I study three forms of misallocation that distort production and lower aggregate economic output. Imperfect markets and imperfect governance are hallmarks of a developing country. I assess whether the misallocation caused by such imperfections might explain why developing countries remain poor. My first chapter develops a method to measure and separate the production misallocation caused by factor and financial market failures. When I apply the method to rice farming villages in Thailand I find surprisingly little misallocation. Optimal reallocation would increase output in most villages by less than 15 percent. By 2006 most misallocation comes from factor market failures. I derive a decomposition of aggregate growth that accounts for misallocation. Declining misallocation contributes little to growth compared to factor accumulation and rising farm productivity. I use a government credit intervention to test my measures. I confirm that credit causes a statistically significant decrease in only financial market misallocation. My second chapter studies why the poor have so many economic activities. According to one theory the poor do not specialize because relying on one income source is risky. I test the theory by measuring the response of Thai rice farmers to conditional volatility in the international rice price. Households expecting a harvest take on an extra activity when the volatility rises by 21 percent. I confirm the decrease in specialization costs households foregone revenue. I find no evidence for the alternate explanation that households under-specialize because they cannot afford lumpy business investments. My third chapter builds a model in which bureaucrats favor firms aligned with one political faction. Though they charge such firms fewer bribes the optimal bribe does not distort production in their favor. I test the notion by studying the politics of caste in Indian village councils. I exploit a regression discontinuity created by close elections for council president. Farmers of the same caste as a candidate who barely won are far less likely to have to bribe a bureaucrat than farmers of a caste that barely lost. But I find no evidence that winners use more inputs or produce more output.