DOI: 10.1126/science.1078710 , 758 (2003); 300Science

et al.R. E. Evenson,to 2000Assessing the Impact of the Green Revolution, 1960

This copy is for your personal, non-commercial use only.

. clicking herecolleagues, clients, or customers by , you can order high-quality copies for yourIf you wish to distribute this article to others

. herefollowing the guidelines can be obtained byPermission to republish or repurpose articles or portions of articles

(this information is current as of January 26, 2010 ):The following resources related to this article are available online at www.sciencemag.org

http://www.sciencemag.org/cgi/content/full/300/5620/758version of this article at:

including high-resolution figures, can be found in the onlineUpdated information and services,

found at: can berelated to this articleA list of selected additional articles on the Science Web sites

http://www.sciencemag.org/cgi/content/full/300/5620/758#related-content

97 article(s) on the ISI Web of Science. cited byThis article has been

http://www.sciencemag.org/cgi/content/full/300/5620/758#otherarticles 18 articles hosted by HighWire Press; see: cited byThis article has been

http://www.sciencemag.org/cgi/collection/economicsEconomics

: subject collectionsThis article appears in the following

registered trademark of AAAS. is aScience2003 by the American Association for the Advancement of Science; all rights reserved. The title

CopyrightAmerican Association for the Advancement of Science, 1200 New York Avenue NW, Washington, DC 20005. (print ISSN 0036-8075; online ISSN 1095-9203) is published weekly, except the last week in December, by theScience

on

Janu

ary

26, 2

010

ww

w.s

cien

cem

ag.o

rgD

ownl

oade

d fr

om

Assessing the Impact of the Green Revolution,1960 to 2000

R. E. Evenson* and D. Gollin

We summarize the findings of a recently completed study of the productivity impactsof international crop genetic improvement research in developing countries. Over theperiod 1960 to 2000, international agricultural research centers, in collaboration withnational research programs, contributed to the development of “modern varieties” formany crops. These varieties have contributed to large increases in crop production.Productivity gains, however, have been uneven across crops and regions. Consumersgenerally benefited from declines in food prices. Farmers benefited only where costreductions exceeded price reductions.

T he development of modern or high-yielding crop varieties (MVs) for devel-oping countries began in a concerted

fashion in the late 1950s. In the mid-1960s,scientists developed MVs of rice and wheat thatwere subsequently released to farmers in LatinAmerica and Asia. The success of these MVswas characterized as a “Green Revolution.”Early rice and wheat MVs were rapidly adoptedin tropical and subtropical regions with goodirrigation systems or reliable rainfall. TheseMVs were associated with the first two majorinternational agricultural research centers(IARCs): the International Center for Wheatand Maize Improvement in Mexico (CIM-MYT) and the International Rice Research In-stitute in the Philippines (IRRI). There are now16 such centers that operate under the auspicesof the Consultative Group for International Ag-ricultural Research (CGIAR) (1). These centerscurrently support about 8500 scientists and sci-entific staff, and the annual budget of theCGIAR is currently around $340 million.

A recent study initiated by the SpecialProject on Impact Assessment (SPIA) of theCGIAR’s Technical Advisory Committee(TAC) has compiled the most extensive datayet assembled on the breeding, release, anddiffusion of MVs (2). The SPIA study allowsfor a detailed analysis of the impact of inter-national research for 11 major food crops, byregion and country, for the period 1960 to2000 (3). Here we summarize and report themajor findings of the SPIA study.

In focusing on the impact of internationalresearch, we do not in any sense disparage thework of national agricultural research systems(NARS), which played a crucial role in creating

varieties suitable for farmers. Strong nationalprograms have provided effective research inmany developing countries, and some are lead-ers in the science and technique of plant breed-ing. The SPIA study specifically considers theinteraction between IARC plant-breeding pro-grams and NARS plant-breeding programs andfinds that the two generally fill complementaryroles (4).

