UAB researchers from the Department of Sociology have developed a computer model which simulates, in different situations, the behaviour of taxpayers when faced with the possibility of committing tax evasion. The simulator, described in the journal Advances in Complex Systems, analyses the factors motivating tax evasion and allows to determine which measures are effective in reducing it, among which are an improvement in tax inspections by increasing their frequency and efficacy.
Tax fraud is a very serious problem for society, especially in Spain, where tax evasion represents almost one-fourth of its Gross Domestic Product. On the one hand, evasion is a problem because it produces a loss in public resources, something which is especially difficult in a time of economic crisis with cutbacks in public funding; on the other hand, tax fraud damages the effectiveness of justice within the tax system, since not everyone is equally able to evade taxes, and this leads to injustices between small and large companies and between those self-employed and employees.
The study of the causes behind tax evasion is a relatively young field which has been dominated mainly by economists. According to the Neoclassical economic theory, the decision to evade taxes or not is the result of a rational analysis which takes into account the benefits of evasion (the monetary amount saved in taxes) in relation to the potential costs (of tax inspectors discovering the evasion and being charged with a fine). Nevertheless, it is becoming more and more clear that this view is insufficient in explaining such a complex phenomenon as tax fraud and that is why other explanatory mechanisms have been devised from the viewpoints of psychology and sociology.
Researchers from the UAB GSADI research group (Analytical Sociology and Institutional Design Group) have tried to explain the behaviour of a tax evader through an integral model named SIMULFIS. The model is a agent-based social simulation, a computerised technique which allows creating virtual societies formed by agents which have specific individual and relational characteristics which take decisions by following a series of rules. It is the first time that different factors are integrated into this type of simulator with the aim of creating realistic results.
Researchers calibrated the simulator to make it reproduce real traits found in the Spanish society, such as tax rates, income and occupational distribution.
The SIMULFIC agents, which correspond to taxpayers in real life, decide to make use of tax evasion opportunities after passing through a series of conditions and filters. The first condition is prescriptive: when taxpayers believe the state are treating them justly, their inclination towards tax evasion diminishes. Next, they are passed through the rational election filter: the agents calculate if they will benefit by evading taxes after considering the inspections and sanctions. Lastly, a social influence filter is applied: the more tax evaded by neighbours, the more tax evasions appear in the simulation.
The simulator can be used to predict the effects of different measures taken against tax evasion to try to reduce this phenomenon. Among the conclusions of the study, researchers observed that improving tax inspections -- by increasing their frequency and efficacy -- would be a more effective measure against fraud than raising the amount of fines issued.
"The results of the experiments conducted with SIMULFIS allow us to reaffirm that we are working with a promising tool which will help to explain the level of tax fraud amongst a society based on individual decisions taken by the taxpayers," concludes Toni Llΰcer, researcher at the UAB Department of Sociology and co-author of the study, together with researchers Josι Antonio Noguera (director of the study), Eduardo Tapia and Francesc J. Miguel.
- TONI LLACER, FRANCISCO J. MIGUEL, JOSΙ A. NOGUERA, EDUARDO TAPIA. AN AGENT-BASED MODEL OF TAX COMPLIANCE: AN APPLICATION TO THE SPANISH CASE. Advances in Complex Systems, 2013; 16 (04n05): 1350007 DOI: 10.1142/S0219525913500070
Cite This Page: