Analytical Study of Tax Evasion Dynamics Using Agent-Based Computational Models in Nigerian Informal Sector

Tax non-compliance in Nigeria's large informal sector, which accounts for an estimated 58 percent of GDP according to IMF measurements, represents a fundamental constraint on domestic revenue mobilisation and fiscal space for public investment. This study develops an agent-based computational model to simulate the emergence and evolution of tax evasion behaviour among heterogeneous informal sector agents interacting within a market network structure calibrated to the Aba and Onitsha commercial hubs in South East Nigeria. Agents are characterised by income levels, risk aversion parameters, audit probability beliefs, and social network connections through which information on enforcement experiences propagates. The model incorporates a tax authority agent that adjusts audit rates and penalty structures based on declared income distributions. Simulation experiments across 500 model runs explore how changes in audit probability, penalty rates, tax literacy programmes, and social norm conformity parameters affect equilibrium tax compliance rates and revenue yield. Results demonstrate that a 15 percentage point increase in audit probability raises equilibrium compliance by 22 percentage points, while doubling the penalty rate increases compliance by only 9 percentage points. Social norm interventions that shift the perceived compliance norm generate compliance improvements exceeding audit probability increases at lower administrative cost. Keywords: agent-based model, tax evasion, informal sector, Nigeria, tax compliance.