Breeding of Modern VarietiesThe early successes in breeding rice and wheatMVs reflected the advanced state of research onthose crops in the late 1950s. Researchers atIRRI and CIMMYT had access to rich stocks ofgenetic resources and drew on extensive breed-ing experience in developed countries. For bothcrops, breeders incorporated dwarfing genesthat allowed the development of shorter, stiff-strawed varieties. These varieties devoted muchof their energy to producing grain and relativelylittle to producing straw or leaf material. Theyalso responded better to fertilizer than tradition-al varieties. Farmers adopted the new semid-warf MVs rapidly in some areas—chiefly thosewith access to irrigation or reliable rainfall—and the new varieties yielded substantially moregrain than previous varieties (5, 6).

The early success of these MVs was wide-ly referred to as the “Green Revolution,” andpopular accounts have tended to equate theGreen Revolution with the initial wave ofMV releases in the late 1960s and early1970s. Our findings suggest, however, thatthis early episode of MV adoption was onlythe beginning of the Green Revolution. Overthe following years, the Green Revolutionachieved broader and deeper impacts, extend-ing far beyond the original successes of riceand wheat in Latin America and Asia.

For many other crops, however, breedingwork aimed at the developing world could notrely on prior work in developed countries. Incassava or tropical beans, for example, there

was essentially no research or elite germplasmavailable in the 1960s (7, 8). As a result, thedevelopment of MVs was slower for thesecrops. But over the following decades, interna-tional research led to the development of im-proved varieties in all 11 crops studied. By2000, the SPIA study documents more than8000 MVs released in the 11 crops studied (Fig.1). These MVs were released by more than 400public breeding programs and seed boards inover 100 countries (9). Contrary to some viewsof the Green Revolution, the rate of MV releaseshas actually increased since the 1960s. Thereare, however, a number of important disparitiesin the development of MVs, especially by agro-ecological zone (AEZ). For sorghum, millet,and barley—crops grown primarily under semi-arid and dryland conditions—few MVs werebred until the 1980s. The same was true for themajor pulses and for root crops—especially cas-sava. Regional disparities were also important.Even for maize and rice, few varieties wereavailable until the 1980s for the Middle East–North Africa and for Sub-Saharan African coun-tries (10, 11).

Adoption of Modern VarietiesWhen a farmer chooses to adopt a new varietyto replace an older variety, it reflects the farm-er’s judgment that the new variety offers somenet benefit or advantage. For most crops, inmost regions, MV adoption occurred soon afterMVs were released (Fig. 2). There are, however,important differences across crops and regionsin the date at which significant adoption of MVsfirst occurred and in the subsequent growth ratesof MV adoption. For example, although largenumbers of MVs were released in Sub-SaharanAfrica in the 1960s and 1970s, there was littleMV adoption by farmers, except for wheat. Thedata suggest that in the 1960s and 1970s, na-tional and international programs may havesought to “short-cut” the varietal improvementprocess in Sub-Saharan Africa by introducingunsuitable varieties from Asia and Latin Amer-ica, rather than engaging in the time-consumingwork of identifying locally adapted germplasmand using it as the basis for breeding new vari-eties. This pattern remained until the 1980s,when more suitable varieties finally becameavailable—based on research targeted specifi-cally to African conditions (10, 11).

More generally, diffusion patterns reflect theimportance of location-specific breeding. For

1Department of Economics, Yale University, New Ha-ven, CT 06520, USA. Department of Economics, Wil-liams College, Williamstown, MA 01267, USA.

*To whom correspondence should be addressed. E-mail: robert.evenson@yale.edu

REVIEW

2 MAY 2003 VOL 300 SCIENCE www.sciencemag.org758

on

Janu

ary

26, 2

010

ww

w.s

cien

cem

ag.o

rgD

ownl

oade

d fr

om

most crops, researchers sought first to develop aproductive “plant type” (for example, a high-yielding semidwarf ) for each major AEZ toserve as a platform for local adaptation, and thensubsequently bred for location-relevant traits—such as resistance to diseases, pests, and abioticstresses. This second-stage research proved ex-tremely important. For India, the SPIA studysuggests that the first generation of improvedrice varieties (the basic semidwarf plant type)would have been planted on only �35% ofirrigated and rainfed rice land. The subsequentgenerations of MVs increased adoption to morethan 80%, with large benefits ensuing for bothproducers and consumers (12).

