Application of Markov models in income tax revenue forecasting in Nigeria

Abstract

Accurate tax revenue forecasting is vital for effective fiscal policy and economic planning, especially in contexts marked by data scarcity and economic volatility. Traditional statistical and machine learning models often underperform when confronted with non-linear patterns and limited historical data. This research proposes a novel hybrid Grey-Markov model that synergizes the
trend-extraction strength of Grey System Theory (GM(1,1)) with the stochastic correction capability of Markov Chain Analysis. The Grey model forecasts long-term revenue trends using a minimal data requirement, while the Markov component dynamically adjusts predictions by modeling the probabilistic behavior of forecast errors. Through empirical validation using historical tax revenue data, the hybrid model will be benchmarked against conventional methods such as ARIMA, exponential smoothing, and neural networks. The anticipated outcome is a robust forecasting tool with improved accuracy and reliability, offering policymakers a more resilient framework for fiscal decision-making under uncertainty. This approach holds particular relevance for developing economies where data irregularities and economic shocks are prevalent.