Download files
Abstract
This dissertation exploits recent advancements in panel data econometrics, as well as newly available sources of data, to answer questions about the causes and effects of rising inequality. The first chapter investigates the reasons why some advanced economies have experienced a sharp increase in the top 1% income share over the last three decades. Using a panel cointegration model it finds that both policy shifts and structural changes have contributed to this increase. The results suggest that the deregulation of labour, financial, and trade markets have had serious side effects on the level of equity in the economy.
The second chapter explores a link between this increase in inequality and the crisis of confidence in democratic political institutions that has emerged in recent decades, particularly following the Global Financial Crisis. A theoretical model is developed which argues that the income share of the top earners affects political confidence through its impact on the prevailing economic institutions. This prediction is tested empirically using multi-level modelling techniques, and finds a significant relationship between the two even after multiple robustness checks.
The third chapter argues that high levels of inequality among U.S. states are responsible for half of the lethargy that characterises the recent economic recovery. It creates a business cycle dataset of state employment over the last century, and then utilises survival analysis techniques to explore the determinants of the speed of economic recoveries. It finds that the speed of recovery and the distribution of income are strongly related.
The fourth chapter extends the Common Correlated Effects approach to estimating heterogeneous coefficients in panel data models that suffer from cross-sectional dependence. The extension replaces Ordinary Least Squares with General Method of Moments estimation, and uses lags of the variables to form the instrument set. By so doing the estimator is now robust to endogenous regressors, and based on an extensive Monte Carlo simulation study exhibits better small sample properties in dynamic models. The estimator is then applied to the topic of a long run relationship between inequality and violent crime using U.S. state data, and finds evidence for a positive relationship.