Three Essays on Empirical Industrial Organization and Time Series Forecasting
MetadataShow full item record
This dissertation consists of three chapters on empirical industrial organization and time series forecasting. In the first chapter, I evaluate the price effects of the 2010 United-Continental (UA-CO) merger in the U.S. airline industry. I first estimate air travel demand using a nested logit model, assuming travelers are homogenous. Then marginal costs are recovered. Based on demand estimates and recovered constant marginal costs, I employ a standard merger simulation method to predict post-merger prices, assuming a Bertrand-Nash equilibrium. The results suggest that this methodology underpredicts the price effects of the UA-CO merger. In the second chapter, I evaluate the impact of the United-Continental merger on consumer welfare. An important assumption in this chapter is that there are two types of consumers in the U.S. airline industry. Consumers are heterogeneous-business and leisure travelers. A discrete mixture nested logit model is applied to estimate passengers’ heterogenous air travel demand. Then air travel demand is used to measure the welfare effects of the merger on business and leisure travelers separately. I find both business and leisure travelers in the merger-affected markets suffer welfare loss in the post-merger time period, comparing to those in the merger-unaffected markets. As of 2015, the UA-CO merger had not yet reached its expectation. In the third chapter, we first identity economic factors that influence two specific classes of wheat: hard red winter (HRW) and soft white (SWW) wheat, and develop models to improve the forecast performance of wheat basis in Washington State. The models we estimate include: 1) a simple three-year moving average model to serve as benchmark; 2) an econometric fundamental model; 3) an ARMA time series model; and 4) an ARMAX hybrid model. An econometric fundamental model and an ARMAX model include supply/demand factors suggested by economic theory and literature. We find the best model for both HRW and SWW is a function of forecast horizon. In addition, the ARMAX models perform better than the ARMA models in most cases, except SWW in Odessa, WA.