88490 JUNE 2014 • Number 145 Public Research Organizations and Agricultural Development in Brazil: How Did Embrapa Get It Right? Paulo Correa and Cristiane Schmidt One of the most extraordinary events in Brazil in the past 30 years has been the country’s “agricultural revolution.” In the 1970s, food scarcity was a concrete risk in a country experiencing rapid urbanization and middle class expansion. Food scarcity concerns reemerged during the following decade when short-lived spikes in real wages temporarily increased households’ demand for those goods. One of the government’s initiatives to address Brazil’s stagnant agriculture sector and food scarcity was Embrapa (Empresa Brasileira de Pesquisa Agropecuária). Embrapa has succeeded in adapting, creating, and transferring technologies to Brazilian farmers for the past 30 years, helping transform Brazil into one of the world’s largest food exporters. How did Embrapa get it right when similar organizations failed? Embrapa was created in 1973 as an agricultural research orga- • The development of cottonseeds adapted to the semihu- nization under Brazil’s Ministry of Agriculture and was al- mid tropical conditions made it possible to obtain much most entirely funded by government resources. Pursuing a higher yields per hectare (which tripled between 1983 clear vision of recuperating and boosting the agricultural sec- and 2010) and a fiber quality equivalent to that of the tor, Embrapa has developed and transferred more than 9,000 imported product. This represented an important turn- technologies to Brazilian farmers. Researchers working at Em- around for this sector that employed large contingents of brapa have created over 350 cultivars1 and obtained more unskilled labor and had suffered from low productivity, than 200 international patents. Embrapa’s key contributions plant disease, and international competition. to agricultural development include: By the late 2000s, the Brazilian agricultural sector not • “Agricultural liming” techniques that turned the acidic only met the country’s consumption needs amid robust pov- soil of the Brazilian Cerrado into arable land by neutral- erty reduction, it was also leading global markets in agricul- izing the soil’s pH levels; the transformation of the Cerra- ture goods and livestock (such as coffee, sugar, orange juice, do—a biome covering about 22 percent of Brazil’s sur- beef, and poultry). According to some researchers, agricul- face area—kept the price of marginal land low and the ture development in Brazil also had large impacts in terms of expansion of agriculture at internationally competitive social inclusion (Bonelli 2002). prices possible. The modernization of Brazilian agriculture resulted • Cross-breeding techniques that led to the development from the convergence of many factors. For example, efficient of soybean varieties more tolerant to the Cerrado’s acidic producers in the Cerrado tend to be large farms that can seize soils and with a lifecycle up to 12 weeks shorter than that scale economies and are well integrated, vertically and hori- of the typical plant, enabling two harvests per year. zontally, in the market.2 Well-functioning markets for arable 1 POVERTY REDUCTION AND ECONOMIC MANAGEMENT (PREM) NETWORK    www.worldbank.org/economicpremise land enabled the emergence of large agricultural units and the Furthermore, Embrapa strengthened its international achievement of economies of scale at production level. Trade links by establishing “virtual labs abroad” on three conti- liberalization—which substantially reduced the prices of agri- nents to institutionalize knowledge generation and ex- cultural inputs (herbicides, pesticides, insecticides, fertiliz- change. ers, tractors, and others) is another factor contributing to the (iv) A mission orientation and IPR policy. Embrapa was created sector’s rapid modernization. Created as part of the govern- with “the mission to provide feasible solutions for the de- ment response to food scarcity and growing macroeconomic velopment of Brazilian agribusiness through knowledge imbalances due to growing imports of agricultural goods, the and technology generation and transfer.” Curiosity-driv- Brazilian agricultural research company Embrapa is arguably en research was discouraged by using agricultural re- a major factor contributing to the systematic increases in Bra- search as a means to solve concrete problems faced by the zil’s agricultural productivity. 3 Brazilian farming sector. Pursuing an open innovation Agricultural productivity gains in Brazil in the 1980s system and IPR policy in the agricultural sector facilitat- and 1990s were closely related to improved tropical agricul- ed technology transfer, diffusion of new cultivars, and ture knowledge and its effective use by local farmers. Embra- the filing of international patents. An IPR policy that fa- pa’s new technologies built on these developments in at least vored social well-being rather than benefiting just corpo- two ways: first, they enabled the expansion of agriculture and rations allowed new technology to be distributed at pro- cattle ranching activities into Brazil’s Cerrado, one of the larg- duction costs only. est reserves of arable land in the world. This process helped These factors may be used as guiding principles for the keep the cost of the marginal land down and the growth of design of public organizations providing agricultural research Brazilian agricultural production internationally competi- and extension services