green Revolution led to the sea-sonal migration of more than a million agricultur-al workers each year from the eastern states toPunjab and Haryana (Westley 1986). Moreover,in a study of the impact of the Green Revolution ina sample of Asian villages, David and Otsuka(1994) found that seasonal migration played animportant role in spreading the benefits betweentechnology-adopting and -nonadopting regions.But even though migration can buffer wideningincome differentials between regions, it is rarelysufficient to avoid it. In India, poverty in many low-potential rainfall areas has changed little evenwhile irrigated and high-potential rainfall areashave progressed (Fan and Hazell 2000).Regional inequalities have also worsened inChina in recent years (Knight and Song 1992;Zhang and Fan 2001).

11

Impact on the NonfarmEconomy

Agricultural growth generates important rounds ofincome and employment growth within the non-farm economy. These are driven by (1) increaseddemands for additional farm inputs, investmentgoods, and marketing services (demands thatoften increase per hectare with technologicalchange); (2) increased rural household demandsfor consumer goods and services as farm andwage incomes rise; and (3) generation of farmsavings, foreign exchange earnings, and lowerprices for foods and other primary commodities.

This paper distinguishes between the ruralnonfarm economy, including its market towns, andthe broader urban-based national economy. Therural nonfarm economy has more transparent linksto agriculture and is especially important to therural poor. However, there are also powerful linksbetween agricultural and national economicgrowth that benefit society at large, including therural and urban poor.

The Rural Nonfarm EconomyThe rural nonfarm economy is of special impor-tance to the rural poor and to the urban poor liv-ing in rural towns. Landless and near-landlesshouseholds everywhere depend on nonfarm earn-ings. Those with less than half a hectare earnbetween 30 and 90 percent of their income fromnonfarm sources (Haggblade, Hazell, and Brown1989; Rosegrant and Hazell 2000). Nonfarmshares are strongly and negatively related to farmsize. Low-investment manufacturing and servic-es—including weaving, pottery, gathering, foodpreparation and processing, domestic and per-sonal services, and unskilled nonfarm wagelabor—account for a greater share of income forthe rural poor than for the wealthy (Hazell andHaggblade 1993). The reverse is true of trans-port, commerce, and such manufacturing activitiesas milling and metal fabrication, which requiresizable investments.

Nonfarm income is also important to the pooras a means to help stabilize household income indrought years (Reardon et al. 1998). In a study ofseveral villages in the semi-arid tropics of India, forexample, Walker and Ryan (1990) found thatnonagricultural self-employment and labor marketearnings became increasingly important sources ofincome during the 1980s, increasing mean incomeand dampening household income variability.

Numerous studies have shown that agriculturalgrowth generates important income and employ-ment multipliers within the surrounding nonfarmeconomy. The multipliers are particularly large inAsia, with $0.5 to $1.0 of additional income cre-ated in the local nonfarm economy for each dollarof additional income created in agriculture (Bell,Hazell, and Slade 1982; Hazell and Ramasamy1991; Haggblade and Hazell 1989). The multipli-ers are about half as large in Africa and LatinAmerica (Haggblade and Hazell 1989).

The multipliers are predominantly driven byincreased rural household demands for consumergoods and services as farm incomes rise. Small,informal, and labor-intensive rural nonfarm firmssupply many of these goods and services. Thisleads to high nonfarm employment elasticitieswithin rural regions. Often, each 1 percentincrease in agricultural output is associated with a1 percent increase in rural nonfarm employment(Gibb 1974; Hazell and Haggblade 1991).

The strength of the regional growth linkages ishigher in labor-abundant regions and increaseswith regional development and per capitaincomes. Irrigated regions dominated by medium-sized farms and modern input-intensive farmingsystems generate the largest multipliers. The multi-pliers are smaller in rainfed farming systems andin regions dominated by very small farms or largeestates (Haggblade and Hazell 1989). In India,for example, the multipliers are largest in GreenRevolution states like Punjab and Haryana andsmallest in less developed states like Bihar andMadhya Pradesh (Hazell and Haggblade 1991).As regions develop and labor markets tighten,

12

employment impacts become less important. Thelower multipliers in Africa are attributable to lowper capita incomes (which are mostly spent onfood), poor infrastructure, and farming technolo-gies that require few purchased inputs(Haggblade and Hazell 1989). In Latin America,higher per capita rural incomes ought to lead tolarger multipliers, but often fail to do so becausethe distribution of income is highly inequitable andthe richer households have much stronger con-sumption linkages to cities than the rural nonfarmeconomy.

Rural income multipliers and employment elas-ticities of the sizes observed in Asia mean thattechnological change in agriculture has the poten-tial to generate significant new nonfarm income-earning opportunities for the poor. These mayarise in the form of greater nonfarm employment,higher wages, and opportunities for the poor tostart or expand nonfarm businesses of their own.Increasing competition for labor between agricul-ture and the local nonfarm economy can also con-tribute to higher agricultural wages for the poor.

The benefits of growth in the rural nonfarmeconomy are concentrated in towns more than vil-lages, affecting an important segment of the non-farm poor residing in towns. Distribution ofbenefits between rural farm areas and townsdepends on the state of infrastructure connectingthe two, on population density, on governmentpolicies, and on average per capita income levels(Haggblade, Hazell, and Brown 1989).

The National EconomyAgricultural growth has broad general equilibrium(GE) impacts on the national economy that varyaccording to stage of economic development. Inpoorer countries, agriculture accounts for the lion’sshare of national income, employment, andexport earnings. Under these conditions, even amodest growth rate for agriculture can have sig-nificant leverage on the national economy. Rapidagricultural growth contributes to the economictransformation of a country in a number of impor-tant ways. It:

• supplies basic foods, raw materials for agroin-dustry, and exports, and frees up foreignexchange for the importation of strategicindustrial and capital goods.

• releases labor and capital (in the form of ruralsavings and taxes) to the nonfarm sector.Generates purchasing power among the ruralpopulation for nonfood consumer goods andservices, supporting growth in services andtrade.

• provides a nascent market for an emergingmanufacturing sector.

• reduces poverty by increasing labor productiv-ity and employment and by lowering foodprices for all.As the transformation of an economy

advances, agriculture’s share in national incomefalls and its importance for national economicgrowth diminishes. The nonagricultural sectorbecomes the primary engine of growth and is nolonger as dependent on resource flows from agri-culture or on agriculture’s demand linkages.However, agriculture’s share of total employmentfalls more slowly than its share of national income,with the inevitable result that agricultural laborproductivity, and hence per capita farm incomes,lag behind the nonagricultural sector. The problemis then to absorb workers out of agriculture at asufficiently rapid rate to stop their average pro-ductivity (and hence their incomes) from laggingtoo far behind the levels achieved in the nonagri-cultural sector. Few countries have been able tomanage this transition successfully. Either ruralpoverty has persisted until late in the developmentprocess or governments have engaged in expen-sive farm-income support policies.

Empirical studies confirm the importance ofthese GE effects in developing countries. In India,the fact that the nonfarm share of total nationalemployment did not change for over a centuryuntil the full force of the Green Revolution wasunderway in the 1970s provides strong circum-

13

stantial evidence of the importance of agriculturalgrowth as a motor for India’s nonfarm economy.Rangarajan (1982) confirmed this, estimating thata 1 percent increase in the agricultural growthrate stimulated a 0.5 percent rise in the growthrate of industrial output and a 0.7 percent rise inthe growth rate of national income.

Computable general equilibrium (CGE) mod-eling studies show that these GE effects arestronger in economies that have more liberalizedtrade and that invest in adequate levels of ruralinfrastructure and service provision (Robinson,Roe, and Yelden 1998). Modeling results alsoconfirm that agriculture-led growth strategies aremore beneficial for overall economic growth thanindustry-led strategies, particularly in agrarianeconomies (Adelman 1984; De Franco andGodoy 1993).

Impact on Food Prices and Diet Quality

Technological change contributes to increases inthe aggregate output of affected commodities andoften lower unit costs. This has proved to be oneof the most important ways through which poorpeople have benefited from technological changein agriculture (Scobie and Posado 1978;Rosegrant and Hazell 2000; Fan, Hazell, andThorat 1999).

If the demand for these products is downwardsloping (that is, export opportunities are con-strained by trade policy or by high transportcosts), the output price will fall. The more elasticthe supply relative to demand, the greater theprice decline will be (Alston, Norton, and Pardey1995). Lower food prices benefit rural and urbanpoor alike. Because food accounts for a largeshare of their total expenditures, the poor gainproportionally more than the nonpoor from adecline in food prices (Pinstrup-Andersen andHazell 1985). These price effects may be muted inopen economies with low transport costs. Morecountries now fall into this category than before

because of recent rounds of market liberalizationpolicies. But many poor countries still face hightransport costs because of poor infrastructure,remoteness from world markets, or inefficient mar-keting institutions. Hence, domestic prices are stillresponsive to local supply even after market liber-alization. In many landlocked African countries,for example, domestic prices still fall sharply whendomestic food production increases suddenly.Some traditional food crops are not traded inworld markets, and therefore yam, millet, taro,and teff prices continue to be endogenously deter-mined within the countries that grow them.

Food-price benefits may be enhanced if tech-nological change leads to lower production costsper unit of output. Farmers can then maintain orincrease profits even selling at lower prices. Butwhether consumers benefit from these lower costsdepends on the food marketing and distributionsystem being sufficiently competitive so that costsavings at the farm gate are passed up throughthe marketing chain. In some cases, the cost sav-ings are simply captured as additional profits inthe marketing chain.

Technological changes that smooth seasonalfood supplies, such as irrigation and short-seasonrice varieties, can help smooth seasonal price vari-ation. This can be of considerable benefit to thepoor. The rural poor may become more food-secure from increased local production by reduc-ing the need to purchase food from outside theregion. Locally grown produce is cheaperbecause there is no need to cover high trans-portation costs.

Food-price declines amount to an increase inreal income for net food-purchasing households.Real-income increases can be used to increaseconsumption of important staples and to purchasemore diverse, nutritionally rich diets. What is thebest way to improve the nutrient content of thepoor people’s diets through agricultural research?By concentrating on increasing incomes throughproductivity-enhancing investments in staple foodsor by concentrating on decreasing the relativeprices of micronutrient-rich foods?

