Three essays in development economics

The first essay estimates the effect of losing a banking relationship on a firm's employment growth. Using a differences-in-differences estimation strategy that controls for unobserved firm heterogeneity, I find a large and significant negative impact on the employment growth of affected firms in the post-banking crisis period. I investigate two potential channels of this effect: the information view, which states that banking relationships embody private information that can not be easily reproduced to new lenders, and the looting view which suggests that close relationships between banks and firms create incentives to loot depositors. I find evidence in support of the information view: older firms and firms that do not produce hard information are worse affected. I find weak evidence in support of the looting view: insider firms have a larger negative growth even after controlling for the size of outstanding debt of these firms.; The second essay uses microeconomic evidence from a randomized intervention of a novel Tsetse control technology in Western Kenya to evaluate the impact of an improved disease environment on productivity. Using a differences-in-differences estimation, I show that the technology has a positive and significant impact on the health of imported high yielding cattle breeds. I find weak evidence that treatment animals are less likely to suffer from Trypanosomosis. I am unable to show statistically significant differences in productivity between treatment and control groups. Additional evidence suggests that differences in actual and potential productivity are explained by lack of access to credit and information. In line with the predictions of the investment under uncertainty theory, I find weak evidence that treatment farmers invest more in intensive dairy farming.; The third essay (joint with J. Das, S. Dercon and P. Krishnan) investigates the circumstances under which the relationship between school inputs and educational outcomes can be identified. We recognize that households respond optimally to changes in school inputs and hence the relationship between school inputs and cognitive achievement. We present a household optimization model relating household resources and cognitive achievement to school inputs. In this framework, if household and school inputs are technical substitutes in the production function for cognitive achievement, the impact of unanticipated inputs is larger than that of anticipated inputs. We test the predictions of the model for non-salary cash grants to schools as the inputs of interest using a unique dataset from Zambia. We find that household educational expenditures and school cash grants are substitutes with a coefficient of elasticity between -0.45 and -0.7. Consistent with the optimization model, anticipated grants have no impact on cognitive achievement, but unanticipated grants lead to significant improvements in learning. This methodology has important implications for educational research and policy.