in developing countries. The adequacy tive. Second, the development of seeds that were more suit- of the Embrapa model, however, will depend upon a number able for tropical climate conditions (and the Cerrado’s soil) of country-specific factors, including the country’s develop- helped increase land productivity for a number of crops, espe- ment level (and its corresponding technological needs) and cially those originally grown in temperate climate regions. the size of its economy, which ultimately defines to what ex- Embrapa’s successful experience is at odds with the per- tent such organization is affordable, among other issues (dis- formances of many other public research institutes in devel- cussed in detail in conclusion of this note). It would be incor- oping countries, which often struggle to generate high-quality rect, therefore, to assume that replicating Embrapa’s research and effectively transfer technology to farmers. It is experience in a different context would necessarily lead to also at odds with the expected performance of organizations similar results. This note also presents three examples of how that share Embrapa’s general governance structure: a publicly Embrapa’s technologies and techniques dramatically affected owned company under the Ministry of Agriculture and essen- the Brazilian agriculture, and discusses how certain institu- tially funded by public resources. In most cases, this gover- tional characteristics enabled Embrapa to succeed while oth- nance structure ensures failure. Why did Embrapa succeed er similar organizations did not. where other research organizations failed? Is its experience Technology and Productivity in replicable? Brazil’s Agriculture Briefly, this note argues that Embrapa’s success is due to four main factors: This section presents three case studies. The first highlights (i) Adequate levels of public funding. Embrapa’s expendi- the transformation of the Cerrado’s soil, which opened up a tures in the last 20 years, at around 1 percent of Brazil’s vast amount of comparatively cheap, idle, but potentially ara- agricultural gross domestic product (GDP), compare ble land, helping to competitively produce agricultural (in- well with figures of public spending on agricultural re- cluding livestock) exports for international markets. The oth- search and development (R&D) in more developed coun- er two cases outline the introduction of new soybean and tries, such as Canada, the United States, and Australia cotton seeds that were adapted to local characteristics of the (1.2, 1.4, and 0.8 percent, respectively, for 2006–9). soil and climate and which raised the land productivity for (ii) Sustained investment in human capital. Twenty percent of those crops. Although these cases offer insight for other coun- Embrapa’s budget was invested in the education and tries, it would be simplistic to assume that replicating Em- training of its employees between 1974 and 1982 alone. brapa’s experience in a different context will have the same Currently, three-fourths of Embrapa’s 2,000 researchers result. hold a PhD. Turning the Cerrado into an arable land (iii) International collaboration and research excellence. From Brazil’s Cerrado covers an area of 2 million square kilometers the beginning, researchers were drawn from leading uni- in the central region of the country, equivalent to approxi- versities, setting a high standard of research excellence. mately 22 percent of the country’s total area. It is the second 2 POVERTY REDUCTION AND ECONOMIC MANAGEMENT (PREM) NETWORK    www.worldbank.org/economicpremise largest biome in Brazil, surpassed only by the Amazon (figure national competitiveness of Brazilian beef exports.6 Other 1). Until the early 1970s, the area was used mainly for low- important technical solutions promoted by Embrapa in- productivity activities, such as extensive cattle ranching. Dur- clude promoting soil recuperation, minimum tillage, and ing this period, Brazilian commercial agriculture, particularly crop and cattle integration. grain crops, was located primarily in the states of Rio Grande By increasing the fertility of the Cerrado, Embrapa do Sul, Parana, and São Paulo. By the late 1970s, however, a opened up for cultivation one of the largest reserves of arable rise in land prices in southern states made the cultivation of land in the world. This, in turn, helped keep the price of mar- land-intensive crops such as grains increasingly unviable from ginal land down and the cost of Brazilian agricultural exports a commercial standpoint.4 Expanding agriculture toward the competitive in international markets. The U.S. Midwest and Cerrado was a natural option for southern farmers, given the Argentina still possess the most productive land for temper- availability, land prices, and its overall climate.5 The problem, ate climate crops (such as soybeans). But, as the availability of however, was the soil, which was extremely poor in nutrients land for agricultural expansion declined in those countries, and high in acidity, both of which made it unfit for commer- Brazil’s Cerrado became one of the most productive marginal cial agriculture. lands worldwide (Rezende 2002). Not surprisingly, about 20 To reduce the soil’s toxicity, Embrapa employed a tech- percent of the land in the Cerrado is owned by foreign inves- nique called agricultural liming, a process in which indus- tors. Figure 2 shows the spectacular increase of the share of trial quantities of lime are poured onto the soil to reduce