TAX AMNESTIES AND COMPLIANCE IN THE LONG-RUN - A TIME-SERIES ANALYSIS

This paper discusses time series methods that can be used to examine the long run effects of a tax amnesty, and applies these methods to the 1985 Colorado amnesty. Several time series models are estimated: simple ordinary least squares time trend models, univariate time series models, and multivariate intervention models. The empirical results from all models strongly indicate that the Colorado amnesty had no long run impact on either the level or the trend of tax collections. This result suggests that a typical amnesty seems unlikely to generate significant new revenues, but also seems unlikely to compromise voluntary compliance.