14

Figure 3 shows a downward trend in the priceof rice in Bangladesh. However, it also showsupward trends in the real prices of other foods thatare richer in micronutrients. This may reflect under-investment in technologies for the production ofnutrient-rich foods such as fruits, vegetables, andnonruminant livestock. Areas that are more remotetend to have less access to perishable foods viathe marketplace, so investment in these crops maywell deliver the highest return in terms of micronu-trients delivered per dollar of research resourcesspent. In areas with good market access, the ques-tion is more complicated, since access to micronu-trient-rich foods is dependent on food-priceelasticities, the extent to which women control ofhousehold income, and the quality of informationand education programs related to diet.

Enhancing the Nutrients inStaple Food Crops

Agricultural research that enhances the nutrientquality of foods poor people eat can directlyimprove their diets. Breeding maize for higherquality proteins is an early example. Unfortunately,

the rapidly changing consensus in the nutritioncommunity as to the limiting factors in the diet (fromprotein to calories and micronutrients) made thequality-protein maize (QPM) experience somewhatdemoralizing for the plant-breeding community(Tripp 1990).

Despite this recent history, a new generationof plant-breeding efforts is underway (see Grahamand Welch 1996 for a good summary). The focusthis time is not on protein, but on micronutrients.There are three broad goals: (1) increase themicronutrient concentration in the crop, (2)decrease the concentration of absorptioninhibitors such as phytic acid, and (3) increase theconcentration of promoter compounds (for ironand zinc in particular) such as sulphur-containingamino acids (Ruel and Bouis 1998). The twobroad technologies are traditional breeding (look-ing for naturally occurring genetic variation inmicronutrient content) and biotechnology (geneticmodification of foods and the creation of newfoods).

The breeding approaches face many chal-lenges. Can high nutrient-density cultivars befound with (1) little or no yield trade-off so thatfarmers will be interested in adopting them, (2) lit-

15

1973–75 1976–78 1979–81 1982–84 1985–87 1988–90 1991–93 1994–960.5

1.0

1.5

2.0

2.5

3.0

Rice paddy

Spinach

Tomatoes

Pumpkin

ChickenLentils

Index (1973–75 = 1.0)

Figure 3—Food price trends in Bangladesh, 1973–96

Source: Bouis 2000.

tle impact on consumer acceptance (storage,cooking, appearance, and taste), and (3) no neg-ative impact on bioavailability (for the strategiesthat increase micronutrient density)? These chal-lenges are similar to those faced by other food-based interventions, but with the added baggageof the QPM experience.

Compared to the traditional breedingapproach, the biotechnology work is at a muchearlier stage. But it is yielding promising results.The Swiss Federal Institute of Technology’s Institutefor Plant Sciences has demonstrated some successin introducing genes that increase iron and vita-min A concentrations in rice (the so-called “gold-en” rice). The Swiss team plans to collaborate withthe International Rice Research Institute (IRRI) totest the health and environmental consequences ofthe technology and to evaluate the acceptability ofthe rice to farmers in terms of yield impacts.

Impact on the Vulnerability, Assets,and Empowerment of the PoorPoverty is more than a lack of sufficient income orfood to meet basic material needs. It is also a stateof social deprivation, involving vulnerability andlack of participation in decisionmaking and incivil, social, and cultural life. It places a widearray of limitations on the capacity of the poor tosubstantially improve their lives. The lack of capac-ity or power is itself a fundamental characteristicof being poor (Carney 1998).

The assets that individuals, households, andcommunities control are critical for their capacityto cope with vulnerability and to establish securelivelihoods. In many developing countries, thepoor are highly dependent on natural resourceswithin their local environments. Poverty thereforecan be exacerbated by not having access to thoseresources. The cycle is self-perpetuating whenpoor people have no access to technologies andinputs that enable better use of resources, or whenthe poor do not participate in the design and eval-uation of those technologies. In addition, lack ofaccess to financial and human capital, social net-works, and political power compound conditions

of poverty and vulnerability. Access to assets isimportant, but so is the ability to use them in com-bination to create secure livelihoods. Assets canbe seen as a base of power, enabling people “toact and to reproduce, challenge, or change therules that govern the control, use, and transforma-tion of resources” (Bebbington 1999).

Recognizing that the lack of capacity is a com-ponent of poverty, development practitioners arefocusing on empowerment. Kerr and Kolavalli(1999) define empowerment as “a developmentstrategy that seeks to bring about change throughmodifications in the power structure, changing thesocial order in which the poor live.” They distin-guish between two types of empowerment: moti-vational, which involves enhancing peoples’abilities thereby contributing to greater confidenceand self-reliance, and relational, which implieschanging power structures and gaining access topolitical decisionmaking. The objectives ofempowerment are increased self-reliance and self-determination—essential tools for breaking out ofthe poverty trap as well as pursuing a prosperous,fulfilling life.

Empowerment tools can offer users of localresources better access to and control over thelocal environment as well as associated produc-tion and conservation technologies. Some of themost fundamental empowerment tools includeproperty rights (which strengthen the asset base),local collective action and organizations (such ascooperatives and microcredit groups), and politi-cal organization and advocacy. But agriculturaltechnology can also contribute to empowerment,particularly if the poor participate in technologydevelopment. Who decides research prioritiesand technology development approaches are twoimportant aspects of how research systems oper-ate. In most countries, resource-poor farmers inunfavorable regions are passive recipients of tech-nologies; they have no control over the prioritiesof the research systems that serve them. Theresearch process ignores farmers’ knowledge andexperience even though they may offer insightsthat could help develop effective technologies for

16

unfavorable areas. Such systems may perpetuatea sense of helplessness among resource-poorfarmers who wait in vain for effective technologi-cal solutions to come from outside.

Participatory research and disseminationstrategies offer an alternate approach based onempowerment, building farmers’ own capabilityto innovate and giving them greater influence overdecisions in agricultural research. Participatoryresearch involving women farmers can also con-tribute to their empowerment. Participatoryresearch and dissemination approaches wherefarmers communicate results within and betweencommunities can enhance the asset base of peo-ple and communities by building social capitaland organizational capacity.

Net Impacts on the Poor

Poor people have complex livelihood strategies.The rural poor, for example, are often part farm-ers, part laborers, and part nonfarmers—andalways consumers. As such, they may gain or losein different dimensions at the same time, so thatthe net impact of technological change on poorhouseholds can remain ambiguous. A poor farmermight be able to gain from increased on-farm pro-duction as a technology adopter, but may lose orgain from increased agricultural wages orreduced food prices depending on whether he orshe is a net buyer or seller of labor or food. Asmall, nonfarm, business entrepreneur might gainfrom cheaper food, but business profits might fallor rise depending on whether or not hired laborcosts rise faster than sales. There is also a tempo-ral dimension to these outcomes, with some costsand benefits being realized before others havehad time to work through (for example, GEimpacts take longer to materialize than on-farmbenefits). Understanding household livelihood

strategies and the dynamics of change is thereforefundamental for assessing the impact of techno-logical change at the household level.

Assessing how technological change affectsthe poor in aggregate is even more complex anduncertain. Not only is there complexity in assess-ing the net impacts on different types of poorhouseholds, but the various direct and indirectimpacts also affect different types of poor house-holds in different ways. Some poor householdsmay be net gainers while others are net losers. Itis possible, for example, that the indirect growthand food-price benefits are strong and generallyfavorable for the urban poor, but that they areweaker in rural areas and perhaps are even offsetby adverse direct impacts because of inequitableland distribution or poor service support for smallfarmers. Whether there is a net gain or loss to thepoor in aggregate will then depend on how manyof them are rural and the relative size of the gainsand losses that different groups experience.

It is also necessary to distinguish betweenlong- and short-term impacts on the poor. Directbenefits within technology-adopting regions areappealing because they can almost immediatelyand transparently affect the rural poor. Indirectgrowth and food-price benefits take longer tomaterialize, and their links to agricultural technol-ogy are less apparent. However, the indirect ben-efits can be much more important in the long termfor reducing both urban and rural poverty. In Asia,for example, although the Green Revolution hadmixed direct impacts on the rural poor, its contri-butions to productivity enhancement were a majorfactor in reducing food prices and in launchingthe rapid economic growth of the region. Thisgrowth led to significant increases in per capitaincomes (especially in Southeast Asia) and adecline in the number of people living in poverty(Asian Development Bank 2000; Rosegrant andHazell 2000).

17

Chapter 3 highlights the important role that a num-ber of national and local characteristics play inconditioning the ways in which agricultural tech-nologies affect the poor and the difficulty of gen-eralizing about when technological change willbenefit the poor. To target pro-poor agriculturalresearch more effectively, the types of researchundertaken need to be tailored to specific countryand region conditions. A key question is whetherone can go beyond pure site specificity and iden-tify sufficient commonalities across sites to con-struct a typology that can serve as a filteringdevice for selecting appropriate types of agricul-tural research for different socioeconomic con-texts.

A Typology

This chapter develops a typology of agriculturalregions based on agroclimatic and socioeconom-ic factors that condition the size and distributionof the benefits that might be obtained from tech-nological change. Although such a typologywould seem essential for helping to target pro-poor agricultural research, its construction doesnot seem to have been attempted before. Thisattempt is largely exploratory (even speculative)but it is hoped that it lays a useful foundation forfuture work. To construct the typology, the moreimportant conditioning factors of the size of thedirect and indirect benefits identified in the previ-ous chapter are used as classification criteria,with differentiation between national and local orregion-specific factors.

National Characteristics1. Market liberalization policies and nation-

al rural infrastructure development. Countries thatpursue liberal trade policies and liberalize theirdomestic markets typically grow faster than othercountries and this contributes to greater long-termreductions in poverty (World Bank 2000). Thesegrowth effects can be even stronger in agrarian-based economies (for example, much of Africaand South Asia) when trade liberalization is com-plemented by increased investment in rural infra-structure (Robinson, Roe, and Yelden 1998).National investments in agricultural research andrural infrastructure also contribute to agriculturaland rural nonfarm economic growth and to ruralpoverty reduction in their own right, even whenmarkets are not widely liberalized (Fan, Hazell,and Thorat 1999; Fan, Zhang, and Zhang 2000).Agricultural research can be expected to con-tribute more to growth and therefore to long-termpoverty reduction in countries with liberalizedmarkets and good levels of rural infrastructure. Byconnecting farmers to larger markets, both domes-tic and international, and enhancing economicgrowth, farmers may benefit from greater oppor-tunities to diversify into higher-value products.These are typically employment-intensive and canbenefit smallholder farms and landless workers.Moreover, the prices farmers receive for staplesare less likely to fall as production increases,enabling food-surplus farmers to capture largershares of the productivity increases arising fromtechnological change. The downside to this is thatconsumers and food-deficit farmers are less likelyto benefit from price reductions for staple foods as

4. Targeting Agricultural Research to Benefit the Poor

18

production increases, though they may benefitfrom more plentiful supplies of higher value andmore nutritious foods, such as vegetables, fruits,and livestock products. Farmers living in remoteregions also face potential hardships if they areunable to compete in more liberalized marketsbecause of higher transport and marketing costs.Smallholder farms everywhere may also beunable to compete in markets for higher-valueproducts if they are unable to organize and mar-ket their smaller volumes of output competitivelywith large farms.

