Essays on Evaluating the Economic Effects of Agricultural Research and Development

Abstract

In this dissertation I investigate the economics of agricultural research and development (R&D) from two perspectives: first a meta-review and reassessment of the costs and reported returns to agricultural R&D in Latin America and the Caribbean (LAC), and second an analysis on the implication of including parameter uncertainty into rate-of-return measures benchmarked off the prior returns-to-research evidence worldwide. In the first essay, the reported economic returns to agricultural and food R&D in LAC are recalibrated and re-examined to provide insights for policy makers and analysts in their decision process to prioritize agricultural R&D investments. The recalibrated rates of return showed that the imputed MIRRs (modified internal rate of returns) generated as part of this study are systematically smaller than the corresponding IRRs (internal rate of returns) reported in the prior literature. Nonetheless, the newly estimated MIRRs are still sufficiently large to be indicative of an underinvestment in agricultural R&D throughout the LAC region. The large discrepancies between IRR and MIRR can send mixed signals to policy makers so careful interpretation of different rates of return measures is important. Developing plausible estimates of the returns to research is reliant on economic values that may be questionable. The second essay explores the implications of incorporating uncertainty into the parameters used to estimate the benefits attributable to agricultural R&D. The measured returns to agricultural R&D are especially sensitive to variations in the (estimated or assumed) unit cost reductions (K-shift) attributed to R&D. For example, for the data used to benchmark this study, a further 1% decrease in the relative median K-shift value (from 0.1295 to 0.1308) results in a 1.09%, 1.01%, 0.35%, and 0.17% increase in the NPV, BCR, IRR, and MIRR respectively. Furthermore, variation on the timing of initial benefits, the length (and by implication, the timing of the termination of benefits), and the shape of the adoption lag structure also have consequences for the measured returns to research. However, variation on the timing of initial benefits has the largest percentage changes into the rates-of-return; for instance, starting benefits a year earlier increase NPV and BCR by 7%, 12% for IRR and only 1% for MIRR. Moreover, IRR and MIRR measures of the benefits attributable to R&D appear to be more sensitive to changes in the structure of the R&D lag than changes in the magnitude of the K-shift, whereas the reverse holds for NPV and BCR. For instance, the coefficient of variation of uncertain K-shift values (0.98) is almost the same for NPV and BCR (1.05, and 1.00 respectively), whereas IRR and MIRR have smaller coefficients of variation (0.39 and 0.32 respectively), more than 60% smaller.