the Center-West Region in national agricultural production, acidity levels. In 1990, between 14 million and 16 million going from 10 percent to between 60 and 70 percent in just a metric tons of lime were spread on Brazilian fields. Embra- decade. pa also developed varieties of rhizobium, a bacterium that Although new techniques introduced by Embrapa helps fix nitrogen in legumes (such as soy), specifically played a major role in the success of agribusiness in that re- adapted to the Cerrado soil, thereby reducing the need for gion, other factors also had significant impacts, including (i) fertilizers. Embrapa cross-bred an African grass called bra- good geographic conditions (a topography suited to mechani- chiaria with a native Cerrado grass to engineer a variety zation and rainfall patterns suitable for summer crops); (ii) that produced 20–25 tons of grass feed per hectare, many improvements in the transportation infrastructure; and (iii) times the native grass yield and three times the yield in Af- tax incentives.7 rica. This allowed parts of the Cerrado to be transformed Biannual harvest of soybeans into high-yielding pasture, helping reduce the average time The soybean plant is native to the more temperate regions of needed to raise an animal for slaughter from four years to northeast Asia (Japan, the Korean peninsula, and northeast 18–20 months, expanding Brazil’s beef herd and the inter- China). It is a short-day (long-night) crop that is naturally bet- ter suited to grow in latitudes above 30 degrees. In Brazil, the Figure 1. Brazil's Cerrado production of soybeans in the South Region, where climate conditions are similar to those in temperate regions, picked up in the late 1960s. In the decades that followed, the soy- bean subsector established itself as a core activity of Brazilian agribusiness. Increases in cultivated area (from 1.3 million to 8.8 million ha) and incremental productivity gains (from 1.14 to 1.73 tons/ha) caused output volumes to grow tenfold Figure 2. Regional Distribution of Agricultural Production 80 share of production (%) North South-East 70 North-East South 60 Center-West 50 40 30 20 10 0 1976/77 1980/81 1984/85 1988/89 1992/93 1996/97 2000/1 2004/5 2008/9 Source: IBAMA (Instituto Brasileiro do Meio Ambiente e dos Recursos Naturais Renováveis) Source: Compania Nacional de Abstecimento (CONAB). 3 POVERTY REDUCTION AND ECONOMIC MANAGEMENT (PREM) NETWORK    www.worldbank.org/economicpremise (from 1.5 million to more than 15 million tons) between innovation in recent times for the genetic improvement of 1970 and 1980. In the 1970s, soybean production was con- soybeans in Brazil.8 The second crop significantly expanded centrated in the southern states. While cheaper farmland land productivity in the soybean industry, reaching an aver- made the Center-West Region attractive, the predominantly age of about 2.75 tons/ha in 2006–10 (figure 3). By 2013, tropical climate of the region limited the expansion of soy- more than 70 percent of total soybean production was gener- bean farming in the Cerrado area. ated in the Cerrado. As a first step, and in addition to introducing general Revival of the cotton industry techniques to reduce soil toxicity such as the agricultural lim- Until the mid-1980s, the Brazilian cotton crop, concentrat- ing, Embrapa promoted the development of cultivars with ed in the South and South-East regions, was characterized by better agronomic adaptation to the tropical climate and more low productivity. At that time, the emerging boll weevil tolerance for the Cerrado’s acidic soil. Cross-breeding tech- plague hit the sector hard. Moreover, beginning in 1990, niques eventually led to the development of soybean varieties trade liberalization significantly increased import competi- with a lifecycle 8–12 weeks shorter than that of the typical tion. These two events combined with devastating effects: plant. The shortened lifecycle enabled two harvests per year. cotton production dropped from nearly 1 million tons in Embrapa also developed cultivars more resistant to diseases 1981 to 420,000 tons in 1992, while cultivated areas di- like frogeye, stem canker, mildew, and red root rot, reducing minished from 4.1 million ha in 1981 to 1.3 million ha in crop losses as well as expenditures on insecticides and contrib- 1995. The cotton crisis had both economic and social im- uting to higher yields. The inoculation of soybean seeds with pacts, given the importance of the sector for the employ- nitrogen-fixing bacteria has almost eliminated the need for ni- ment of low-skilled workers. In response, Embrapa started to trogen fertilizers, leading to savings of R$7.5 billion per year, work on the development and adaptation of cottonseeds in according to Embrapa’s estimations (The Economist 2010). the state of Mato Grosso, the heart of Brazil’s Cerrado. The development of short-photoperiod soybeans and The Cerrado’s climate conditions are generally favorable the development of nitrogen-fixing rhizobia were game for growing cotton. Temperatures in Mato Grosso remain in a changers. The discovery of the long juvenile period, which narrow band throughout the year, from 73–82° F (23–28° delays flowering under short-day conditions, was the greatest C). The result is a long growing season, up to 210 days, de- Figure 3. Soybean Production, Productivity, and Acreage 90 3.5 80 3.0 70 2.5 60 50 2.0 40 1.5 30 1.0 20 0.5 10 0 0 91 93 95 97 99 01 03 05 07 09 11 13 19 19 19 19 19 20 20 20 20 20 20 20 productivity (kg/ha, in thousands, left axis) production (in thousands of tons, right axis) area (in millions of ha, right axis) Source: Compania Nacional de Abstecimento (CONAB). 