Direct and Indirect Contributions ofIARC ProgramsFor most of the MVs in thestudy, complete or near-com-plete genealogies could beconstructed. The SPIA studyanalyzed these genealogies tolook for two types of interna-tional contributions to varietalimprovement. Direct contri-butions were defined as vari-eties developed in internation-al institutions and then re-leased by national programswithout further crossing. Indi-rect contributions were de-fined to include varieties thatwere crossed in NARS pro-grams but that had parents orancestors bred in IARCs.

The evidence on contri-butions points to severalstriking results:

1) Large IARC contri-butions. More than 35% ofMVs released and adopted

were based on crosses made in IARCs. Fif-teen percent of NARS-crossed MVs had anIARC-crossed parent, and an additional 7%had another IARC-crossed ancestor (13). Va-rieties with IARC ancestry were also morewidely planted than other varieties.

2) Low international flows of NARS-crossed MVs. For rice, where such data wereavailable, only 6% of MVs originated whenone national program released a variety thatwas crossed by a NARS in another develop-ing country. By contrast, most IARC-crossedMVs were released in several countries.

3) Negligible developed country contribu-tions. Fewer than 1% of MVs included intheir genealogies any crosses made in publicor private sector plant-breeding programs indeveloped countries.

4) Small private sectorcontributions. Private sec-tor contributions were lim-ited to “hybrid” varietiesof maize, sorghum, andmillet. Private sectorbreeding programs forthese crops were devel-oped only after “platform”varieties were developedin IARC and NARS pro-grams. It should be notedthat genetically engineeredMVs appeared only after1996 and have been plant-ed in only three or fourdeveloping countries.

5) IARC research com-plemented NARS breed-ing. By providing im-proved germplasm forNARS breeding programs,international breeding ef-

forts increased the productivity of nationalprograms. Because of this IARC-NARScomplementarity, the existence of the inter-national centers actually stimulated nationalinvestment in NARS research.

Production, Area, and Yield Growth:MV ContributionsUltimately, the release of new MVs is not ameasure of research success. Farmers must firstadopt MVs. This will lead to increased produc-tion and yield (14). Table 1 provides data on theproduction impacts of MVs over the past 40years. Not all of the production growth from1961 to 2000 was due to MVs; this table showshow production growth can be disaggregatedinto area growth and yield growth. Yield growthin turn can be decomposed into the contributions

Fig. 1. Modern variety production by decade and region.

Per

cen

t ar

ea

pla

nte

dto

mo

der

n v

arie

ties

Per

cen

t ar

ea

pla

nte

dto

mo

der

n v

arie

ties

Per

cen

t ar

ea

pla

nte

dto

mo

der

n v

arie

ties

Per

cen

t ar

ea

pla

nte

dto

mo

der

n v

arie

ties

Latin America Asia

Middle East-North Africa Sub-Saharan Africa

1960s 1970s 1980s 1990s 1960s 1970s 1980s 1990s

1960s 1970s 1980s 1990s 1960s 1970s 1980s 1990s

0—

20—

40—

60—

80—

100—

Protein Crops Root Crops Other Cereals Maize Rice Wheat

0—

20—

40—

60—

80—

100—

0—

20—

40—

60—

80—

100—

0—

20—

40—

60—

80—

100—

Fig. 2. Modern variety diffusion by decade and region.

R E V I E W

www.sciencemag.org SCIENCE VOL 300 2 MAY 2003 759

on

Janu

ary

26, 2

010

ww

w.s

cien

cem

ag.o

rgD

ownl

oade

d fr

om

of MVs and the contributions of all other inputs(e.g., fertilizer, irrigation, mechanization, and la-bor) (4).

One striking feature of the data in Table 1is that the gains from MVs were larger in the1980s and 1990s than in the preceding twodecades—despite popular perceptions thatthe Green Revolution was effectively over bythis time. Overall, the productivity data sug-gest that the Green Revolution is best under-stood not as a one-time jump in production,occurring in the late 1960s, but rather as along-term increase in the trend growth rate ofproductivity. This was because successivegenerations of MVs were developed, eachcontributing gains over previous generations.

