Factors to Curb Tax Evasion: Evidence from the TAXSIM Agent-Based Simulation Model

Agent based models are proposed as an adequate tool for analysing tax payer decisions and, thereby, the consequences of such decisions as they manifest themselves at the macro level. TAXSIM models the conduct of agents of three types, i.e. employers, employees and the government, in an economy of a single sector. Using this model, we examine the conduct of each individual agent and the impact of their decisions on the size of tax evasion and aggregated tax revenues. The main objective is the analysis of the relations between the government and tax payers, as well as the identification of how this relationship affects tax evasion. In this paper, the operation of several factors affecting the incidence of non-registered employment is studied with the help of an improved version of the TAXSIM agent-based simulation model. We analyse the effects of unemployment experience, the quality of government services and the audit strategy of the tax authority on totally hidden, mixed (hidden and legal) and totally legal payments. We start by a brief overview of the TAXSIM model and introduce the novel components in detail. This is followed by a thorough analysis of a vast series of computational experiments, analysing the behavior of the base model, as well as the effect of the newly introduced components (i.e., unemployment experience, adaptive audit strategy, etc.) and their various parameters. Our computational results suggest that the government can use effective tools to improve tax morale and curb tax evasion. Beyond improving the frequency and precision of tax audits the implementation of an adaptive tax audit strategy, the reduction of the exposure to unemployment or, on the motivation side, the improvement of the quality of government services can also be effective. Revised version of the paper presented in Münster at the Shadow 2013 Conference (http://www.wiwi.uni-muenster.de/shadow2013/)