4 POVERTY REDUCTION AND ECONOMIC MANAGEMENT (PREM) NETWORK    www.worldbank.org/economicpremise pending on the timing of monsoonal rains. From October to Most cotton production is currently outside the South and March, monthly precipitation ranges from four to eight inch- South-East regions, in Mato Grosso and Bahia, which lie es, tapering off to virtually zero in July, the peak harvest within the vast area of the Cerrado. These two states ac- month. The combination of regular rainfall throughout the count for more than 90 percent of Brazil’s total cotton pro- growing season and sandy, well-drained soils means that cot- duction (ICAC 2011). ton yields in the Cerrado surpass irrigated yields in many Factors Contributing to Embrapa’s parts of the world. The virtual absence of rain during harvest Performance: Some Hypotheses minimizes crop damage, and the well-drained soils mean that fieldwork is seldom impeded by rainfall. One drawback of the This section explores factors contributing to Embrapa’s ca- climate is the lack of a cold period to help kill off harvest pests. pacity to effectively develop and transfer new technologies to As a result, insecticide expenditures per hectare are among the Brazilian farming sector to help increase agricultural pro- the world’s highest. ductivity. Research productivity (the generation of excellent The development of the CNPA ITA 90 seed increased research within a reasonable time and at reasonable costs) and cotton yields and made the fiber quality equivalent to that market relevance are common challenges faced by public re- of the imported product, beginning the expansion of cot- search organizations like Embrapa. Efficient technology trans- ton cultivation in the Cerrado. The first experiments in fer—that is, technology transfer accomplished within a short Chapadão do Parecis were expanded to southern Mato period of time and at costs affordable to the producers—is an- Grosso and received the support of the Mato Grosso Foun- other. How did Embrapa address those issues and get it right? dation, which went on to encourage the adoption of cotton Four key factors have contributed to Embrapa’s performance: across the Cerrado in the state. Research was funded by (i) adequate public funding; (ii) sustained investment in hu- programs such as Facual, Fialgo, Fundeagro, Pluma, and re- man capital; (iii) international collaboration and research ex- gional cotton quality improvement programs such as PRO- cellence; and (iv) mission orientation and IPR policy. ALMAT, PROALGO, and PROALBA. Figure 4 shows the Adequate public funding remarkable turnaround in Brazil’s cotton production: by The federal government’s allocation of the necessary funding 2010, cotton output had reached precrisis levels, and al- for a sufficiently long period was a critical factor to success. though the area cultivated remained essentially unchanged, Brazil’s federal government funds up to 95 percent of Em- the land yields of 1,500 kg/ha were three times greater. In brapa’s total budget.9 Figure 5a shows the sustained and in- fact, the productivity of the Brazilian cotton sector became creasing investment made by the Brazilian government in the third highest in the world after Israel and Australia. Embrapa between 1974 and 2008. Particularly striking is the Figure 4. Output, Area, and Yield of Brazil's Cotton Industry, 1976–2011 4.5 2.5 area (millions of ha, left axis) 4.0 yields (thousands of kg/ha, right axis) production (millions of tons, right axis) 3.5 2.0 3.0 1.5 2.5 2.0 1.0 1.5 1.0 .5 .5 0 0 2000/1 2002/3 2004/5 2006/7 2008/9 1976/77 1978/79 1980/81 1982/83 1984/85 1986/87 1988/89 1990/91 1992/93 1994/95 1996/97 1998/99 2010/11 Source: Compania Nacional de Abstecimento (CONAB). 5 POVERTY REDUCTION AND ECONOMIC MANAGEMENT (PREM) NETWORK    www.worldbank.org/economicpremise Figure 5. Brazil's Budget a. In real terms b. Share of agricultural GDP (values adjusted by IGP-DI) 1,800,000 1.6 1,600,000 1.4 Embrapa expenditures as share 1,400,000 1.2 1,200,000 of agriculture GDP 1.0 1,000,000 0.8 800,000 600,000 0.6 400,000 0.4 GDP share 200,000 share = 1 0.2 0 0.0 1974 1978 1982 1986 1990 1994 1998 2002 2006 **2008 1974 1978 1982 1986 1990 1994 1998 2002 2006 2010 Source: Embrapa. rapid growth of Embrapa’s expenditures in the nascent stage percent in the 1990s, reaching a peak of 1.4 percent in the mid (1974–82), when important initial investments were need- 1990s. Embrapa’s expenditures since 1990, at around 1 per- ed. A decline occurred from 1997 to 2002, greatly influenced cent of agricultural GDP, compare well with figures of public by the macroeconomic adjustments of the Real Plan—the eco- spending on agricultural R&D in more developed countries, nomic stabilization program introduced in Brazil in July such as Canada, the United States, and Australia (1.2 percent, 1994 that was successful in bringing inflation down from the 1.4 percent, and 0.8 percent, respectively, for 2006–9). extremely high levels that had prevailed.10 Embrapa’s budget Sustained investment in human capital cuts were much less severe than those imposed on other pub- Embrapa’s human resources policy of aggressively developing lic organizations, however, in fact, in a period of macroeco- the capacity of its researchers is another reason for its success. nomic imbalances and tight fiscal policies, it is revealing