Omitted Variable Bias in Time Series Estimates of Capital Gains Realizations

Abstract

This paper addresses the omitted variable problem in time series studies of individual capital gains realizations. While not directly identifying the omitted variables, I use the realizations of corporate capital gains to test for common omitted variables. Jointly estimating individual and corporate realizations provides more efficient estimates than traditional single-equation models. The results provide empirical support for omitted variables being important in time series models. In both single-equation or multi-equation models controlling for omitted variables greatly reduces the estimated elasticities of individual and corporate capital gains.

Keywords: capital gains, tax

JEL Classification: H2, C3