Three empirical essays in industrial organization.
详细信息
文摘
This dissertation contains three independent chapters in empirical industrial organization. Chapter one is on ecommerce industry. Theory predicts that seller reputation should have a stronger effect on an experience good than a search good. This chapter seeks to provide empirical evidence on different impacts of seller reputation on price between these two types of products in the online retail industry,using data from the Chinese ecommerce website,Taobao. By implementing a robust standard error ordinary least squares OLS) model and a weighted least squares model,this chapter shows that consumer ratings have a strong positive influence on an experience goods price,while the impact on a search good is not significant. Both Chapters two and three are on electricity retail. Chapter two attempts to measure the impacts of electricity restructuring on retail rates for all sectors in Massachusetts. The first part of the chapter disentangles historical price trends using other states in New England as a benchmark,and seeks to depict the slow customer migration and seemingly high retail prices after Massachusetts launched the restructuring in 1998. The second part of the chapter constructs two econometric models,a random effect model and difference-in-difference DD) model,to precisely estimate the effects of restructuring and the potential influences of switching from Standard Offer Service SOS) to Default Service DS). The conclusion is that Massachusetts restructuring effectively promoted competition and increased total welfare by strengthening the correlation between retail rates and fuel costs. The administrative controls accompanying with the reform efficiently lowered the retail rates compared to the rates that would have existed if full synchronization with energy costs was allowed. Chapter three analyzes various default/standard offer pricing models,proposes a profit maximizing problem,and specifies the hypothesis of a predictive relationship between default prices and competitive prices. It first tests unit roots in default and competitive rates across all year-state observations,then utilizes a dynamic panel data regression to evaluate causality. There is no significant causal or forecasting relationship found from the regression results. However,it discovered that competitive rates more closely follow previous changes in fuel costs,but changes in default rates are less dependent on changes in fuel costs.