that Figure 6a highlights the gradual increase in the number of Embrapa’s budget did not falter much. Spending resumed a professionals with postgraduate qualifications, initially at the growing trend in 2003, and the 2008 budget was one of the master’s degree and then at the PhD level. Investments in four biggest during 1974–2008. training between 1974 and 1982 accounted for approximate- Figure 5b illustrates the evolution of Embrapa’s budget as ly 20 percent of the budget, totaling US$214.6 million fund- a share of agriculture GDP. Expenditures increased from less ed via the federal government or international institutions, than 0.2 percent of the agricultural GDP in 1974 to about 1 including the World Bank, the Inter-American Development Figure 6. Increase in Highly Qualified Researchers a. Researchers' qualifications b. Employment: researchers, support staff, and all employees 2,000 BS 12,000 researchers 1,800 MS support PhD all employees 1,600 10,000 1,400 8,000 1,200 1,000 6,000 800 600 4,000 400 200 2,000 0 1973 0 1977 1981 1985 1989 1992 1996 2000 2004 2008 1973 1977 1981 1985 1989 1992 1996 2000 2004 2008 Source: Embrapa. 6 POVERTY REDUCTION AND ECONOMIC MANAGEMENT (PREM) NETWORK    www.worldbank.org/economicpremise Bank, and the Food and Agricultural Organization of the ganization had become a leader in south–south cooperation United Nations (see Alves [1984]). Embrapa currently em- and conducted technology transfers with markets in Africa ploys over 9,200 people, including more than 2,000 research- (including Ghana, Senegal, Mozambique, and Mali) and the ers, three-fourths of whom have doctoral degrees.11 These in- Americas (including the República Bolivariana de Venezuela, vestments in human capital have helped Embrapa accumulate Ecuador, Colombia, and Panama). Embrapa also created Vir- a critical mass of scientists who have formed a robust research tual Labs Abroad (Labex) as another step toward increasing network with diverse talents. international collaboration. The United States (USDA-ARS) Figure 6b illustrates the steady increase in the number of was the first partner in this project. Subsequently, Labex Eu- employees and the proportion of highly qualified researchers rope was created in Montpellier, France, and expanded into at Embrapa. It shows the decline in the number of nonre- other countries including the Netherlands and the United search personnel and the stability of the number of research- Kingdom. ers in total employment. Embrapa prioritized the retention of Mission orientation and IPR policy research talent by promoting a meritocratic culture, particu- From the outset, Embrapa’s organizational focus has been on larly in its recruitment and promotion of researchers. Each improving agricultural productivity through applied research research center had clear objectives outlined for the long and technology transfer to farmers. Its creation as a mission- term, short-term goals, well-defined deadlines, and metrics to oriented institute dedicated to using agricultural research as a measure results. As a company, Embrapa could have a salary means to solve concrete problems faced by the Brazilian farm- structure different from the standard structures of public ad- ing sector helped it avoid the temptation of investing in curi- ministration in Brazil, allowing it to reward results by allocat- osity-driven research and, to some extent, its capture by pure- ing bonuses for high-performance centers. An evaluation sys- ly academic interests (Alves 2010). Mission-orientation led to tem has been in place since 1996 to judge merit, but the the creation of National Product Programs, which were very incorporation of meritocratic advancement procedures has effective at identifying the needs of specific crops and direct- been difficult, in many cases hampered by cronyism and labor ing resources toward the development of technological solu- legislation. tions. International collaboration and research excellence Embrapa was also conceived as a broad network of re- International collaboration, such as the transfer of research search entities, each specializing in a particular topic, with results, the development of joint research projects and train- decentralized control over decision making. The agency’s de- ing, also played a dynamic role in Embrapa’s ability to achieve centralized model of applied research split it among national effective results. Being able to borrow from the international commodity, regional resource, and “thematic” centers, en- pool of available knowledge was probably a major component abling both a national and local focus. Close connection with of Embrapa’s success. the farming sector and a solid feedback system helped focus For example, the breeding program for short-photoperi- activities on the final goal of improving agriculture productiv- od soybeans benefited from collaboration between Embrapa ity (rather than generating academic publications) through and USDA-ARS, including from the access to the germ- technology transfer. As of December 2013, Embrapa had a plasm12 of soybeans. Similarly, the International Center for presence in almost all 27 states in Brazil. It has 38 research Tropical Agriculture (CIAT) provided the germplasm of for- centers, three service centers, and 13 central divisions.15 Its ages such as the bracharia and panicum species. Having access widespread presence facilitates a working relationship be- to this pool of knowledge was vital for subsequent adaptation. tween farmers and the