2. National per capita income level. Ascountries grow and national per capita incomesrise, agriculture’s share in national income dimin-ishes. This has a number of important implicationsfor growth-poverty relationships. First, becausehouseholds allocate increasing shares of theirincome to nonfoods as they get richer, rising percapita incomes have a powerful impact on boththe level and composition of demand for foodsand nonfoods. Rural nonfarm income multiplierstend to be larger with higher per capita incomes,a situation that benefits the rural poor, but broad-er general equilibrium effects on the nationaleconomy get smaller as the agricultural sectorbecomes relatively less important. Second, thestructure of poverty also changes with increases inaverage per capita incomes, with clearer demar-cation of hard-core poor poverty groups that donot easily benefit from agricultural and nationaleconomic growth. These poverty groups tend toconcentrate in urban areas and less-favored ormarginal regions. Third, since agriculture’semployment share typically diminishes at a slowerrate than its income share as per capita incomesrise, labor productivity lags behind that of othersectors. This can lead to worsening inequalitybetween farm and nonfarm incomes, though notnecessarily to worsening poverty. Technologiesand policies to increase average labor productivi-ty therefore become more important as countriesdevelop.

3. The share of poor who are urban. Thedirect benefits of agricultural technology for the

poor will be smaller if the poor are predominantlyurban-based. In these cases, indirect effects suchas food price changes and growth of the nonfarmeconomy will be much more important, especiallyin countries that are not well integrated with glob-al food markets. It is then appropriate to givegreater attention to pure productivity gains indomestic agriculture (with an eye to the relativeprices of foods that are key for the eradication ofdiet-quality problems of the urban poor) ratherthan worrying about their social distribution with-in the rural sector. The opposite holds when thepoor are predominantly engaged in agriculture.

4. Population growth. Population growthadds to the rural labor supply in agrarian coun-tries and, other things being equal, acts to reducewages. Rapid population growth can lead to stag-nant living standards and can mask gains fromtechnological change. Population growth typicallydeclines as per capita incomes rise, but where thisis not true, then agricultural growth should contin-ue to be labor intensive until late in the develop-ment process.

Regional or Local Characteristics1. Agroclimatic conditions. The agroclimatic

conditions of a region determine whether it hashigh or low potential for agriculture. The use ofmodern inputs and the potential to achieve higherlevels of output per hectare and per worker aretypically greater in high potential areas (HPAs),creating greater growth opportunities for reducinglocal poverty. But even many low-potential areas(LPAs) can achieve high levels of output perhectare if they have good infrastructure andaccess to markets (through livestock and tree cropproduction, for example, rather than staple foods).Unfortunately, many LPAs have also been neglect-ed through past patterns of public investment andhave weak infrastructure and market access,which greatly restricts their ability to improve agri-cultural productivity and reduce poverty. Becausethe development of the local nonfarm economyalso hinges crucially on agricultural growth to gen-erate markets for most of its products, there are

19

typically fewer opportunities for farmers and land-less workers to diversify into productive nonfarmactivities in LPAs. In short, the rural nonfarm econ-omy tends to be much weaker in LPAs andstronger in HPAs, reinforcing the initial inequitiesin agroclimatic conditions (Hazell and Reardon,1998).

2. Labor market situation. Agricultural wageearnings for the poor tend to be greater whenagriculture has a high employment elasticity. InSouth Asia, employment elasticities of around0.7–0.8 were observed during the peak of theGreen Revolution (that is, each 1 percent increasein agricultural output led to a 0.7–0.8 percentincrease in agricultural employment), but employ-ment elasticities are probably about half that leveltoday. The rural nonfarm economy also has a highemployment elasticity (often around 1.0), so thatthe combination of an employment-elastic agricul-ture and a vibrant rural nonfarm economy canlead to strong growth in demand for rural labor,which can be very beneficial for the poor. On theother hand, an abundant or elastic supply of labor(such as occurs with rapid population growth butslow agricultural growth) keeps wages down,thereby diluting the benefits of growth in employ-ment opportunities for the poor. Promotion oflabor-intensive crop and livestock technologies areparticularly important in labor-abundant regions,as is the need to avoid policies and investmentsthat lead to premature mechanization of farming.

3. Land distribution and the incidence oflandlessness. An unequal distribution of landreduces the percentage of the rural poor who cangain from own-farm productivity increases andtypically contributes to excessive mechanizationand low employment elasticities in agriculture. Italso leads to weaker demand for local nonfoodgoods and services, and hence to smaller ruralincome and employment multipliers. Technologicalchange rarely contributes to rural poverty reduc-tion in technology-adopting regions when the dis-tribution of land is badly skewed, as amplydemonstrated by some experiences with theGreen Revolution.

4. Infrastructure and the provision of keyagricultural services. These are essential for ensur-ing that poor farmers can adopt new technologiesand take advantage of new income-earningopportunities such as diversification into highervalue products. They also affect the size of thenonfarm income multipliers within an adoptingregion. Regions with poor infrastructure and serv-ice support are not only likely to be poorer, butpoor farmers living within those regions are lesslikely to be able to adopt improved technologies.

5. Local institutions and empowerment.Cultural and institutional factors that work againstthe poor are not easily overcome by economicchanges alone. These factors can include racial,tribal, and religious discrimination, gender biases,exclusive property rights arrangements, localpower structures that favor the rich, and poordelivery of public services to needy groups andless developed regions. In regions where suchbiases are strong, technological change in agri-culture may sometimes work against the interestsof the poor. Improved governance arrangementsand greater empowerment of the poor may thenneed to be a prerequisite for successful pro-pooragricultural growth.

The list of criteria is too long for a practicaltypology, and it is necessary to cluster some ofthese criteria to obtain a more manageable num-ber. Fortunately, some important national charac-teristics are strongly correlated. The World Bankclassifies countries by per capita income level,and two groups of countries are important for thisstudy: low-income countries (LICs) that had 1999per capita incomes below $755 and middle-income countries (MICs) that had 1999 per capi-ta incomes of between $756 and $9,265. AsTable 4 shows, this classification reveals that LICsnot only have lower per capita incomes than MICs($410 versus $2,000) but they also have smallerurban population shares, higher populationgrowth rates, and larger agricultural sector sharesin national output. The income classification of acountry therefore is used as a proxy for these threevariables. But Figure 4 provides a cautionary

20

reminder that there is still considerable variation inthese relationships within the two country groupsand that one must be careful in practice not toassume too much for a particular country situation.

It is also reasonable to merge concern for thelevel of national investment in rural areas andlocal or regional development of infrastructureand agricultural services and to define this vari-able at the regional level. Finally, a strong andpositive association between an inequitable distri-bution of land and a high incidence of landless-ness with institutional and cultural biases againstthe poor is hypothesized. Taken together, thesefactors can be considered an index of empower-ment of the poor at the regional level, defined interms of their access to land and other resources,and to public services.

With these changes, a regional typology canbe defined based on two country characteristicsand four local characteristics:

• region located in a low- versus middle-incomecountry;

• region located in a country with liberalizedversus unliberalized market and trade policies;

• region has high versus low agroclimatic poten-tial for agricultural growth;

• region has good vs. poor rural infrastructure,service provision, and market access;

• region has low wages and abundant laborsupply versus high wages and scarce laborsupply; and

• region has favorable versus unfavorableempowerment of the poor.

Implications for AgriculturalResearch Priorities

Table 5 shows how the typology can be used toset priorities for agricultural research that will bemost appropriate for reducing poverty in differenttypes of regions and country situations. Two typesof priorities are captured in this summary: (1) inter-regional allocation of research resources toachieve the largest reduction in poverty at thenational level and (2) the kinds of research thatare most appropriate within individual types ofregions. Reading across a row (corresponding toa type of country) in Table 5, the shaded cellsdepict the types of regions that should receivehighest research priority within such countries. Thedigital codes in the cells depict priority researchactivities within each type of region. Of course,not all countries have all the types of regionsshown in the table, in which case there are fewerinterregional choices to be made. The unshadedcells in the table depict regions where research ismore likely to lead to smaller net reductions innational poverty than if the same resources were

21

Table 4—Comparison of low- and middle-income countries

Indicator Low-income countries Middle-income countries

GNP per capita (1999 US$) 410 2,000Urban population (% total in 1999) 31 50Annual population growth rate (%), 1990–99 2.0 1.2Agriculture’s share in GDP (% in 1999) 27 10Population density (people/square kilometer in 1999) 73 40Arable land per capita (hectares in 1995–97) 0.19 0.23

Source: World Bank 2000.

22

Figure 4—Relationships between GNP per capita and population growth,

agriculture’s share in GNP, and urbanization in low- and middle-income countries

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

4.5

5.0

1000 2000 3000 4000 5000 6000 7000 8000 9000

0.0

5.0

10.0

15.0

20.0

25.0

30.0

35.0

40.0

45.0

50.0

1000 2000 3000 4000 5000 6000 7000 8000 9000

0.0

10.0

20.0

30.0

40.0

50.0

60.0

70.0

80.0

90.0

100.0

1000 2000 3000 4000 5000 6000 7000 8000 9000

% of population in urban areas

Population growth (%)

GNP per capita (US$)

GNP per capita (US$)

GNP per capita (US$)

Agricultural share in GNP (%)

Source: World Bank 2000.

allocated to shaded regions. But research in theseregions may still benefit important poverty groups.

Although empowerment of the poor was iden-tified as a classification criterion in constructingthe typology, the cells in Table 5 are not disag-gregated by this criterion. This is because thetypes of research considered in the table aredefined at a high level of generality, and theappropriate choice at this level is less one ofselecting the type of research or technology need-ed to empower poor people than of choosing theway in which agricultural research and extensionis conducted. Agricultural research can contributeto the empowerment of the poor if conducted inparticipatory ways, but on its own it has limitedcapacity beyond labor market impacts to directlyovercome the problems of seriously disempow-ered people who are denied access to land andother key resources and public services.

