Central Government Revenue and Expenditure Relationship in Indian Economy: An Alternative Approach Based on Beveridge-Nelson Decomposition

This paper examines the Granger causal relation between government revenue and expenditure over the period 1971-2010. The study emphasizes on the idea that the relationship can truly be captured when the relation between anticipated parts and that between unanticipated parts of the variable are studied separately. In-sample one-step ahead ARIMA(3, 1, 1) forecasts for expenditure and ARIMA(2, 1, 0) forecasts for revenue constitute the anticipated parts for expenditure and revenue (xta and Rta) respectively. The respective forecast errors constitute the unanticipated parts of the variables. xta and Rta both I(1) variables, are found to be cointegrated. VAR(2, 3) model for these variables testifies to the unidirectional Granger causality from expenditure to revenue. Variance decomposition analysis based on this VAR(2, 3) model supports this finding. White noise unanticipated parts in the VAR(2, 1) model further testify to unidirectional causality from expenditure to revenue. This finding draws large support from the variance decomposition analysis for xtu and Rtu. The findings negate the ‘hypothesis of independence’ given forth by the study with observed series on xt and Rt.