We find it useful to distinguish between an“early Green Revolution” period (1961 to 1980)

and a “late Green Revolution” period from 1981to 2000. Table 1 shows that in the early GreenRevolution, MVs contributed substantially togrowth in Asia and Latin America, but relativelylittle in other areas. For all developing countries,MVs accounted for 21% of the growth in yieldsand about 17% of production growth in the earlyGreen Revolution period. Area expansion ac-counted for about 20% of the increases in pro-duction; the rest came from intensification ofinput use.

The late Green Revolution period (1981 to2000) differed from the early period in severalimportant respects. In part because prices tofarmers were declining, production growth waslower in all regions except Sub-Saharan Africa.The area under food crop cultivation remainedflat overall, with declines in Latin America off-

setting the continued expansion of agriculturallands in Sub-Saharan Africa and the MiddleEast–North Africa region. Yield growth ac-counted for almost all of the increases in foodproduction in developing countries (86%). Fur-thermore, the MV contribution to yield growthwas higher in the late Green Revolution periodthan in early Green Revolution period, account-ing for almost 50% of yield growth and 40% ofproduction growth for all developing countries.This indicates that in the late Green Revolutionperiod, production gains were more dependenton MVs than in the early period, and that MVcontributions were greater in the late period.

Although input use intensified in the lateGreen Revolution period, productivity gainsfrom MVs allowed food production to in-crease dramatically with only modest increas-es in area planted to food crops—and withrelatively slow growth in the use of inputssuch as fertilizer and irrigation.

The Sub-Saharan African region is unusu-al in both periods. Yield growth made onlyminor contributions to production growth inboth periods, and the MV contributions toyield growth were also low—although con-siderably higher in the more recent period.Production growth was based almost entirelyon extending the area under cultivation. Inshort, this region achieved a very partial andincomplete Green Revolution, with a numberof countries realizing virtually no MV contri-butions to food production growth.

The limited scope of the Green Revolu-tion in Sub-Saharan Africa was in part due tothe mix of crops grown in the region (whereroot crops and tropical maize are dominantfood crops) and in part due to the agroeco-logical complexities of the region and asso-ciated difficulties in producing suitable MVs.The table shows that such yield growth aswas realized in Sub-Saharan Africa was al-most entirely contributed by MVs, with littlecontribution from fertilizers and other inputs.

Why did Sub-Saharan Africa get so littlegrowth from varietal improvement until the1990s? The inherited state of knowledge andthe preexisting stocks of improved germplasmwere important factors in differential regionalperformance. Clearly, institutional and politicalfailures also mattered. But Fig. 1 and the un-derlying data suggest that some of Sub-SaharanAfrica’s low growth reflected the lack of usableMV technology until the 1980s and 1990s.Recent evidence is more promising, however.Varietal improvement appears finally to bemaking an impact in Sub-Saharan Africa inrice, maize, cassava, and other crops (7, 10, 11).

More generally, the differences in productiv-ity impacts across regions reflect dramatic un-derlying disparities in the availability and impactof suitable MVs across different agroecologicalzones. The largest initial impacts (in wheat andrice) were in irrigated areas and in rainfed low-lands with good water control. But outside of

Table 1. Growth rates of food production, area, yield, and yield components, by region and period.Data on food crop production and area harvested are taken from FAOSTAT data, revised 2003(http://apps.fao.org/page/collections?subset�agriculture), on total cereals, total roots and tubers,and total pulses. Asia consists of “Developing Asia” excluding the countries of the “Near East inAsia.” Africa consists of “Developing Africa” excluding the countries of the “Near East in Africa” andthe countries of “North-West Africa.” The Middle East–North Africa consists of “Near East in Africa,”“Near East in Asia,” and “North-West Africa.” Latin America includes Latin America and theCaribbean. Crop production is aggregated for each region using area weights from 1981. Estimatesof production increases due to MVs are from (4). Growth rates of other inputs are taken as aresidual. Growth rates are compound and are computed by regressing log time series data on aconstant and log trend variable. The totals for “All Developing Countries” are derived by weightingthe regional figures by 1981 area shares.

