The Identification of Regulatory Impacts Using Panel Data

This dissertation broadly explores the use of panel data methods for identifying the effects of regulation. Chapter 1 applies dynamic panel models to control for potential feedback effects when examining the relationship between environmental regulations and firm performance. Specifically, this chapter uses data from the U.S. manufacturing industry to explore the effects of the Clean Air Act (CAA) on firm performance outcomes related to profitability, employment, and capital expenditures. The evidence suggests that feedback effects are present in the data and, more critically, appropriately modeling these effects can alter conclusions regarding the effects of environmental regulation.

Chapter 2 uses panel data logit-class models to analyze whether the stringency of environmental regulation motivates mergers among regulated firms. Using comprehensive data set that capture both environmental and financial behaviors from a variety of data sources, results in this chapter provide new evidence on the impact of environmental regulation on manufacturing firm M&A activities, suggesting that more stringent environmental regulation (CAA) will motive manufacturing firms conduct more M&As, and market value of M&A firms will increase.

In Chapter 3, my coauthor and I use Monte Carlo experiments to provide insight on how power is influenced by data analysis methods, such as the choice of econometric estimator or whether all or a subset of data is used. Surprisingly, except in very few extreme cases, we find that power does not vary much if one relies on simple difference-of-means tests or instead on robust or structured (i.e. assumption-laden) panel data estimation. However, the simulations provide strong evidence of power loss when uses a subset of the collected data, and power gains through the inclusion of non-experimental control variables (e.g. demographic variables).