In these cases, new technologies may bringsignificant changes for the poor only if comple-mentary social and political changes are under-taken. Appropriate actions might need to includeland redistribution programs, rental marketreforms, microfinance to increase the assets of thepoor, and improved delivery of public services tothe poor. In some cases, social action may berequired to organize poor people for greater polit-ical voice in local decisions that affect them, or forbetter access to and management of commonproperty like communal grazing and woodlotareas. This often requires direct interventions bygrassroots organizations such as NGOs or thereform of local government. Agricultural researchsystems have little capacity or mandate to under-take this kind of work, but they may need to devel-op new kinds of partnerships with other types ofdevelopment agencies if their technologies are tobenefit the disempowered poor.

In LICs, the importance of staple foods in rurallivelihoods and the national diet requires that highnational priority be given to agricultural researchthat increases staple food production in high-potential areas. This will create larger marketedsurpluses and benefit large numbers of poor peo-

ple. But where national staple food surpluses havealready been achieved, or where infrastructureand markets are weak, production increases willneed to be spread more broadly across regions(including some LPAs) to achieve widespread foodsecurity. Because national demands for higher-value crop and livestock products are limited atlow per capita income levels, significant researchon these products may be relevant only wherethere are major export opportunities. These aremore likely to occur in countries that have liberalmarket and trade policies. However, attempts toimprove the nutrient quality of food staples mayhave high payoffs for improving the nutritionalwell-being of the poor, especially in regions withinadequate infrastructure and market access.

Research priorities for different types ofregions within LICs need to be adjusted to theirkey characteristics if they are to be most helpful tothe local poor. In labor surplus regions, emphasisshould be placed on the development of staplefood technologies that are labor-intensive and areattractive to smallholder farms. In labor-scareregions, technologies should enhance labor pro-ductivity, and this may require focus on someforms of mechanization. Technologies that requireintensive use of fertilizers, purchased seeds, andpesticides will be more relevant for regions thathave high agricultural potential and good infra-structure and market access, while improved natu-ral resource management and low external-inputfarming technologies will be more relevant forLPAs and/or regions with weak infrastructure andinadequate market access. Poor farmers are leastlikely to be able to afford modern inputs in LPAs,a pattern that is reinforced when they are disem-powered in their access to land, other resources,and public services.

In MICs, high priority should be given toincreasing agricultural productivity through diver-sification into higher-value crops and livestock inmost kinds of regions with reasonable infrastruc-ture and market access. Staple foods are stillimportant, but except in MICs that have not liber-alized their trade and markets, staple food

23

Tab

le 5

—P

riori

ties

for

ag

ricu

ltu

ral

rese

arc

h t

o r

ed

uce

na

tion

al

pove

rty

by

typ

e o

f a

dop

tin

g r

eg

ion

Reg

iona

l cha

ract

eris

tics

Goo

d in

frast

ruct

ure

Poor

infr

ast

ruct

ure

Surp

lus

labor

Scarc

e la

bor

Surp

lus

labor

Scarc

e la

bor

Low

Hig

hLo

wH

igh

Low

Hig

hLo

wH

igh

Coun

try

sett

ing

pot

entia

lpot

entia

lpot

entia

lpot

entia

lpot

entia

lpot

entia

lpot

entia

lpot

entia

l

Mid

dle-

inco

me

coun

tryM

arke

ts lib

eral

ized

1, 2

, 3, 5

2, 3

, 5, 8

1, 4

, 64,

6, 8

1, 3

, 5, 7

3, 5

, 81,

4, 6

, 74,

6, 8

Mar

kets

not l

iber

aliz

ed

1, 2

, 3, 5

1, 2

, 3, 5

, 81,

4, 6

1, 4

, 6, 8

1, 3

, 5, 7

, 91,

3, 5

, 81,

4, 6

, 7, 9

1, 4

, 6, 8

Low

-inco

me

coun

tryM

arke

ts lib

eral

ized

3, 5

1, 2

, 3, 5

, 82,

4, 5

, 81,

2, 4

, 5, 6

, 81,

3, 5

, 7, 9

1, 3

, 5, 7

, 91,

4, 5

, 7, 9

1,

4, 5

, 7, 9

Mar

kets

not l

iber

aliz

ed1,

3, 5

, 91,

3, 5

, 8, 9

1, 4

, 5, 8

, 91,

4, 5

, 8, 9

1, 3

, 5, 7

, 91,

3, 5

, 7, 9

1, 4

, 5, 7

, 91,

4, 5

, 7, 9

Prio

ritie

s fo

r agr

icul

tura

l res

earc

h:1.

Sta

ple

food

pro

duct

ion

2. H

igh-

valu

e cr

ops,

tree

s an

d liv

esto

ck3.

Em

ploy

men

t int

ensiv

e gr

owth

4. In

crea

sed

labo

r pro

duct

ivity

5. S

mal

lhol

der f

arm

s6.

Med

ium

and

larg

e fa

rms

7. L

ow e

xter

nal-i

nput

farm

ing

8. H

igh

exte

rnal

-inpu

t far

min

g9.

Nut

ritio

nal c

onte

nt o

f foo

d sta

ples

24

research should receive greatest priority in LPAsthat have poor infrastructure and their own foodsecurity problems, and in HPAs that have contin-uing comparative advantage in growing thesecrops. Smallholders continue to deserve priorityin all labor-surplus areas, but research prioritiesin labor-scarce regions should be consistent withthe need to increase holding sizes, particularly inregions that have good infrastructure and vibrantlocal nonfarm economies and that are well con-nected to urban areas. Labor-intensive technolo-gies remain important in labor-surplus regions,but diversification into higher-value crops andlivestock can also add significantly to localemployment. High external-input farming is morewidely applicable in MICs, except perhaps in

LPAs that have poor infrastructure and marketaccess.

National labor markets typically becometighter and wages rise as MICs develop, and farmincomes generally grow more slowly than non-farm incomes. Maintaining reasonable paritybetween farm and nonfarm incomes requires thatresearch shift toward the development of tech-nologies that increase labor productivity. Lessfavored areas also tend to get left behind andrequire targeted research and other public invest-ments to reduce poverty and food insecurityamong their populations. Investments that are win-win for growth and poverty reduction are neededto avoid significant tradeoffs against the interestsof the poor outside targeted LPAs.

25

Chapter 4 identified several key topics for a pro-poor agricultural research agenda. These include:

1. Increasing production of staple foods in coun-tries where food price effects are still importantand/or that have comparative advantage ingrowing these crops. This includes most LICs,but also many MICs that have not liberalizedtheir trade and markets.

2. Increasing agricultural productivity in manyless-favored lands. Special attention is neededfor heavily populated LPAs, but also someHPAs that are constrained by poor infrastruc-ture and market access. Appropriate technolo-gies for these areas will often have to bebased on low use of external inputs.

3. Helping smallholder farms in all kinds of areasto diversify into higher value products, includ-ing livestock products, especially in countrieswith rapidly growing domestic markets forsuch products (most MICs) and/or access tosuitable export markets.

4. Increasing employment and income-earningopportunities for landless and near-landlessworkers in labor-surplus regions. This is espe-cially important in LICs with growing popula-tions and land scarcity.

5. Developing more nutritious and safer foods toenhance the diets of poor people by investingin agricultural technology that reduces theprice of micronutrient-rich foods in urban and

well-integrated rural areas, increases physicalaccess in remote rural areas, or increases thenutrient content of food staple crops via tradi-tional or transgenic technologies. This is espe-cially important for LICs where poor dietquality generates micronutrient deficienciesthat impair the health of current and future gen-erations.

6. Undertaking agricultural research in ways thatare more empowering to the poor.

If these objectives could be achieved withouthaving to trade off much agricultural growth at thesector level, this would lead to larger indirect ben-efits in the nonfarm economy that would contributeto further rounds of longer-term poverty reduction.In short, agricultural research must target poorpeoples’ problems in ways that are win-win forgrowth and poverty reduction. The following is adiscussion of appropriate research strategies forachieving each of these objectives.

Research Strategies for StapleFood Production

Irrigated and high-potential rainfed areas willcontinue to be the major producers of staplefoods in most countries, and improved technolo-gies and natural resource management (NRM)practices for these areas will be critical for main-taining ample and affordable food supplies.Growth in staple food production will also beimportant in many less-favored lands as a way of

5. Strategies for Pro-Poor Agricultural Research

26

targeting poverty and food security (see below).In high-potential areas with abundant labor in theform of landless workers or small farms, tech-nologies and NRM practices are needed that areemployment intensive (Lipton with Longhurst1989; IFAD 2001). On the other hand, in high-potential areas that must compete for labor intightening labor markets, technologies thatincrease both labor and land productivity arerequired. Technologies also need to reduce pro-duction costs per unit of output so that farmerscan better compete in world markets (includingcompeting with imports) and consumers can ben-efit from cheaper foods and raw materials.

The Green Revolution technologies achievedmany of these goals, but they have now largelyrun their course. There is a need for moreupstream germplasm improvement work, includ-ing biotechnology. Yet at the same time,germplasm improvement needs to be integratedwith better natural resource management prac-tices to contain environmental problems that arenow common in many intensively farmed areas.The more efficient use and better management ofexternal inputs such as chemical fertilizer, irriga-tion water, and/or land-use diversification maylead to higher productivity and reduced environ-mental degradation. Such knowledge-intensivefarming will require improved extension systemsand greater investment in farmer education.

Focusing on staple foods production can bewin-win for growth and poverty reduction whenthe research benefits regions that have a compar-ative advantage in these crops. Countries that lackcomparative advantage in foodgrains but thatnevertheless pursue self-sufficiency in their produc-tion may achieve greater national immunity fromthe vagaries of world food markets, but at the costof economic growth opportunities. Such opportu-nity costs are smaller when the foodgrains aregrown in LPAs rather than HPAs, and the trade-offmay be worthwhile if this leads to significant gainsin food security in those regions.

Research Strategies for Less-Favored Areas

While some types of commodity improvementwork seem vital for less-favored areas—improvingdrought tolerance, yield response to scarce plantnutrients, food nutrient content, pest and diseaseresistance, and livestock health and productivity—there is a growing consensus that major produc-tivity improvements will first come from improvednatural resource management (NRM) practicesand technologies. In areas of poor soils and harshclimate, NRM can lead the way in improving soildepth, organic matter, fertility, and moisture con-tent, and eventually higher yields from fertilizersand improved varieties.

The poor infrastructure and market access thatcharacterize many less-favored areas make theuse of high levels of external inputs uneconomic,placing a premium on low external input (LEI) tech-nologies. LEI technologies are typically labor inten-sive, both seasonally and in total, and this canconstrain their use. Fallows and green manuresalso keep land out of crop production, and com-posting and manuring compete with householdenergy use for scarce organic matter. The chal-lenge is to develop LEI technologies that boostboth labor and land productivity.

