Revenue forecasting of corporate income tax (CIT) in India

Revenue forecasting is an integrated part of annual budgeting exercise of the government. Literature on revenue forecasting is sparse in India. To fill this gap in literature, an attempt is made in this paper to forecast the revenue of Corporate Income Tax (CIT) collection. Based on available quarterly data of CIT collection, real Gross Value Added (GVA) and statutory CIT rate for the period Q1 of 2011–12 to Q1 of 2023–24, we develop a conditional model of CIT revenue forecasting using a Vector Auto Regression (VAR) model. In order to address the issue of seasonality, we separately model the seasonal component using a univariate Seasonal Auto Regressive Integrated Moving Average (SARIMA) model. The non-seasonal component of the CIT revenue series is modeled following the VAR framework using CIT revenue, real GVA and CIT rate series. The Theil inequality index based on in-sample forecast error for the growth in the CIT revenue series is found to be 0.17. The Theil inequality index based on out-of-sample forecast error for the growth in the CIT revenue for the period Q1: 2021–22 to Q1: 2023–24 is found to be 0.25. The average absolute percentage out-of-sample forecast error of the level of CIT revenue for the period Q2: 2022–23 to Q1: 2023–24 is estimated to be 0.12. The VAR model out performs the univariate SARIMA model in terms of out-of-sample RMSE and Theil Inequality Index. However, using Diebold-Mariano test, we find that the both VAR-based model and the univariate model have similar predictive powers. Possibility of further improvement of the model will be explored in the future research.