Essays on Productivity, Uncertainty, and Firm Activities

Open Access
Li, Shengyu
Graduate Program:
Doctor of Philosophy
Document Type:
Date of Defense:
March 03, 2014
Committee Members:
  • Mark John Roberts, Dissertation Advisor
  • Mark John Roberts, Committee Chair
  • James R. Tybout, Committee Member
  • Paul L E Grieco, Committee Member
  • Spiro E Stefanou, Special Member
  • uncertainty
  • Bayesian learning
  • productivity
  • export dynamics; production functions
  • unobserved price bias
  • productivity dispers
This dissertation studies the interaction between firm heterogeneity in productivity and demand uncertainty and firm export participation. Chapter 1 develops a dynamic model to empirically study how firms' export decisions depend on productivity evolution and learning about firm-specific foreign demand. Chapter 2, a joint work with Paul L. E. Grieco and Hongsong Zhang, develops a new method to consistently estimate production functions when prices of intermediate input are heterogenous across firms but are not observed by researchers. Chapter 1: An Empirical Structural Model of Productivity, Uncertain Demand, and Export Dynamics This paper develops a structural model of export dynamics to empirically study how firms' market-level export decisions depend on productivity evolution and Bayesian learning about demand in foreign markets. Firms have uncertainty about foreign demand and they gradually learn about it based on the prices and quantities they observe in their individual export transactions in the Bayesian style. Firms' export decisions are dynamic and depend on the evolution of both productivity and beliefs about foreign demand. I empirically identify the role of each process in determining firm-market-level export participation and estimate the dynamic model. The identification and estimation use data on both firm shipment-level exports and firm-level production information for the Chinese ceramics industry. The empirical results indicate substantial firm heterogeneity in both productivity and demand uncertainty. Demand uncertainty is the dominant difference between potential entrants in export markets and experienced exporters. In particular, experienced exporters have higher expectations and face less uncertainty about foreign demand. Both the learning process and productivity evolution are driving forces of export participation for experienced exporters but for potential entrants the former plays a more important role. A further counterfactual exercise shows that reducing the level of uncertainty of potential entrants to that of experienced exporters causes the number of exporters to fall by 11%. Chapter 2: Production Function Estimation with Unobserved Input Price Dispersion We propose a method to consistently estimate production functions in the presence of input price dispersion when intermediate input quantities are not observed. The traditional approach to dealing with unobserved input quantities---using deflated expenditure as a proxy---requires strong assumptions for consistency. Instead, we control for heterogeneous input prices by exploiting the first order conditions of the firm's profit maximization problem. We show that the traditional approach tends to underestimate the elasticity of substitution and biases estimates of the distribution parameters. Our approach applies to a general class of production functions. It can accommodate both heterogeneity in input prices and a variety of heterogeneous intermediate input types. A Monte Carlo study illustrates that the omitted price bias is significant in the traditional approach, while our method consistently recovers the production function parameters. We apply our method to a firm-level data set from Colombian manufacturing industries. The empirical results are consistent with the predictions that the use of expenditure as a proxy for quantities biases the elasticity of substitution downward. Moreover, using our preferred method, we provide evidence of significant input price dispersion and even wider productivity dispersion than is estimated using proxy methods.