Although poorer LFAs need improved tech-nologies for food crops for subsistence and localneeds, sustained increases in per capita incomewill hinge on diversification into higher-value agri-cultural products and nonfarm activities.Successful examples include dairying in India,growing olives in North Africa, and raising horti-cultural crops in parts of Central America (Hazell,Jagger, and Knox 2000).

A key lesson from many past research invest-ments in LFAs is the importance of social and insti-tutional constraints, particularly the effectivenessof indigenous property rights systems and localcapacity for organizing and sustaining collectiveaction for managing natural resources. Figure 5

27

(taken from Knox, Meinzen-Dick, and Hazell1998) plots increasingly secure property rights onthe horizontal axis and increasing levels of collec-tive action on the vertical axis. Some of the mostsuccessful agricultural technologies lie close to theorigin in this figure. For example, the benefits ofhigh-yielding cereal varieties (HYVs)—the lynch-pin of the Green Revolution—could be capturedwithin a single agricultural season and hence didnot require secure property rights. Even share-cropping tenants with single-season leases wereable to adopt these technologies. Moreover, sinceindividual farmers could adopt regardless of whattheir neighbors decided to do, collective actionwas not necessary. These features made adoptiondecisions relatively simple, and they help explainwhy Green Revolution technologies spread soquickly and widely despite considerable diversityin local socioeconomic conditions.

But where research agendas must focus onthe sustainable use of natural resources, localinstitutional issues become much more prominent.

Integrated pest management, for example,requires that all farmers in an area work togeth-er; the technology is not effective if some farmersspray indiscriminately or if planting dates are notsynchronized. However, the returns are relativelyquick, so secure property rights are less of anissue. For these reasons, IPM appears in theupper left corner of Figure 5. In contrast, plantingtrees on farms (agroforestry) is a long-term invest-ment that requires secure property rights. Butsince trees can be planted by individual farmersregardless of what their neighbors do, “farmtrees” appears in the lower right-hand corner. Butmany other technologies for improved NRMrequire both secure property rights and effectivecollective action and therefore appear in theupper right-hand quadrant. Watershed develop-ment, for example, requires secure property rightsbecause it involves long-term investments in checkdams, land contouring, and tree planting in watercatchment areas, and it can be successfully doneonly if the entire community living within the rele-

28

•

• Soil erosion

• Farm trees

•

• Social forestry

•

Security of property rights

Leve

l of c

olle

ctiv

e ac

tion

Watershed development

Integrated pest management

High-yielding varieties

Figure 5—Links between property rights, collective action, and technology adoption

Source: Knox, Meinzen-Dick, and Hazell 1998.

vant landscape is mobilized to support collectiveaction. If these institutional conditions are not met,the technology is not likely to be adopted andmaintained, regardless of its profitability and sci-entific soundness.

Difficulties arise because institutions for prop-erty rights and collective action are rarely formal-ized in developing countries, and communitiesvary widely in their ability to organize and sustainsuch institutions on their own. Socioeconomicresearch has much to contribute toward ourunderstanding of local institutions and the condi-tions under which they are likely to be effective.Without such knowledge, and a correspondingability on the part of policymakers to interveneand to strengthen local institutions when neces-sary, many promising NRM technologies are notlikely to be adopted.

NRM research for less-favored lands shouldbuild on farmers’ own indigenous knowledge andpractical innovations. Some NGOs have beensuccessful in pursuing this agenda and in workingwith local communities to overcome social andinstitutional constraints. The more effective linkingof formal research to these kinds of grassrootsdevelopment activities could lead to real improve-ments in the relevance and uptake of much NRMresearch. But sustained growth in on-farm produc-tivity will also require improvements in cropgermplasm and disease and pest control. Withoutcontinuing improvements in knowledge and genet-ic resources in less-favored areas, productivity willquickly stagnate again. The potential to breedhigher-yielding varieties in many less-favoredlands may be constrained by the low-yield poten-tial of the existing range of genetic material.Existing material has often been selected by manand nature for robustness under harsh and riskygrowing conditions. Transgenic biotechnologymay prove an essential tool for achieving thewider crosses that will be needed (IFAD 2000).

Many agricultural researchers are skepticalabout the efficiency of research in less-favoredareas, arguing that it is better to invest in HPAsbecause these give higher returns to research and

contribute more to agricultural growth perresearch dollar. Growing conditions are seen asdiverse in LFAs, and so improved technologiesmay not have widespread application (in contrastto Green Revolution technologies that spread overtens of millions of hectares of irrigated land).Technologies are also perceived to be more diffi-cult and perhaps more costly to develop. There isundoubtedly some basis for these concerns, espe-cially for commodity improvement research. Onthe other hand, recent evidence from India andChina shows that, dollar for dollar, agriculturalresearch investments in LFAs can outperforminvestments in HPAs in productivity growth andpoverty reduction in some types of less-favoredareas (Fan, Hazell, and Haque 2000; Fan,Zhang, and Zhang 2000).

As should be expected, the returns to addi-tional research investments vary by agroclimaticzone and are very low for both growth and pover-ty in some of the most marginal zones. But therange of conditions under which research invest-ments have significant impact is impressive (Fan,Hazell and Haque, 2000). One reason for theseresults is that India and China have already invest-ed heavily in their irrigated and high-potentialrainfed areas, and productivity growth has slowedin many of these regions. With diminishing returnsin HPAs and relatively little research investment inmany LFAs, it is not surprising that the latter shouldnow give higher returns to research investments onthe margin. But care must be taken in extrapolat-ing these results to other countries that have invest-ed a great deal less in both HPAs and LFAs, forexample, much of Sub-Saharan Africa. In thesecases there could well be some important trade-offs between sectorwide growth and regionalpoverty reduction from targeting researchresources on the problems of LFAs.

One way to reduce the cost of research inLFAs is to focus on NRM problems that are com-mon to a significant number of poor people—butonly problems that can be scaled up from bench-mark sites. The scaling up need not mean that allsites have to be homogenous, just that improved

29

NRM practices can be easily adapted by localpeople and institutions to different site-specific cir-cumstances. The research also should involve rele-vant socioeconomic work to understand the socialand economic constraints on adopting improvedNRM practices, and partnerships need to beformed with institutions with the capacity to over-come these constraints at the grassroots level.

Research Strategies forSmallholder Farmers

As farms get smaller, it becomes increasinglyimportant to develop technologies that increasetheir labor as well as their land productivity.Otherwise, they will simply be working harderand harder to achieve the same level of per capi-ta income.

In order to maintain their incomes, smallhold-er farms will need to diversify beyond traditionalcrop production into profitable activities that arecompatible with their diminishing land/laborratios. Countries with rising per capita incomesoffer expanding opportunities to diversify into live-stock and horticultural production to meet rapidlygrowing domestic demands. These activities typi-cally give high returns per unit of land and arelabor-intensive, and hence well suited to small-farm conditions. Such diversification is alreadyhappening in many countries, especially in Asiaand Latin America. Rising per capita incomes arealso associated with expanding rural nonfarmeconomies, which offer opportunities for small-holder farm families to diversify into nonfarmemployment or into nonfarm businesses. InSoutheast Asia, small farmers seem to be follow-ing the development pattern that occurred inJapan: becoming part-time farmers with regularemployment in nonfarm, often industrial, occupa-tions (Rosegrant and Hazell 2000).

The opportunities for income-enhancing diver-sification are much more constrained in countrieswith low and stagnant per capita incomes. Inthese cases, attention needs to be given to devel-

oping cash crops for export or expanding oppor-tunities for seasonal migration to cities. Otherincome-augmenting measures include creation ofrural processing facilities to enable higher-value-added from agricultural output.

Targeting small rather than large farms willcomplement agricultural sector growth if smallfarms are at least as economically efficient aslarge farms. An impressive body of empirical evi-dence confirms that land productivity is inverselyrelated to farm size in many developing countries(see Heltberg 1998 for a recent review). One rea-son for this is that hired labor is less efficient andmore costly to manage than family labor, givingsmaller farms a competitive advantage. Anotherreason may be the higher management intensitythat is possible on smaller farms. But becausesmall farms use more labor-intensive technologies,their average labor productivity is also lower. Ascountries develop and agriculture must competewith higher-paying nonfarm employment, mecha-nization and more capital-intensive farming meth-ods become important to increase laborproductivity. This requires that farms become big-ger. There is little hard information about whenthis crossover occurs, and it is likely to vary withthe rate of rural population growth (which slowsthe transition), with credit subsidies and other agri-cultural support policies (which often encouragepremature mechanization), and with opportunitiesfor income diversification into nonfarm activities inrural areas (which can have mixed effectsbecause greater opportunities encourage contin-ued part-time farming by small farmers but alsoraise wage rates). Nor does a crossover neces-sarily occur for some employment-intensive typesof farming—for example, high-value horticulturalcrops and intensive livestock production. It seemslikely that the crossover is important for foodgrainproduction by the time countries reach middle-income status, but small farms may retain theircompetitive edge indefinitely for some types ofhorticultural and livestock farming.

Another concern is that, with increasing ruralpopulations but scarce land, small farms may

30

eventually become too small and fragmented to beefficient. Carter and Wiebe (1990) have providedevidence from Kenya showing that profits perhectare decline when farms get too small. It ispremature to draw any definite conclusions aboutthis concern, and, even if true, it is likely that smallfarms first lose their competitive advantage forgrowing foodgrains rather than labor-intensivelivestock and horticultural products.

Even though small farms are generally themore efficient producers, this advantage may beoffset by problems in reaching markets. Becausesmall farms typically trade only in small quantities,both inputs and outputs, they cannot command thesame price advantages as larger farms dealing ingreater bulk. They are also less likely to have thesame access to market information and contacts.Such disadvantages can be offset by coopera-tives, contract farming, and other marketingarrangements, but these often introduce addition-al costs and challenges. Resolving such marketingproblems—once the mandate of now-disbandedmarketing parastatals—may be critical to achiev-ing greater complementarity between targetingsmallholder farms and agricultural productivitygrowth.

