The dissertation consists of three essays: "Benefits of diversification in agriculture: Evidence from Malawi", "Corruption and economic growth revisited" and "How close is your government to its people? Worldwide indicators on localization and decentralization". All three essays focus on issues, which are particularly important for developing countries: food security, corruption, decentralization.In the first essay I use data from a farmers' survey in Malawi to compare two agricultural technologies: monoculture maize and crop diversification (maize-legume intercrop). I match farmers locations with data on rainfall and air temperature to test whether more biodiverse agriculture is better at absorbing weather shocks. The data make it possible to compare variation not only over time, but also over different plots within the same time period, which helps reduce omitted variable bias. The instrumental variable method is used to eliminate rainfall measurement error. For a number of specifications, and controlling for fertilizer use, crop diversification is both more productive than monoculture maize and more resistant to weather shocks. Although I am not able to identify the average population effect, I show that the effect I identify is likely to prevail if the Malawian government decides to shift the focus of its agricultural subsidy at the margin from fertilizer to legume seeds and education.The focus of the second essay is on corruption. While literature finds many channels through which corruption can hurt economic growth, the link proved hard to establish in empirical cross-country studies. In this paper I show that part of the explanation of this puzzle is that there is a reverse causality: everything else equal, exogenously-driven economic growth can increase corruption. The reason is that the boost to output increases tax revenue, and hence pool of resources that corrupt public officials can embezzle. I show the workings of this channel in a simple stylized model, which is then accompanied by numerical simulations in a dynamic general equilibium overlapping-generations model, which allows for corruption and tax evasion. I also present empirical evidence, which supports my findings.The third chapter assembles and analyzes a unique data base on local governance that provides a first approximation of the institutional architecture of local governance that has emerged as a result of the silent revolution (decentralization reforms) of the last three decades on moving governments closer to people. An important feature of this data set is that, for comparative purposes, it measures government decision making at the local level i.e. the order of government that is closest to the people and hence providing a better indicator of decentralized decision making as compared to the "sub-national governments" that also include intermediate tiers of provinces and states, used by the existing literature. This unique data set for 182 countries attempts to capture institutional dimensions of political, fiscal and administrative autonomy enjoyed by local governments under diverse multi-order governance regimes using a common framework. These dimensions are aggregated to develop a "decentralization index" and then adjusted for heterogeneity to develop a "government closeness index" that provides an overall ranking of countries on the closeness of their government to the people.
Read
