penditures and activities as well as industrial spill-ins (Table 8-17). These studies report relatively high rates of return and are roughly equal to the social RORs to public agricul­tural research.

8.2.8 Impacts of public sector agricultural R&D investments on poverty
Several recent studies (Fan et al., 2000; 2004ab, 2005; Fan and Zhang, 2004) clearly indicate that public investment in agricultural R&D is the most efficient public sector invest­ment (with the exception of Ethiopia). In terms of number of poor people moving out of poverty, agricultural R&D investments ranked among the top three. Although limited, this evidence indicates that the investment in agricultural R&D performs equally as well or better than the other pub­lic sector investments and contributes significantly to pov­erty reduction. These studies measured the effects of public spending on growth and poverty reduction in selected Asian and African countries using pooled time-series and cross-region data (Table 8-18).To assess the impact of public in­vestment on poverty, the number of poor people who would come out of poverty for a fixed investment across different

sectors was estimated. Similar to estimate the economic ben­efit of the investment the benefit/cost ratios were estimated at the national level based on the increase in household in­come and/or productivity per unit of investment. In Asian countries, the growth effects of investments in agricultural research, roads and education are found to be large. Re­gional differences were observed within the countries. This demonstrates that there is an opportunity to improve the growth and poverty impacts of total public investments through better regional targeting of specific types of invest­ment (Fan et al., 2005).
         Effects of improved technology on income distribution across farms with different resource endowments have been ambiguous. Poverty reduction is largely about distribution. The benefits of agricultural research investments are large and undisputed, but their actual levels and distributional effects remain under discussion (Alston and Pardey, 2001). Measurement of distributional effects can, in principle, be made using economic surplus methods (Alston et al., 1995), although such measurement is not common. One reason why the debates continue may be that discussions of research im­pacts on poverty implicitly refer to only one of two separate