Research Strategies forLandless Laborers

Landless laborers are difficult to reach through agri-cultural growth except through the labor market(both locally and through inter-regional migration)and the rural nonfarm economy. In irrigated andhigh-potential areas, agriculture’s employment elas-ticity has shrunk in the post-Green Revolution era inmany countries. In India, for example, the employ-ment elasticity has fallen from about 0.7 in the1970s to about half that level today. This meansthat a 1 percent increase in agricultural outputgenerates a much smaller percentage of addition-al employment than before. The recent removal ofcredit subsidies and policy distortions as part ofstructural adjustment and market liberalization

reforms in many countries may discourage inap-propriate investments in mechanization and helpshift agriculture toward more employment-intensivepatterns of growth, helping landless and near-land-less farmers. Increased investments in improvednatural resource management in less-favored areasshould also help create additional employmentbecause many of these investments are labor inten-sive. In countries where rising per capita incomesare spurring demand for livestock and horticulturalproducts, agricultural employment is likely to growas farms diversify into these activities.

The landless also can benefit from types oflivestock production where animals are kept installs or cages or fed from common propertygrazing resources and purchased feeds (forexample, poultry, rabbits, goats, and dairycows). Rehabilitation and better management ofcommon property resources in general can alsohelp the landless, since the use of these resourcesis often an important element in their livelihoodstrategies.

Increasing employment and labor productivitythrough improvements in labor and capital mar-kets is also good for economic efficiency andoffers win-win opportunities. Targeting researchon the own-agricultural activities of landless work-ers probably adds little to overall agriculturalgrowth rates in the short term, but it could lead tolonger-term growth by increasing worker produc-tivity and the economic contributions of their chil-dren (see next section).

Research Strategies for MoreNutritious Foods

Conventional research approaches can helpimprove productivity and lower the price of non-staple crops that are rich in micronutrients. Thispotentially powerful indirect effect should be seri-ously considered in regions that experience highmicronutrient deficiency and are poorly linked tointernational or domestic markets. The followingapproaches should be considered.

31

New Technology for Small-ScaleHome Gardening of Micronutrient-Rich FoodA recent review of food-based interventions tocombat vitamin A deficiency points to the potentialof home gardening combined with promotionaland education interventions. With regard to iron,production and education interventions toincrease the supply and intake of iron from plantfoods have not been as popular as those for vita-min A. Experience with food-based approaches toincrease production and/or consumption of hemeor non-heme iron-rich foods is limited, but somelessons are clear. In addition to the well-knownproblems of bioavailability with iron from plantsources, the experience with animal productionsuggests trade-offs between increased incomefrom selling home-produced animal products andincreasing one’s own consumption of these prod-ucts to improve dietary quality. Similar to homegardening interventions, a strong nutrition educa-tion component is critical to achieving improveddietary diversity through animal production inter-ventions.

Postharvest Technologies Postharvest activities can affect nutrient availabili-ty by increasing the general use of nutrient-richfoods (for instance, beta-carotene-rich varieties ofsweet potatoes), increasing the nutrient density offoods consumed by infants, and decreasing nutri-ent losses from the processing of widely availablefoods. Postharvest activities include storage, com-mercial processing, in-home processing, andpreparation. Food processing includes physicalprocesses (heat/cold treatment, mechanical sepa-ration such as milling, and reduction of wateractivity), chemical processes (addition of acid,alkaline, oxidizing, and reducing agents), enzy-matic processes (hydrolysis of proteins and inacti-vation of toxins), and biological processes(fermentation and germination) (WHO 1998). Agood example of work that has the potential toincrease the general use of nutrient-rich foods viaprocessing is provided by the work of the

International Potato Center (CIP) with sweet pota-toes. A series of papers have been written thatdescribe the technical and economic feasibility ofderiving and using sweet potato chips and floursin chapati and bread processing, as well as con-sumer acceptance (Hagenimana and Owori1997a and 1997b). The goal of this work is main-ly to reduce the costs of production and the cost tothe consumer. Sweet potatoes performed well inevery area. The successful application of suchtechniques to orange- and yellow-flesh varietiesthat are richer in beta-carotene has directly helpedimprove nutrition in Kenya (Hagenimana andOyunga 1999).

Breeding Techniques to Improve theNutrient Content of Staples There are several ways to improve the nutrientavailability of food staples: (1) increase the densi-ty of micronutrients in the grain, (2) decrease thedensity of factors that inhibit human absorption ofmicronutrients already in the grain, and (3)increase the level of promoters of human absorp-tion. Breeding methods can be conventional ortransgenic.

The CGIAR micronutrients project involvingthe International Rice Research Institute (IRRI), theInternational Center for Tropical Agriculture(CIAT), CIP, and IFPRI is beginning to producesome promising results (Bouis, Graham, andWelch 2000). First, two high-yielding, high-ironrices were identified among improved linesalready being tested by IRRI and now being sub-jected to human feeding trials. Second, 24 select-ed genotypes of beans from CIAT were found tohave substantial variation in iron concentrationand a constant level of bioavailability. The devel-opment of low phytic acid mutants of maize andother cereals for use in food and feed is anotherapproach to improving the micronutrient contentof staples. An important advantage of low phyticacid mutants is that the bioavailability of a rangeof minerals may be improved. Yields of the bestlow phytic acid lines, first developed in the mid-1990s, now range between 5 and 15 percent

32

below those of the highest-yielding commercialvarieties. Because of the benefits for nutrition, lowphytic acid crops may have higher total benefits tothe poor, although the rate of adoption will remaina drawback as long as yields are compromised(Raboy 2000).

Enhancing the nutrition status of the poor willlead to additional productivity growth (both inagriculture and other activities) today and in futuregenerations. Undernutrition involves serious eco-nomic costs. Both stunting and anemia are knownto lower productivity in physical labor. Adults whoare moderately stunted are 2 to 6 percent less pro-ductive, and those who are severely stunted are 2to 9 percent less productive, than those of normalstature (Horton 1999, using estimates in Haddadand Bouis 1991 and Alderman et al. 1997). Irondeficiency anemia is associated with a loss of pro-ductivity of 5 percent in light blue-collar work and17 percent in heavy manual labor (Ross andHorton 1998).

Evidence suggests that childhood stunting andiron deficiency, as well as maternal iodine defi-ciency, are also associated with lower cognitiveoutcomes in children and hence lower adult pro-ductivity in the next generation. The effects onadult earnings and productivity is estimated at 10percent for low-height-for-age (from an older studyby Selowsky and Taylor 1973), 4 percent forchildhood anemia (Ross and Horton 1998), and10 percent per child born to a mother with goiter(Ross 1997). These losses are conservatively 2 to3 percent of GDP in low-income countries. InSouth Asia, the estimated losses associated withiron deficiency alone are estimated to be $5 bil-lion per year (Ross and Horton 1998).

Technologies that improve the nutritional con-tent of the foods that poor people eat need notinvolve trade-offs with other desirable traits.Researchers at the International Crops ResearchInstitute for the Semi-Arid Tropics (ICRISAT), forexample, concluded some years ago that breed-ing for high protein and lysine content in foodgrains was not a priority because calorie con-sumption was a more severe problem for the poor

than low protein or lysine consumption in SouthAsia’s semi-arid tropical areas. As a result, breed-ing for high yield was the best way to help poorpeople and there was no trade-off between effi-ciency and poverty alleviation goals.

However, this is not so clearly the case inbreeding for greater micronutrient content, such asiron and zinc, for regions where poor people suf-fer from serious deficiencies. Such breeding workruns the risk of drawing resources away fromother objectives such as strengthening tolerance tovarious stresses. Even so, the cost is likely to bemuch lower (at least for iron and zinc) than for acapsule program that has to be sustained on along-term basis (Ruel and Bouis 1998).

Research Methods thatEmpower the Poor

Participatory research, in which poor people playa role in setting the research agenda and carryingit out, has potential to make agricultural researchmore effective in empowering poor people. It cangive them more influence over the research systemto address their needs and provide them with theskills needed to solve many of their own problems.

Participatory research is still new, and therehas been little attempt to assess its impact on agri-cultural productivity or poverty reduction. Initialattempts at participatory research have givengreater priority to involving poor people in evalu-ating new technologies (such as specific traits ofimproved varieties) than in setting priorities forresearch itself (such as which kinds of varieties todevelop, and for what types of crops, regions,and farmers). Involving farmers in problem diag-nosis and field-testing of technologies can provideuseful information to researchers and result inmore useful products for farmers (Farrington andMartin 1988). Using participatory research topromote empowerment as a direct goal requiresworking with communities that are organized andskilled in working together, solving problems, andresolving conflicts. Kerr and Kolavalli (1999)

33

argue that “participatory research should gohand-in-hand with participatory community devel-opment that can help improve access to credit andmarkets and can teach local people the skills theyneed to organize themselves, analyze and solveproblems as a group, and resolve conflicts.”

Increasing pressures on research organiza-tions to improve their effectiveness in reaching thepoor will lead to increasing collaboration withfarmers. But Kerr and Kolavalli (1999, p. 137-138) identify several constraints:

More participatory research requires mul-tidisciplinary work that may be difficult toorganize; this was a constraint in earlierfarming systems research (Farrington andMartin 1988). Some scientists are reluc-tant to learn from indigenous knowledge,and economists shy away from participa-tory methods because they do not alwaysyield quantitative data. Scientific journalsare less receptive to research based onparticipatory than traditional methods. Asa result, major changes are still needed in

both researchers’ perceptions and theincentives that guide them. Another con-straint is that initial costs of participatoryresearch can be high because travel andtraining budgets rise. In a project to devel-op pest control measures in Ghana,farmer participation increased projectcosts by 66 percent and accounted for 80percent of researchers’ time (Magrath etal. 1997). However, the Rwanda caseshows that the extra cost can lead to high-er returns by reducing the time needed toidentify promising technologies. The costper variety released actually fell in theRwanda case (Sperling et al. 1993). Also,developing the farmers’ own capabilitiesin developing improved pest managementsystems, or conducting field trials, or evenbreeding can be a cost-effective way toadapt technology to local needs whereconditions are spatially diverse. Whereparticipation may mean the differencebetween success and failure in developingtechnologies, there are no cost trade-offs.

34

Public research and extension systems will have akey role in implementing the pro-poor agriculturalresearch strategy described in the previous chap-ter. Because targeting the poor is not always win-win for growth and poverty alleviation, countriesthat can afford large research budgets, or enjoysignificant private-sector investment in productivityenhancing research, will be best placed to under-take this agenda. Where the trade-offs are high, itis appropriate to consider alternative policies forpoverty alleviation. Technology is only one instru-ment for helping the poor, and it is not always themost effective one. Its role must be seen within thebroader context of rural development and grass-roots development efforts.