Early GreenRevolution

Late GreenRevolution

1961 to 1980 1981 to 2000

Latin America

Production 3.083 1.631Area 1.473 -0.512Yield 1.587 2.154

MV contributions to yield 0.463 0.772Other input per hectare 1.124 1.382

Asia

Production 3.649 2.107Area 0.513 0.020Yield 3.120 2.087

MV contributions to yield 0.682 0.968Other input per hectare 2.439 1.119

Middle East–North Africa

Production 2.529 2.121Area 0.953 0.607Yield 1.561 1.505

MV contributions to yield 0.173 0.783Other input per hectare 1.389 0.722

Sub-Saharan Africa

Production 1.697 3.189Area 0.524 2.818Yield 1.166 0.361

MV contributions to yield 0.097 0.471Other input per hectare 1.069 �0.110

All developing countries

Production 3.200 2.192Area 0.683 0.386Yield 2.502 1.805

MV contributions to yield 0.523 0.857Other input per hectare 1.979 0.948

R E V I E W

2 MAY 2003 VOL 300 SCIENCE www.sciencemag.org760

on

Janu

ary

26, 2

010

ww

w.s

cien

cem

ag.o

rgD

ownl

oade

d fr

om

these environments, varietal improvement wasslower and more limited. This was not for lack ofeffort: IRRI and CIMMYT, along with manynational programs, sought to adapt rice, wheat,and maize MVs to “marginal” environments.And newer IARCs—such as the InternationalCenter for Agriculture in the Dry Areas(ICARDA), the International Center for Re-search in the Semi-Arid Tropics (ICRISAT), theInternational Institute for Tropical Agriculture(IITA), and others—were specifically directed tomarginal environments. But this research tooktime to yield dividends, and the diffusion ofMVs into less favorable agroecologies was slow.Ultimately, however, the effort to broaden theGreen Revolution has been successful, as shownin Fig. 2 and Table 1.

Welfare Effects and CounterfactualScenariosThe SPIA study attempted to consider whatwould have happened had international re-search not taken place between 1960 and 2000.The analysis was conducted with an interna-tional multimarket model developed by the In-ternational Food Policy Research Institute (theIFPRI/IMPACT model) (15, 16).

The counterfactual scenarios consideredwere the following:

1) “No Green Revolution” (NGR): Howwould the food and agricultural situation in2000 have differed if poor countries hadfailed to achieve any productivity gains fromcrop breeding over the period 1965 to 2000?In this scenario, it was assumed that richcountries would have continued to achieveproductivity gains, but the developing worldwould have been constrained to use the sametechnologies available in 1965.

2) “No IARC” (NIARC): How would thefood and agricultural situation in 1999 havediffered in the absence of internationallyfunded research, assuming that national pro-grams would have responded by increasingtheir efforts to some degree? The NIARCscenario is intermediate between the NGRcase and the actual experience.

Both cases feature “high” and “low” scenar-ios. They are compared to a base case, whichincorporates productivity growth components forcrops and countries on the basis of actual expe-rience (Table 2) (16). The simulations indicatethat without international research in developingcountries, crop yields (for all study crops) wouldhave been higher in developed countries by 2.4 to4.8%—primarily because lower production inthe developing world would have driven upprices and given farmers in rich countries anincentive to intensify production. Crop yieldsin developing countries would have been 19.5to 23.5% lower—with price effects again play-ing a mitigating role. The model indicates thatequilibrium prices for all crops combinedwould have been from 35 to 66% higher in2000 than they actually were. Because real

grain prices actually fell by 40% from 1965 to2000, this means that prices would have re-mained constant or risen modestly in the ab-sence of international research (16 ).

Higher world prices would have contributedto an expansion of area planted to crops in allcountries, with attendant environmental conse-quences. Taking area and yield effects together,crop production would have been from 4.4 to6.9% higher in developed countries and 13.9 to18.6% lower in developing countries. The worldwould not necessarily have experienced a cata-strophic “food crisis”—as reflected in worldprices—in the absence of international research;developing countries would have increased theirfood imports by 27 to 30%, partly offsetting theirproduction decreases.

The model does indicate, however, that inthe absence of international research, theworld would have experienced a “humanwelfare” crisis. Caloric intake per capita inthe developing world would have been 13.3to 14.4% lower, and the proportion of chil-dren malnourished would have been from 6.1to 7.9% higher. Put in perspective, this sug-gests that the Green Revolution succeeded inraising the health status of 32 to 42 millionpreschool children. Infant and child mortalitywould have been considerably higher in de-veloping countries as well (16 ).

