Essays on Multisourcing, Trade and Education

Abstract

This dissertation contains essays concerned with multisourcing, firm dynamics and education. The first chapter examines the empirical evidence and motivations for multisourcing -- the purchase of the same input from multiple suppliers -- by U.S. importers of manufacturing inputs, using the confidential customs trade data from the U.S. Census. The paper highlights a striking and novel sourcing pattern that sourcing on the intensive and extensive margins both follow power laws. To explain the empirical pattern, I model a firm's tradeoff in paying higher prices and fixed costs for engaging with multiple suppliers versus diversifying from idiosyncratic supplier risks. Empirically, the paper confirms the model predictions in the cross-section of the data by assessing the relative importance of the determinants of sourcing. Second, using the Japanese Earthquake in 2011 and other sudden supplier closures as quasi-natural experiments, I find evidence that in response to exogenous supply shocks, U.S. importers substitute towards the unaffected suppliers. As such, the multisourcing firms are in better positions to minimize production disruptions than single-sourcing firms. The second chapter explores the dynamics of intermediate input sourcing using the confidential U.S. customs trade data from the U.S. Census. I document three key empirical facts on sourcing dynamics, and explain them using a model on supplier uncertainty and learning. I find first that new suppliers have high entry and exit rates, and second, they are given lower shares on average than incumbent suppliers. Third, over time, suppliers' shares tend to converge to an equal level. To explain these facts, I test the two key empirical implications from the Rauch and Watson model. First, supplier uncertainty leads to small orders to new suppliers, and second higher search cost leads to longer relationships. I find evidence consistent with the model and hypothesis of learning. The third chapter studies the impact of changing tuition on students' preferences for college degrees. In particular, it analyzes the selective tuition policy changes for certain degrees in Australia in the 2000s. Using a logit discrete choice model, I find that while tuition increases discouraged students from choosing those degrees, the effect was small as the estimated price elasticity of students' preferences is low. The estimated responses are however heterogeneous across income groups; students in the lower socioeconomic groups are more responsive to tuition changes.