2001) (Table 8-9). The benefit exceeded cost in SSA almost 15 years later than was the case for Latin America and Asia, causing the low IRR. The available evidence suggests that the economic RORs to agricultural R&D are high (Even-son, 2001). The broad scope of the evidence for high re­turns suggests considerable international spillovers. Eco­nomic RORs to agricultural research are likely to be above most public and private rates (Fuglie et al., 1996; Alston et al., 2000a; Evenson, 2001). Recently studies have been carried out on the impact of natural resource management research. In most cases however these were satisfactory but lower than in germplasm improvement research (Waibel and Zilberman, 2007).
          Due to a variety of market failures, private returns to R&D are far smaller than economic returns as private de­velopers cannot appropriate many of the benefits associated with their research (Evenson and Westphal, 1995; Scotch-mer, 1999; Shavall and van Ypserle, 2001) (see 8.1.2). In agriculture in particular, firms often have difficulty in cap­turing much of the economic benefits of their investments (Huffman and Evenson, 1993); e.g., in the US seed com­panies retained 30 to 50% of the economic benefits from enhanced hybrid seed yields and 10% of benefits from non-hybrid seed during 1975-1990 (Fuglie et al., 1996). A key market failure that inhibits developers from recovering the

cost of R&D in agriculture is the potential for resale of seeds (Kremer and Zwane, 2005). The gap between social and private returns may be more acute in tropical agriculture, where market failures are particularly severe and the intel­lectual property rights (IPR) environment is weaker (Pray and Umali-Deininger 1998; Kremer and Zwane, 2005).
          Very often the reported higher economic ROR is attrib­uted to the selectivity bias. First, highly successful programs are likely to be evaluated. Second, the unsuccessful evalua­tions are less likely to be published than evaluations show­ing impact. However, one can compare the studies covering aggregate programs, which includes both successful and un­successful, with studies of specific commodity programs and the evidence is based on a substantial part the world's agri­cultural research and extension programs (Evenson, 2001). Returns to AKST investments vary across continents, com­modities, types of research, methods of estimation, public versus private, and over time (Alston et al., 2000a) (Tables 8-10,8-11,8-12).
            The distribution of ROR for crops, livestock, and mul­tiple commodities is similar to that for the entire sample (Table 8-11). A substantial difference in the distribution of ROR is observed for resources research; these estimates mostly include forestry research, for which the research lags are relatively long, contributing to the relatively low average

NARSs

IARCs

Estimated benefits

Estimated

Lower range

IRR

B/C

IRR

B/C

Latin America

31

56

39

34

Asia

33

115

115

104

West Asia-North Africa

22

54

165

147

sub-Saharan Africa

9

4

68

57