The simulations for the “No IARC” sce-narios show that aggressive NARS researchprograms would have succeeded in producinga “muted” Green Revolution. As a roughgeneralization, this “lite” Green Revolutionwould have been about 60% of the magnitudeof the one actually achieved.

The SPIA study included benefit-cost cal-culations for IARC and NARS programs.These calculations showed very high benefit-cost ratios for IARC programs and for mostNARS programs (4, 17 ).

Summary: Evaluating InternationalResearch

The comprehensive picture that emergesfrom the SPIA study supports a nuanced viewof internationally funded agricultural re-search. On the positive side, it is clear thatproductivity growth associated with MVs hadimportant consequences. Increased food pro-duction has contributed to lower food pricesglobally. Average caloric intake has risen as aresult of lower food prices—with correspond-ing gains in health and life expectancy.

Critics of further investment in research havenoted that grain prices are at or near historiclows, and they question the need for furtherimprovements in technology. They have alsoraised concerns about the sustainability of inten-sive cultivation—e.g., the environmental conse-quences of soil degradation, chemical pollution,aquifer depletion, and soil salinity—and aboutdifferential socioeconomic impacts of new tech-nologies (18–21). These are valid criticisms. Butit is unclear what alternative scenario wouldhave allowed developing countries to meet, withlower environmental impact, the human needsposed by the massive population expansion ofthe 20th century. Nor is it true that chemical-intensive technologies were thrust upon thefarmers of the developing world. Both IARCand NARS breeding programs attempted to de-velop MVs that were less dependent on pur-chased inputs, and considerable effort has beendevoted to research on farming systems, agro-nomic practices, integrated pest management,and other “environment-friendly” technologies.But ultimately it is farmers who choose whichtechnologies to adopt, and many farmers in de-veloping countries—like those in developedcountries—have found it profitable to use MVswith high responsiveness to chemical fertilizers.

The end result, as shown in Table 2, is thatvirtually all consumers in the world have

Table 2. Counterfactual simulations.

Comparisons to base case(percent changes)

With no MVs indeveloping

countries (NGR)

With no IARCprograms(NIARC)

Crop yieldsDeveloped countries 2.4 to 4.8 1.4 to 2.5Developing countries �19.5 to –23.5 �8.1 to –8.9

Cropped areaDeveloped countries 2.8 to 4.9 1.6 to 1.9Developing countries 2.8 to 4.9 1.6 to 1.9

Crop productionDeveloped countries 4.4 to 6.9 1.0 to 1.7Developing countries �15.9 to �18.6 �6.5 to �7.3

Crop prices, all countries 35 to 66 18 to 21Increase in imports by

developing countries27 to 30 6 to 9

Percent of children malnourished,developing countries

6.1 to 7.9 2.0 to 2.2

Calorie consumption per capita,developing countries

�13.3 to �14.4 �4.5 to �5.0

R E V I E W

www.sciencemag.org SCIENCE VOL 300 2 MAY 2003 761

on

Janu

ary

26, 2

010

ww

w.s

cien

cem

ag.o

rgD

ownl

oade

d fr

om

benefited from lower food prices. Many farmfamilies also benefited from research-drivenproductivity gains—most clearly those whoseproductivity rose more than prices fell, butalso those who produce much of their ownfood. But some farmers and farm workersexperienced real losses from the Green Rev-olution. Those who did not receive the pro-ductivity gains of the Green Revolution(largely because they were located in lessfavorable agroecological zones), but whononetheless experienced price declines, havesuffered actual losses of income. The chal-lenge for the coming decades is to find waysto reach these farmers with improved tech-nologies; for many, future green revolutionshold out the best, and perhaps the only, hopefor an escape from poverty.

