Agent-based Modeling of Tax Evasion

Chapter 9Agent-Based Simulations of Tax Evasion: Dynamics by Lapse of Time, Social Norms, Age Heterogeneity, Subjective Audit Probability, Public Goods Provision, and Pareto-Optimality

Sascha Hokamp and Andrés M. Cuervo Díaz

9.1 Introduction

“Essentially, all models are wrong, but some are useful!,” a well-known claim attributed to George E. P. Box (Box and Draper, 1987, p. 424). Of course, all models necessarily are a brief representation of a real problem and, thus, simplify the complexity of the real world. In addition, concerning tax evasion and the shadow economy, another issue arises, that is, how to measure what needs to be hidden in the shadow.

The early attempts to measure the shadow economy (Gutmann, 1977; Feige, 1980; Tanzi, 1980; Klovland, 1984) date back to the equations on currency demand by Cagan (1958) and the transaction approach by Feige (1979). Kirchgässner (2017, p. 99) identifies three approaches, which dominate the literature, (i) the “direct measurement” employing a survey method (Isachsen and Strøm, 1982; Feld and Larsen, 2005, 2012), (ii) the “indirect measurement” applying, a modified currency demand approach (Tanzi, 1980; Klovland, 1984), and, (iii) a model approach that originates from Weck (1983) and Frey and Weck-Hannemann (1984), well known as the (DYnamic) Multiple Indicators, MultIple Causes, (DY)MIMIC, approach. The latter is often used by Friedrich Schneider (with various co-authors, for example, Schneider, 2005, 2014; Schneider et al., 2011, ...