Where productivity trade-offs exist but are notoverwhelming, targeting research can sometimesoffer more cost-effective and long-term solutions topoverty than alternative interventions. For exam-ple, research to improve soil nutrient and watermanagement practices or to develop moredrought-tolerant cereal varieties might lead to sub-stantial long-term savings in relief aid in manypoor, drought-prone areas. In these cases, thetrade-off against agricultural productivity growthmay be less important than the cost savingsachieved in alternative relief programs, especiallyif government and donors can be persuaded toallocate some of those savings to additional agri-cultural research investments.

If public research and extension systems are tobe effective in undertaking targeted pro-poorresearch, they will also have to make some insti-tutional adjustments. Their past successes lay indeveloping and spreading Green Revolution tech-nologies in high-potential areas. Because yield

ceilings had already been reached and were inneed of basic biological breakthroughs, andbecause crops were grown in monoculture sys-tems under homogenous conditions with goodaccess to markets, the new technologies could bewidely adopted with little if any local adaptation.

But as discussed in previous chapters, thechallenges facing public research and extensionsystems today are rather different. In irrigated andhigh-potential areas, crop diversification andimproved environmental management havebecome key challenges for small and large farmsalike, and the problems facing less-favored landsare much more diverse and location-specific andinvolve changes in complex natural resource man-agement systems that have been developed overgenerations to cope with uncertain rainfall andweather conditions, poor and often fragile soils,and the high costs of external inputs, given poormarket access. Moreover, research needs to beundertaken in more participatory ways if it is tobecome more effective in empowering the poor.At the same time, public research systems alsoneed to become more engaged in biotechnologyresearch to ensure that at least part of its potentialis used to enhance agricultural growth and to helppoor farmers.

To meet these challenges, there is a need fora more client-oriented, problem-solving approachthroughout the agricultural research system, anapproach not limited to a particular kind of tech-nology or a particular type of agriculture or zone.This approach will often translate into a need formore on-farm research under conditions that aredifficult and diverse. Not all of the technologicalchallenges facing poor people will be solved by

6. The Role of Public Research and Extension Systems

35

more on-farm work; biotechnology conducted in astrict laboratory environment may be critical, forexample, in raising yield ceilings or for improvingdrought tolerance. However, even biotechnologywill be more effective if it addresses prioritiesbased on a client-oriented, problem-solvingapproach that draws many of its insights frominteraction with farmers.

Institutional reforms are necessary to changeincentive structures within public research andextension systems so that scientists and extensionofficers are responsive to the needs of their clients.But to be effective, these changes will need toextend to all levels of management (Byerlee andAlex 1998). The kinds of changes needed in pub-lic agricultural research and extension systems willalso require the forging of new partnershipsbetween the public system and NGOs, private-sector firms, and farmers.

Private seed companies and input suppliersare playing a larger role as many countries liber-alize and privatize their agricultural input mar-kets. Many of these companies not only developimproved products of their own (including under-taking agricultural research), but also advisefarmers about the use of products they sell.NGOs have also become important actors inspreading natural resource management prac-tices regarding soil and water management,watershed development, and social forestry. Theyhave a particular advantage in helping commu-nities take collective action to implementimproved natural resource management practicesat the landscape level (Kerr et al. 2000).Farmers, including smallholder and women farm-ers, also need to be actively involved in thedesign and evaluation of research intended for

their farming conditions. Participatory researchmethods are proving to be a fruitful way ofachieving this goal (Kerr et al. 2000).

Because of the public-goods nature of muchagricultural research and extension, market forcesalone will be insufficient to integrate the roles ofthe public and private sectors and NGOs andmake them fully accountable to their clients. Newinstitutional mechanisms will also be required.Competitive research grants offer one interestingapproach, particularly if they are open to all rele-vant public, private, and NGO agencies, and iffarmers (including poor farmers and women) arerepresented in the decisionmaking process(Farrington and Martin 1998). Local researchcommittees composed of relevant stakeholdersshow some promise as a device for settingresearch priorities for the public sector (Alsop1998).

Even as agricultural research and extensionsystems are being asked to take on more diverseand difficult challenges, their budgets are beingcut in many countries (Pardey and Beintema2001). National agricultural research systems(NARS) in Africa have been particularly affected,and the availability of resources per scientist hasfallen sharply. On average, developing countriesspend about one half of 1 percent of their agri-cultural GDP on public research, which is muchless than the 2 percent averaged by the industri-alized countries. If new technologies are to bedeveloped to address the poverty and environ-mental problems of developing countries, and toenable them to capture some of the potential ben-efits of biotechnology, then there is an urgent needto increase available funding and to implementneeded institutional reforms.

36

ACC/SCN (United Nations AdministrativeCommittee on Coordination/Sub-Committee onNutrition). 2000. Fourth report on the worldnutrition situation. Geneva: ACC/SCN in col-laboration with IFPRI.

Adelman, I. 1984. Beyond export-led growth.World Development 12 (9): 937–949.

Ahluwalia, M. S. 1978. Rural poverty and agricul-tural performance in India. Journal ofDevelopment Studies 14 (3): 298–323.

Alderman, H., J. Behrman, V. Lavy, and R. Menon.1997. Child nutrition, child health, and schoolenrollment. Policy Research Working Paper1700. Washington, D.C.: World Bank.

Alsop, R. 1998. Coalitions and the organization ofmultiple-stakeholder action: A case study ofagricultural research and extension inRajasthan, India. Environment and ProductionTechnology Division Discussion Paper No. 34.Washington, D.C.: IFPRI.

Alston, J., G. Norton, and P. Pardey. 1995. Scienceunder scarcity: Principles and practice for agri-cultural research evaluation and priority setting.Ithaca, N.Y.: Cornell University Press.

Asian Development Bank. 2000. Rural Asia: Beyondthe Green Revolution. Manila.

Bardhan, P. K. 1973. On the incidence of poverty inrural India in the sixties. Economic and PoliticalWeekly 4, 5, and 6 (February).

Bebbington, A. 1999. Capitals and capabilities: Aframework for analyzing peasant viability, rurallivelihoods and poverty. World Development27 (12): 2021–2044.

Bell, C., P.B.R. Hazell and R. Slade. 1982. Projectevaluation in regional perspective: A study of anirrigation project in northwest Malaysia.Baltimore, MD.: Johns Hopkins University Press.

Blyn, G. 1983. The Green Revolution revisited.Economic Development and Cultural Change31 (4): 705–725.

Bouis, H.E. 2000. Development of modern varieties ofrice: Impacts on food security and poverty. Paperprepared for presentation at the International RiceResearch Institute Conference, Los Banos, Laguna,Philippines, March 31-April 3.

Bouis, H., R. Graham, and R. Welch. 2000. TheCGIAR Micronutrients Project: Justification andobjectives. Food and Nutrition Bulletin 21 (4):374–381.

Byerlee, D., and G. E. Alex. 1998. Strengtheningnational agricultural research systems: Policyissues and good practice. Washington, D.C.:World Bank, Rural Development Department.

Carney, D., ed. 1998. Sustainable rural livelihoods:What contribution can we make? London:Department for International Development.

Carter, M. R., and K. D. Wiebe. 1990. Access andcapital as impact on agrarian structure and pro-ductivity in Kenya. American Journal ofAgricultural Economics 72 (5): 1146–1150.

CGIAR (Consultative Group on InternationalAgricultural Research)/Technical AdvisoryCommittee (TAC). 2000. CGIAR research pri-orities for marginal lands. Washington, D.C.

Conway, G. 1997. The Doubly Green Revolution.London: Penguin Books.

References

37

Datt, G., and M. Ravallion. 1997. Macroeconomiccrises and poverty monitoring: A case study forIndia. Review of Development Economics 1 (2):135–152.

David, C., and K. Otsuka, eds. 1994. Modern ricetechnology and income distribution in Asia.Boulder, Colo., U.S.A.: Lynne Reinner.

De Franco, M., and R. Godoy. 1993. Potato-ledgrowth: The macroeconomic effects of techno-logical innovations in Bolivian agriculture.Journal of Development Studies 29 (3):561–587.

Fan, S., and P. Hazell. 2000. Should developingcountries invest more in less-favored lands? Anempirical analysis of rural India. Economic andPolitical Weekly 35 (17): 1455–1464.

Fan, S., P. Hazell, and T. Haque. 2000. Targetingpublic investments by agroecological zone toachieve growth and poverty alleviation goals inrural India. Food Policy 25 (4): 411–428.

Fan S., P. Hazell, and S. Thorat. 1999. Linkagesbetween government spending, growth andpoverty in rural India. Research Report 110.Washington, D.C.: IFPRI.

Fan, S., L. Zhang, and X. Zhang. 2000. Growthand poverty in rural China: The role of publicinvestments. Environment and ProductionTechnology Division Discussion Paper No. 66.Washington, D.C.: IFPRI.

Farmer, B. H. 1977. Green Revolution? Technologyand change in rice growing areas of TamilNadu and Sri Lanka. London: Macmillan.

Farrington, J., and A. Martin. 1988. Farmer partici-pation in agricultural research: A review of con-cepts and practices. Agricultural AdministrationUnit Occasional Paper 9. London: OverseasDevelopment Institute.

Frankel, F. R. 1976. India’s Green Revolution:Economic gains and political costs. Princeton,N.J.: Princeton University Press.

Freebairn, D. K. 1995. Did the Green Revolutionconcentrate incomes? A quantitative study of

research reports. World Development 23 (2):265–279.

Gaiha, R. 1989. Poverty, agricultural productionand price fluctuations in rural India: A reformu-lation. Cambridge Journal of Economics 13 (2):333–352.

Ghose, A. K. 1989. Rural poverty and relativeprices in India. Cambridge Journal ofEconomics 13 (2): 307–331.

Gibb, A. 1974. Agricultural modernization, non-farm employment and low-level urbanization: Acase study of a Central Luzon sub-region. Ph.D.thesis, University of Michigan, Ann Arbor.

Graham, R. D., and R. M. Welch. 1996. Breedingfor staple food crops with high micronutrientdensity. Agricultural Strategies forMicronutrients Working Paper 3. Washington,D.C.: IFPRI.

Griffin, K. 1974. The political economy of agrarianchange. London: Macmillan.

Griffin, K., and A. K. Ghose. 1979. Growth andimpoverishment in the rural areas of Asia.World Development 7 (4/5): 361–384.

Haddad, L., and H. Bouis 1991. The impact of nutri-tional status on agricultural productivity: Wageevidence from the Philippines. Oxford Bulletinof Economics and Statistics 53 (1): 45–68.