Yet the prospects for continued green rev-olutions are mixed. On the one hand, theresearch pipeline for the plant sciences is full.Basic science has generated enormous ad-vances in our understanding of plant growthand morphology, stress tolerance, pathogenresistance, and many other fields of science.This understanding should lead in due courseto improvements in agricultural technologies.But on the other hand, IARCs and NARS arefaced with numerous challenges to theirsurvival. The budgets of many IARCs, notto mention many of their national programcounterparts, have declined sharply in realterms over the past decade. The fundingcrunch reflects a number of factors. Devel-opment agencies, faced with public suspi-cions of new agricultural technologies, andperhaps eager to find shortcuts to develop-ment, have tended to shift funding awayfrom agricultural research and toward otherpriorities. Moreover, life science biotechnol-ogy firms have been eager to claim that pri-vate sector research will take over the func-tions formerly occupied by public sector ag-ricultural research.

But if the past offers guidance for thefuture, a strong public sector role will contin-ue to be needed. In most crops and mostregions of the developing world, private sec-tor agricultural research is not likely to gen-erate large impacts on production or socialwelfare. Continued green revolutions will de-pend on strong programs of national andinternational public sector research. The wel-fare of farmers and farm workers not reachedby the Green Revolution ultimately dependson extending the Green Revolution beyond

present boundaries. The IARCs will have animportant role to play in generating and sus-taining future advances in agricultural tech-nology for the developing world.

References and Notes1. The CGIAR is the organization through which most donor

support for international agricultural research is chan-neled. In addition to IRRI and CIMMYT, the other centersinclude CIAT (International Center for Tropical Agricul-ture), CIFOR (Center for International Forestry Research),CIP (International Potato Center), ICARDA (InternationalCenter for Agricultural Research in the Dry Areas),ICLARM (International Center for Living Aquatic Resourc-es Management), ICRAF (International Center for Re-search in Agroforestry), ICRISAT (International Crops Re-search Institute for the Semi-Arid Tropics), IFPRI (Interna-tional Food Policy Research Institute), IITA (InternationalInstitute of Tropical Agriculture), ILRI (International Live-stock Research Institute), IPGRI (International Plant Ge-netic Resources Institute), ISNAR (International Servicefor National Agricultural Research), IWMI (InternationalWater Management Institute), and WARDA (West Afri-can Rice Development Association). Some unaffiliatedcenters (e.g., the International Centre of Insect Physiologyand Ecology; International Fertilizer Development Center)have also engaged in related research.

2. R. E. Evenson, D. Gollin, Eds., Crop Variety Improve-ment and Its Effect on Productivity: The Impact ofInternational Agricultural Research (CAB Internation-al, Wallingford, UK, 2003).

3. The crops are rice, wheat, maize, sorghum, pearlmillet, barley, beans, lentils, groundnuts, potatoes,and cassava. The study did not address the impact ofresearch done by CGIAR centers on livestock, fisher-ies, or forestry, nor on broad-based management,policy, or resource issues.

4. R. E. Evenson, in Crop Variety Improvement and ItsEffect on Productivity: The Impact of InternationalAgricultural Research, R. E. Evenson, D. Gollin, Eds.(CAB International, Wallingford, UK, 2003), chap. 21.

5. D. Dalrymple, Development and Spread of High Yield-ing Wheat Varieties in Developing Countries (Bureaufor Science and Technology, Agency for InternationalDevelopment, Washington, DC, 1986).

6. D. Dalrymple, Development and Spread of High Yield-ing Rice Varieties in Developing Countries (Bureau forScience and Technology, Agency for InternationalDevelopment, Washington, DC, 1986).

7. N. L. Johnson, V. M. Manyong, in Crop Variety Im-provement and Its Effect on Productivity: The Impactof International Agricultural Research, R. E. Evenson,D. Gollin, Eds. (CAB International, Wallingford, UK,2003), chap. 16.

8. N. L. Johnson, D. Pachico, in Crop Variety Improve-ment and Its Effect on Productivity: The Impact ofInternational Agricultural Research, R. E. Evenson, D.Gollin, Eds. (CAB International, Wallingford, UK,2003), chap. 12.

9. Our measure is of varieties released through aformal process. This is admittedly a weak measure,because varieties could be released without beingadopted by farmers, but in practice such “phantomreleases” appear to be rare and do not occur in anysystematic way.

10. V. M. Manyong et al., in Crop Variety Improvement andIts Effect on Productivity: The Impact of InternationalAgricultural Research, R. E. Evenson, D. Gollin, Eds. (CABInternational, Wallingford, UK, 2003), chap. 8.