researchers who want to understand Training for researchers in the best foreign universities their needs. Farmers know, for example, that the unit respon- was a priority for Embrapa in the initial years, when local sible for maize research is the National Research Center for universities had limited capacity in the fields of agriculture Maize and Sorghum, located in Sete Lagoas–Minas Gerais. and agricultural research. So Embrapa hired researchers Another benefit from mission orientation was the effi- with degrees from foreign universities and sent other re- ciency through which scientific output was transformed into searchers abroad for training opportunities, including at proof of concepts: prototypes and innovation. This transition leading universities and research organizations in the United is a common challenge in research commercialization. Re- States, Europe, Canada, and Australia, among others.13 This searchers in public research institutes and universities are of- helped build academic relationships around the world and ten reluctant to engage in commercialization efforts that may led to the inclusion of Brazilian researchers in relevant re- reduce the time available for the development of other re- search networks. search. Embrapa’s mission orientation merged these two sep- By 2013, Embrapa had established 78 bilateral agree- arated activities into one continuous task. Prototypes were ments with 56 countries and 89 institutions, comprising re- evaluated on Embrapa’s testing farms or in collaboration with search partnerships as well as technology transfers.14 The or- similar organizations, including from the private sector. In 7 POVERTY REDUCTION AND ECONOMIC MANAGEMENT (PREM) NETWORK    www.worldbank.org/economicpremise fact, Embrapa realized that close cooperation with other insti- Why did Embrapa succeed where other research tutions in research and commercialization was pivotal to suc- organizations failed? cess, and, depending on the complexity of the productivity Embrapa’s mission orientation, focusing from the outset challenge, Embrapa either produced new varieties by itself or on the improvement of agricultural productivity rather entered into production agreements with select private and than the production of scientific work, was a key driver of public partners.16 its success. Integration into the international flow of Lastly, the open innovation system and IPR policy adopt- knowledge increased research efficiency and accelerated ed by Embrapa, coupled with a network of extension services, training. An open IPR policy—and a network of offices enabled an effective diffusion of research results. While Em- spread throughout the country—facilitated the dissemina- brapa has adopted a flexible approach to IPR throughout its tion of Embrapa’s discoveries. Funding was kept at ade- history—with some emphasis on patenting and licensing in quate levels for more than two decades. Investments in hu- recent years—the distribution of improved seeds at minimum man capital were highly prioritized. A meritocratic culture costs followed somewhat naturally from its original mandate, has been actively promoted by the organization. As a re- that is, improving agricultural productivity. In cases of re- sult, research dealt with the practical problems of agricul- search organizations with a research mandate only, one must ture, while technology and innovations sourced through wonder whether some results simply end up idle on labs’ Embrapa were quickly deployed by farmers. By reacting to shelves. Monopoly pricing—the natural strategy of a private market signals and focusing on activities for which demand research company—would, by definition, restrain access to re- was increasing in international markets, Embrapa avoided search results by producers. By placing farmer’s profitability the usual challenges of purely “supply-push” technology at the center of its objective function (to the detriment of its transfer policies. own financial gains), Embrapa maximized economic returns Is this experience replicable? to public investments in R&D—as measured by the wide- Embrapa’s institutional arrangements and policies—opera- spread productivity gains in agriculture, plus spillovers in tional independence, sustained public investment in mission- terms of stronger export performance and geographic decen- oriented research, investment in human capital, and integra- tralization of growth. tion into international flow of knowledge—are certainly replicable across a variety of policy environments. Setting up Conclusions a public research organization with a similar structure will Embrapa grew out of one of a handful of technology policies require policy and continued financial support as well as a fo- that enabled the development of potentially profitable activi- cus on tangible, practical results. The customization aspect of ties in developing economies, similar to the wine industry in Embrapa’s operations, with researchers focusing on technolo- Chile, the exportation of palm oil in Malaysia, and the pro- gies that address specific challenges, requires some form of duction of cut flowers in Kenya (Chandra 2006). There is, decentralization to establish close contacts with farmers. however, one major difference: the scope of agricultural ac- However, there are important caveats to keep in mind: tivities covered by Embrapa is considerably larger. Since its (i) As indicated by the increase in land area used for founding in 1973, Embrapa has created and transferred