Agent-based Modeling of Tax Evasion

Foreword

We model social systems to better explain, predict, design, and act. Macroeconomic models explain growth rates and patterns. Financial models predict stock prices. Models of correlated assets help the Federal Communications Commission design their spectrum auctions. And industrial organization models of firm competition inform decisions to intervene by the Justice Department.

The aforementioned macroeconomic models differ in their assumptions and methodologies. Macroeconomists gloss over firm level strategic choices that lie at the core of the industrial organization models. Auction models assume strategically rational actors, whereas some financial models assume actors who buy and hold.

Models include some variables and leave out others. They simplify the world, and as a result, err. We can correct for the biases and faults in any one model by constructing complementary models. To be of use, these other models must capture or include different features or relationships. They can then fill in the gaps missed by the original model. By offering more coverage and encompassing a larger number of processes, an ensemble of models provides deeper and broader knowledge and more refined insights than does a single model.

This book exemplifies this many-model approach. The chapters within explore a single policy domain – tax evasion – through the lenses of a variety of agent-based models (ABMs). While each model makes a substantial stand-alone contribution, the volume in its ...