11. T. J. Dalton, R.G. Guei, in Crop Variety Improvement andIts Effect on Productivity: The Impact of International

Agricultural Research, R. E. Evenson, D. Gollin, Eds. (CABInternational, Wallingford, UK, 2003), chap. 6.

12. J. W. McKinsey, R. E. Evenson, in Crop Variety Im-provement and Its Effect on Productivity: The Impactof International Agricultural Research, R. E. Evenson,D. Gollin, Eds. (CAB International, Wallingford, UK,2003), chap. 19.

13. By comparison, IARC programs represented fewerthan 3% of the agricultural scientists and less than5% of agricultural research expenditure in developingcountries.

14. The study included estimates for each crop of theproductivity advantages of converting crop acreagefrom traditional varieties to MVs. In some cases,estimates of productivity advantages of convertingfrom early-generation MVs to later generation MVswere also reported. The project also included threecountry studies for India, China, and Brazil. All ap-proaches to measuring productivity gains reportedsimilar estimates of impact.

15. The IFPRI/IMPACT model includes 18 agriculturalcommodities and covers 37 countries or countrygroups. The model solves for an economic equilibri-um that allows researchers to see how crop yields,crop area, crop production, crop trade, and interna-tional prices would change under different scenarios.This model also calculates two welfare indexes asso-ciated with this equilibrium: the proportion of chil-dren (0 to 6%) malnourished, and average food ca-loric consumption. It is described more fully at www.ifpri.org/themes/impact.

16. R. E. Evenson, M. Rosegrant, in Crop Variety Improve-ment and Its Effect on Productivity: The Impact ofInternational Agricultural Research, R. E. Evenson, D.Gollin, Eds. (CAB International, Wallingford, UK,2003), chap. 23.

17. The lowest benefit-cost ratios were realized in Sub-Saharan Africa.

18. K. B. Griffin, Political Economy of Agrarian Change: AnEssay on the Green Revolution (Harvard Univ. Press,Cambridge, MA, 1974).

19. V. Shiva, Monocultures of the Mind: Perspectives onBiodiversity and Biotechnology ( Zed, Atlantic High-lands, NJ, 1993).

20. C. Hewett de Alcantara, Modernizing Mexican Agri-culture: Socioeconomic Implications of TechnologicalChange, 1940–1970 (United Nations Research Insti-tute for Social Development, Geneva, 1976).

21. See also the Web sites of activist groups such as FoodFirst (www.foodfirst.org) and the Turning PointProject (www.turnpoint.org).

22. This paper summarizes major findings from a studyconducted by researchers from a number of institu-tions and disciplines. The study was commissioned bythe Standing Project on Impact Assessment of theTechnical Advisory Committee of the ConsultativeGroup on International Agricultural Research(CGIAR). We particularly acknowledge the followingindividuals who are authors of component chaptersof the study: F. Afonso de Almeida, A. Aw Hassan,M. C. S. Bantilan, Y. P. Bi, V. Cabanilla, E. Cabrera,S. Ceccarelli, T. Dalton, S. De Silva, U. K. Deb, A. F.Dias Avila, A. G. O. Dixon, H. J. Dubin, W. Erskine, P. C.Gaur, R. Gerpacio, E. Grande, S. Grando, R. G. Guei,P. W. Heisey, M. Hossain, R. Hu, J. Huang, N. Johnson,S. Jin, G. S. Khush, P. Kumar, M. A. Lantican, J. H. Li,V. M. Manyong, J. W. McKinsey Jr., A. McLaren,M. Mekuria, M. Morris, S. N. Nigam, D. Pachico,M. Rosegrant, S. Rozelle, A. Sarker, K. Shideed,R. Tutwiler, T. Walker, and C. S. Wortmann. Inaddition, we acknowledge specific comments on adraft of this article from D. Dalrymple, C. Doss,D. Duvick, G. Gryseels, G. Manners, and M. Morris.

R E V I E W

2 MAY 2003 VOL 300 SCIENCE www.sciencemag.org762

on

Janu

ary

26, 2

010

ww

w.s

cien

cem

ag.o

rgD

ownl

oade

d fr

om