Szabó, Attila, Gulyás, László and Tóth, István János (2008), "TAXSIM Agent Based Tax Evasion Simulator", 5th European Social Simulation Association Conference (ESSA 2008), (pre-print).
This paper introduces the TAXSIM model for the simulation of tax evading behavior in a computational model of a single market sector. The agent -based model defines four kinds of agents: employee, employer, (tax) authority and government. The rate of tax evasion is an agreement between an employer and an employee that is made to reduce costs. During negotiations the expectations of both the employer and the employee depend on the estimated costs (e.g., fines or the costs of missing governmental services) and benefits of evasion and on the level of satisfaction the agents have with the government’s services. The agents’ expectations and satisfaction are results of the agents’ individual learning, based on their own experiences and on those in their social network. The TAXSIM model produces market equilibriums with a wide variety of employment type distributions (there are 23 possible employment types in TAXSIM, ranging from fully legal to completely illegal). Moreover, the model provides means to test the agents’ sensitivity to the frequency and adequacy of tax audits. The series of experiments reported in this paper analyze scenarios in which i) the quality of governmental services increases permanently, ii) a market leader unilaterally adopts the legal position, and iii) multi-national companies with tax allowances enter the market.
agent-based computational economics, agent-based simulation, tax modeling, tax evasion