to crops and livestock preceding the creation of Embra- Brazilian farmers more than 9,000 technologies and built an pa, R&D and technology policy aimed to address bottle- intellectual property portfolio of more than 350 cultivars and necks to the development of existing comparative advan- about 200 international patents. It is currently considered tages in Brazil and avoiding the common challenge of the world’s leading tropical research institute. promoting activities not consistent with the country’s Embrapa generated and transferred new technologies factor endowments. and techniques tailored to Brazil’s climate and soil conditions. (ii) While critical for the enhancement of agricultural pro- The use of these technologies by Brazilian farmers facilitated ductivity, agricultural research and extension services the expansion of Brazilian agriculture and increased exports were part of a broader set of policies and institutional devel- at internationally competitive prices—first, by expanding the opment that positively affected agricultural productivity supply of arable land, and, second, by improving the produc- (from rural credit to trade liberalization). tivity of selected crops. New techniques to improve the quali- — The search for new techniques to improve the Cer- ty of the otherwise inhospitable Cerrado soil opened a vast rado’s soil was in response to the rising price of ara- tract of newly arable land, keeping marginal agricultural costs ble land in the South Region of Brazil in the 1980s. down and enabling an increase in agricultural production, Without a functioning market for arable land, price while improvements in the cultivars of soybeans and cotton signals would not exist. The migration of farmers ultimately yielded biannual harvests. Both activities increased from the south brought preexisting agricultural the productivity of land. knowledge to the Central-West Region, particularly 8 POVERTY REDUCTION AND ECONOMIC MANAGEMENT (PREM) NETWORK    www.worldbank.org/economicpremise with respect to crops such as soybeans, as well as per- Notes haps decades of entrepreneurial experience in agri- 1. Cultivars are new seeds with improved performance due to business. better adaption to specific clime and soil (edaphoclimatic) (iii) Brazil had and has the necessary critical mass and commen- conditions. surate public funding prowess, which permitted invest- 2. In terms of farm size and efficiency, Helfand and Levine ments of the required scale and time horizon. This is not (2004) find a U-shaped curve with decreasing efficiency up to the case for many smaller developing countries (for ex- about 500 ha and then increasing efficiency up to 10,000– ample, in Central America) where other collaborative 20,000 ha. research approaches (for example, the Consultative 3. Rada and Valdes (2012), for example, estimate that each 1 Group on International Agricultural Research [CGIAR] percent increase in Embrapa’s cumulative research invest- model) may be more appropriate. ment raises the productivity of the most efficient farms by 0.2 (iv) Embrapa filled a gap when public investments in science percent. For large farms that were able to seize economies of and technology were essential for the development of a scale, the productivity improvements are estimated to be par- modern commercial agriculture. As agricultural needs ticularly pronounced. change and other players are present 40 years down the 4. For instance, by 1977–79, the price of a hectare of arable road—such as universities and the private sector—the Em- land in the states of Rio Grande do Sul and Parana was about brapa model will need to be revised. three and six times higher, respectively, than in Mato Grosso (v) In this sense, it is important to bear in mind that the fac- (see Rezende [2002]). tors of success identified here refer to a past period 5. The climate conditions are especially suitable for cotton; (roughly 1974–2004), and they do not necessarily re- the steady temperature allows for a longer growing season, the flect the policy and institutional choices of Embrapa at topography is suited to mechanization, and rainfall patterns present. are suitable for summer crops. Acknowledgments 6. The Cerrado had low-yielding varieties of native grass that lowered the productivity of cattle rearing. Embrapa has re- The paper upon which this note was drawn was funded by cently begun experiments to modify brachiaria genetically to the 2013 LAC-Innovation Beam and prepared in May– produce a larger-leafed variety called braquiarão, which prom- June 2013 under the supervision of Esperanza Lasagabaster ises even bigger increases in yields. and Marialisa Motta. The authors wish to thank Braz Henri- 7. Examples of government programs to attract migrants que Nunes Rodrigues, Cláudio Takao Karia, Domingo He- from other regions included the Program Directed Settle- roldo, Rudolfo C. Rinhardt, Eduardo Riedel, Filipe Geraldo ment of Upper Paranaíba (PADAP), the Development Pro- de Moraes Teixeira, Felipe Teixeira, Gervasio Castro de Re- gram of the Cerrado (POLOCENTRO), and the Program for zende, José Renato Faria, Paulo Roberto Galerani, José Ro- Japanese–Brazilian Cooperation for Development of the Cer- berto Rodrigues Peres, and Marcos Esteves for their valuable rado (PRODECER). inputs to the paper. Special gratitude goes to Eliseu Roberto 8. Embrapa has produced the largest worldwide program of de Andrade Alves and José Roberto Mendonça de Barros for soybean improvement in the tropics by developing