This dissertation takes two different research perspectives to address the central theme
of agricultural production and productivity.
The first two essays focus on household production, which, as the primary form of
agriculture to date, not only affects the welfare of individual rural families but also food
supplies on a global scale. Agricultural productivity hinges largely upon farmers’ choice
of technology, inputs, and management strategies. Specifically, the first two essays
investigate land fragmentation, a common farming practice worldwide, and evaluate its
impacts on agricultural production. Chapter 2 argues that land fragmentation enables
farmers to reduce risk by diversifying production among discrete plots of land which
may be subject to heterogeneous growing conditions. Using Tanzanian household survey
data, this essay finds robust evidence to support a risk-reduction hypothesis and
indicates that land fragmentation is positively associated with production efficiency.
Chapter 3 develops a production model that incorporates risk, production efficiency, and
risk preferences and shows that land fragmentation may encourage risk-averse farmers
to increase labor intensity, thereby leading to higher efficiency. It is also shown that
exclusion of risk preferences from efficiency analysis may lead to biased or even
misleading estimates.
The second focus of this dissertation is an assessment of the published evidence on the
payoffs to investments in agricultural research and development (R&D). The related two
essays focus on methodological as well as policy issues underlying the agricultural R&D iii
evaluation literature. Specifically, Chapter 4 scrutinizes the prevailing internal rate of
return (IRR) measure and argues that it is based on implausible assumptions that often
lead to inflated estimates of the returns to research. This essay develops a novel method
for recalibrating the reported rates of return using a more plausible modified internal rate
of return (MIRR) measure and derives more modest estimates. Using the detailed
information collected for each R&D evaluation, Chapter 5 examines how the wide
variation in the reported IRR estimates can be explained by factors such as research
type, research focus, commodity type, institutional aspects of the research, target region,
and methodological specifications. The findings have important implications for future
agricultural R&D policy as well as R&D evaluation methodologies.