We estimate that Brazil received $16 of benefit from every dollar invested by Embrapa in improving upland rice, edible beans, and soybean varieties. The total research benefits over the period 1981–2003 amounted to $14.8 billion in present value (1999 prices) terms—or 6.1 percent of the corresponding value of crop output—of which $3.1 billion were attributed to the efforts of Embrapa. These benefits to Brazil came from either maintaining yields in the face of pressures that would otherwise cause them to fall, or improving the yield performance over time relative to base-year yields. They represent the gains from varietal improvement research alone, abstracting from other factors that can affect yields. The upland rice program has also substantially emphasized the need to improve the quality of the grain. We estimated the benefits arising from this aspect of that research amounted to $233 million, in addition to the $1.68 billion attributable to the yield-enhancing
aspects of the rice research.

Embrapa’s varietal improvement research investments have been profitable overall, primarily because of a very high benefit–cost ratio for soybean research. Although the quantitative details may change, the qualitative pattern was preserved when we investigated the sensitivity
of the estimates to changes in the interest rate used to discount the benefit and cost streams (4 versus 10 percent), and the length of the benefit stream (benefits truncated in year 1998 versus 2003). These benefit–cost ratios are backward looking: they reflect the benefits
accruing to the past investments, specifically the investments made between 1976 and 1998.

Nonetheless, to the extent that the future can be expected to be like the past, they provide an indication that crop improvement research by Embrapa, especially research on developing improved soybean varieties, would be a very profitable investment of public funds in the future.

As such, the results here provide strong support for claims to sustain and even increase funding for all three programs of research, especially varietal improvement research on soybeans. Although the benefits attributable to Embrapa are large absolutely and relative to the size
of the costs of research, the results indicate that sources other than Embrapa contributed significantly to the benefits. In addition, the share of total benefits attributable to Embrapa versus other agencies varied among the different crops and periods considered. The evidence indicates that CNPSo (the Embrapa soybean center) accounted for a sizable and increasing, but not dominant, share of the benefits from improved soybean varieties since 1981—using a geometric attribution rule to apportion the contributions of past breeders to each improved
variety of commercial consequence, 9 percent of the benefits for 1981–85 and 28 percent for 1993–98. The genetic material underpinning all these gains drew heavily on material from non-Embrapa sources (significantly, the United States). The non-Embrapa content of upland
rice varieties was much more reliant on domestic sources compared with soybeans, whereas edible bean varieties drew more heavily than either rice or soybeans on foreign sources—CIAT, Colombia has been a major source of the pedigree material used by CNPAF (the Embrapa
rice and bean center)—and other local breeders, and a nontrivial amount of foreign sourced edible bean varieties were used directly by Brazilian farmers.