Haddad, L., J. Hoddinott, and H. Alderman, eds.1997. Intrahousehold resource allocation:Models, methods, and policy. Baltimore, Md.:Johns Hopkins University Press for IFPRI.

Hagenimana, V., and C. Owori. 1997a. Sweetpotatoes in Chapatis processing: Feasibility andacceptability in rural areas. Journal of FoodTechnology in Africa 2 (1) 4–8.

Hagenimana, V., and C. Owori. 1997b. Feasibility,acceptibility and production costs of sweet pota-to baked products in Lira Municipality, Uganda.In Proceedings of the Fourth Triennial Congressof the African Potato Association. Pretoria,South Africa: African Potato Association.

38

Hagenimana, V., and M. A. Oyunga. 1999. Theeffects of women farmers’ adoption of orange-fleshed sweet potatoes: Raising Vitamin Aintake in Kenya. Research Report Series No. 3.Washington, D.C.: International Center forResearch on Women.

Haggblade, S., and P. Hazell. 1989. Agriculturaltechnology and farm-nonfarm growth linkages.Agricultural Economics 3 (4): 345–364.

Haggblade, S., P. Hazell, and J. Brown. 1989.Farm-nonfarm linkages in rural Sub-SaharanAfrica. World Development 17 (8):1173–1202.

Hazell, P., and S. Haggblade. 1991. Rural-urbangrowth linkages in India. Indian Journal ofAgricultural Economics 46 (4): 515–529.

Hazell, P., and S. Haggblade. 1993. Farm-nonfarmgrowth linkages and the welfare of the poor. InIncluding the poor, ed. M. Lipton and J. van derGaag. Washington, D.C.: World Bank.

Hazell, P., and C. Ramasamy. 1991. The GreenRevolution reconsidered: The impact of high-yielding rice varieties in South India. Baltimore,Md.: John Hopkins University Press for IFPRI.

Hazell, P., and T. Reardon. 1998. Interactionsamong the rural nonfarm economy, poverty,and the environment in resource-poor areas.Paper presented at an IFPRI/World Bank-spon-sored workshop on strategies for stimulatinggrowth of the rural nonfarm economy in devel-oping countries, May, Warrenton, Va., U.S.A.

Hazell, P., P. Jagger, and A. Knox. 2000.Technology, natural resources management andthe poor. Background paper prepared forInternational Fund for AgriculturalDevelopment’s Rural Poverty Report 2000 andavailable in The Report of IFAD’s Workshop onRural Poverty, Rome, January 24–25. Rome:IFAD.

Heltberg, R. 1998. Rural market imperfections andthe farm size–productivity relationship:

Evidence from Pakistan. World Development26 (10): 1807–1826.

Horton, S. 1999. Opportunities for investments innutrition in low-income Asia. AsianDevelopment Review 17 (1, 2): 246–273.

IFAD (International Fund for Agriculture). 2001.Rural poverty report 2001: The challenge ofending rural poverty. Oxford: Oxford UniversityPress.

ILO (International Labor Organization). 1977.Poverty and landlessness in rural Asia. Geneva.

Jazairy, I., M. Alamgir, and T. Panuccio. 1992. Thestate of world rural poverty: An inquiry into itscauses and consequences. Rome: InternationalFund for Agricultural Development.

Kerr, J., and S. Kolavalli. 1999. Impact of agricul-tural research on poverty alleviation:Conceptual framework with illustrations fromthe literature. Environment and ProductionTechnology Division Discussion Paper No. 56.Washington, D.C.: IFPRI.

Kerr, J., G. Pangare, V. L. Pangare, and P. J.George. 2000. Sustainable agriculture and nat-ural resource management in India’s semi-aridtropics. In Tradeoffs or synergies? Agriculturalintensification, economic development and theenvironment, ed. D. R. Lee and C. B. Barrett.Wallingford, U.K.: CAB International.

Knight, J., and L. Song. 1992. Income inequality inrural China: Communities, households andresource mobility. Oxford Applied EconomicsDiscussion Paper Series No. 150. Oxford:University of Oxford Institute of Economics andStatistics.

Knox, A., R. Meinzen-Dick, and P. Hazell. 1998.Property rights, collective action and technolo-gies for natural resource management: A con-ceptual framework. CGIAR System-wideProgram on Collective Action and PropertyRights (CAPRI) Working Paper No. 1.Washington, D.C.: IFPRI.

39

Lipton, M., with R. Longhurst. 1989. New seeds andpoor people. Baltimore, Md.: Johns HopkinsUniversity Press.

Magrath, P. 1997. Cost benefit analysis of clientparticipation in agricultural research: A casestudy from Ghana. ODI Agricultural Researchand Extension Network Paper No. 74. London:Overseas Development Institute.

Mellor, J. W. 1976. The new economics of growth:A strategy for India and the developing world.Ithaca, N.Y.: Cornell University Press.

Owens, T., and J. Hoddinott. 1998. Investing indevelopment or investing in relief: Quantifyingthe poverty trade-offs using Zimbabwe house-hold panel data. Washington, D.C.: IFPRI.Photocopy.

Pardey, P., and N. Beintema. 2001. Losing ground:What’s happened with agricultural researchregarding less-developed countries?Washington, D.C.: IFPRI. Photocopy.

Pearse, A. 1980. Seeds of plenty, seeds of want.London: Oxford University Press.

Pingali, P. L., and M. W. Rosegrant. 2001. Intensivefood systems in Asia: Can the degradation bereversed? In Tradeoffs or synergies? Agriculturalintensification, economic development and theenvironment, ed. D. R. Lee and C. B. Barrett.Wallingford, U.K.: CAB International.

Pingali, P. L., M. Hossain, and R. V. Gerpacio.1997. Asian rice bowls: The returning crisis?Wallingford, U.K.: CAB International in associ-ation with the International Rice ResearchInstitute.

Pinstrup-Andersen, P., and P. Hazell. 1985. Theimpact of the Green Revolution and prospectsfor the future. Food Reviews International 1 (1):1–25.

Pinstrup-Andersen, P., R. Pandya-Lorch, and M.Rosegrant. 1997. The world food situation:Recent developments, emerging issues, andlong-term prospects. 2020 Vision Food PolicyReport. Washington, D.C.: IFPRI.

Raboy, V. 2000. Research on bioavailability: Lowphytic-acid grains. Food and Nutrition Bulletin21 (4): 423–427.

Rangarajan, C. 1982. Agricultural growth andindustrial performance in India. ResearchReport 33. Washington, D.C.: IFPRI.

Reardon T., K. Stamoulis, M. E. Cruz, A. Balisacan,J. Berdugue, and K. Savadogo. 1998.Diversification of household incomes into non-farm sources: Patterns, determinants and effects.Paper presented at an IFPRI/World Bank-spon-sored workshop on strategies for stimulatinggrowth of the rural nonfarm economy in devel-oping countries, May, Warrenton, Va., U.S.A.

Robinson, S., T. L. Roe, and A. E. Yelden. 1998.Macroeconomic policies and the rural nonfarmeconomy: A general equilibrium investigationof alternative development strategies in anarchetype model for Africa and South Asia.Paper presented at an IFPRI/World Bank-spon-sored workshop on strategies for stimulatinggrowth of the rural nonfarm economy in devel-oping countries, Warrenton, Va., U.S.A.

Rosegrant, M., and P. Hazell. 2000. Transformingthe rural Asia economy: The unfinished revolu-tion. Hong Kong: Oxford University Press forthe Asian Development Bank.

Ross, J. S. 1997. Calculating the effects of malnutri-tion on economic productivity and survival.PROFILES guidelines. Washington, D.C.:Academy for Educational Development.

Ross, J. S., and S. Horton. 1998. Economic conse-quences of iron deficiency. Ottawa:Micronutrient Initiative.

Ruel, M., and H. Bouis. 1998. Plant breeding: Along-term strategy for the control of zinc defi-ciency in vulnerable populations. AmericanJournal of Clinical Nutrition 68 (2s):488s–494s.

Saith, A. 1981. Production, prices and poverty inrural India. Journal of Development Studies 19(2): 196–214.

40

Scobie, G. M., and R. Posada. 1978. Technicalchange and income distribution: The case ofrice in Columbia. American Journal ofAgricultural Economics 60 (1): 85–92.

Selowsky, M., and L. Taylor. 1973. The economicsof malnourished children: An example of disin-vestment in human capital. EconomicDevelopment and Cultural Change 22 (1):17–30.

Sharma, M., M. Garcia, A. Qureshi, and L. Brown.1996. Overcoming malnutrition: Is there anecoregional dimension? Food, Agriculture, andthe Environment Discussion Paper 10.Washington, D.C.: IFPRI.

Smith, L. C., and L. Haddad. 2000. Overcomingchild malnutrition in developing countries: Pastachievements and future choices. Food,Agriculture, and the Environment DiscussionPaper 30. Washington, D.C.: IFPRI.

Spencer, D. 1994. Infrastructure and technologyconstraints to agricultural development in thehumid and subhumid tropics of Africa.Environment and Production TechnologyDivision Discussion Paper No. 3. Washington,D.C.: IFPRI.

Sperling, L., M. Loevinsohn, and B. Ntabomvura.1993. Rethinking the farmer’s role in plantbreeding: Local bean experts and on-station

selection in Rwanda. Experimental Agriculture29 (4): 509–519.

Tripp, R. 1990. Does nutrition have a place in agri-cultural research? Food Policy 15 (6):467–474.

Walker, T., and J. Ryan. 1990. Village and house-hold economies in India’s semi-arid tropics.Baltimore, Md.: Johns Hopkins University Press.

Welch, R., and R. Graham. 2000. A new paradigmfor world agriculture: Productive, sustainable,nutritious, healthful food systems. Food andNutrition Bulletin 21 (4): 361–366.

Westley, J. R. 1986. Agriculture and equitablegrowth: The case of Punjab-Haryana. WestviewSpecial Studies in Agriculture Science andPolicy. Boulder, Colo., U.S.A.: Westview Press.

World Bank. 2000. World development report2000/2001: Attacking poverty. Washington,D.C.

World Health Organization (WHO). 1998.Complementary feeding of young children indeveloping countries: A review of current scien-tific knowledge. Geneva.

Zhang, X., and S. Fan. 2001. Public investment andregional inequality in rural China. Environmentand Production Technology Division DiscussionPaper No. 71. Washington, D.C.: IFPRI.

41

Peter Hazell is director of the Environment and Production Technology Division of the International Food Policy ResearchInstitute (IFPRI). Lawrence Haddad is director of the Food Consumption and Nutrition Division of IFPRI.