Tracy J. Noga
Suffolk University
Robert C. Ricketts
Texas Tech University
Abstract: An extensive body of tax literature posits that taxpayers respond to changes in tax laws, especially marginal tax rates. The various elasticity estimates, however, are quite divergent. In order to reconcile past results and propel elasticity research further, it is necessary to discover the source(s) of the divergence and reconcile some of the extant results. This study, through the use of panel data, compares several methods of analysis (random coefficient regression, fixed effects model, random intercept model, first difference regression, difference-in-difference and, pooled and single years ordinary least squares regression) while holding data set, variable selection and variable measurement constant. The study provides insight into the predictive ability of several different models and data sets in an individual taxpayer debt burden context. The conclusions indicate a need for continued compilation of panel data and further research in the areas of variable selection, definition and measurement.
Back to Program
