Econometric simulation of the income tax compliance process for small businesses

Econometric models can be very useful for estimating the marginal impacts of changes in policy. However, their broader application as a tool for micro-simulation analysis poses a number of challenges and limitations. This paper uses the context of modeling taxpayer compliance burden for small businesses to explore some extentions to standard econometric simulation techniques that provide more robust support of the distribution of the characteristics of interest. Key to the approach is explicitly simulating a random draw from the specified error distribution and a pair of calibration factors reflecting some of the technical limitations of a finite simulation. Further technical considerations regarding the retransformation of the dependent variable in a log-linear regression model are also discussed. Final comments include thoughts on potential refinements and implications for simulating the domain area of interest beyond the current scope of small business taxpayers.