cultivars their valuable collaboration and insights. The authors are with higher juvenile periods. also grateful to Gregory V. Wolf, Erick Fernandes, Mark 9. International organizations, such as the Food and Agricul- Dutz, Mark Roland Thomas, Otaviano Canuto, Willem ture Organization of the United Nations, the World Bank Janssen, and two anonymous reviewers for the comments to Group, and the Inter-American Development Bank contrib- previous drafts of this paper. Morten Seja provided research uted regularly to the budget. However, since the initiation of assistance. Embrapa, the vast majority of its budget derives from govern- About the Authors ment funding. For example, the PRODETAB project that was cofinanced by the World Bank contributed around 1 percent Paulo Correa is a Lead Economist at the Innovation Technology of the agricultural R&D expenditure in Brazil (see Pardey, and Entrepreneurship Unit of the Finance and Private Sector De- Alston, and Piggott [2006, 276]. velopment Global Practice of the World Bank. Cristiane Alkmin 10. Inflation rates of over 50 percent per month fell to ap- Junqueira Schmidt has a PhD in economics from the EPGE-FG proximately 20 percent annually. (Rio de Janeiro, Brazil) and currently teaches in the MBA pro- 11. In comparison, less than 20 percent of the more than grams of FGV. Schmidt was also a Visiting Scholar at Columbia 2,300 researchers of the main Argentine agricultural research University. branch, INTA, hold a PhD. 9 POVERTY REDUCTION AND ECONOMIC MANAGEMENT (PREM) NETWORK    www.worldbank.org/economicpremise 12. A germplasm is a collection of genetic resources for an Chandra, Vandana. 2006. Technology, Adaptation and Exports: How organism. Some Developing Countries Got It Right. Washington, DC: World 13. After the 1990s, Embrapa started postdoctoral programs Bank. Evenson, R. E., and Denisard Alves. 1998. “Technology, Climate through Brazilian universities, which provided training at an Change, Productivity and Land Use in Brazilian Agriculture.” international standard. Embrapa now develops its own highly Planejamento e politicas publicas 18: 223–60. trained professionals through permanent cooperation with Helfand, Steven, and Edward Levine. 2004. “Farm Size and the universities, research institutes, and overseas research entities. Determinants of Productive Efficiency in the Brazilian Center- 14. For example, in the United States, Embrapa has relation- West.” Agricultural Economics 31 (2–3): 241–49. ships with several major universities and the Agricultural Re- ICAC (International Cotton Advisory Committee). 2011. “The Cost of Raw Cotton Production.” In Brazil’s Cotton Industry search Service of the U.S. Department of Agriculture; in Economic: Reform and Development, ed. J. Kiawu, C. Valdes, and France, with the INRA, CIRAD, and IRD; and, in Japan, with S. MacDonald. USDA. JICA and JIRCAS. Kouadio, Théodore. 2012. “Africa and Brazil to Share Cotton 15. Embrapa has also been coordinating the National Agricul- Know-How.” Summary posted by Meridian. http://www.merid. tural Research System (SNPA), which includes public and org/en/Content/News_Services/Food_Security_and_AgBio- private entities involved in agricultural research in Brazil. tech_News/Articles/2012/Sep/10/Brazil.aspx. Pardey, Philip G., Julian M. Alston, and Roley R. Piggott. 2006. Agri- 16. Brazilian Innovation Law (federal law 10973, of 2004), cultural R&D in the Developing World:Too Little, Too Late? Wash- article 9º. ington, DC: International Food Policy Research Institute. Portal Brasil. 2010. Research Units: Brazilian Agricultural Research References Corporation (Embrapa). http://www.brasil.gov.br/sobre/ Alves, Eliseu Roberto de Andrade. 1984. “Brazil’s Program for science-and-technology/research-units/brazilian-agricultural- Development of Agricultural Researchers.” Brazilian Agriculture research-corporation-embrapa/br_model1?set_language=en. and Agricultural Research 9 (7): 161–73. Rada, Nicholas, and Constanza Valdes. 2012. “Policy Technology Alves, E. R. A. 2010. “Embrapa: um caso bem-sucedido de inovação and Efficiency in Brazilian Agriculture.” USDA Economic institucional.” Revista de política agrícola 19: 65–73. Research Report No. 137. Bonelli, R. 2002. "Impactos econômicos e sociais de longo prazo da Rezende, G. C. 2002. “Ocupação agrícola e estrutura agrária no expansão agropecuária no Brasil: revolução invisível e inclusão cerrado: o papel do preço da terra, dos recursos naturais e social." In Anais do seminário sobre os impactos da mudança da tecnologia.” IPEA, Texto Para Discussao No. 913, Rio de tecnológica do setor agropecuário na economia brasileira. EM- Janeiro. BRAPA Secretaria de Administração Estratégica, Brasília, 241p. The Economist. 2010. “The Miracle of the Cerrado.” http://www. (Documentos, n. 5). economist.com/node/16886442. Brazilian Agricultural Research Corporation (Embrapa). http:// The Economist. 2011. “No Easy Fix.” http://www.economist.com/ www.embrapa.br/english/embrapa/. node/18200678. The Economic Premise note series is intended to summarize good practices and key policy findings on topics related to economic policy. They are produced by the Poverty Reduction and Economic Management (PREM) Network Vice-Presidency of the World Bank. The views expressed here are those of the authors and do not necessarily reflect those of the World Bank. The notes are available at: www.worldbank.org/economicpremise. 10 POVERTY REDUCTION AND ECONOMIC MANAGEMENT (PREM) NETWORK    www.worldbank.org/economicpremise