This article empirically examines economic and political determinants of the aggregated as well as the channel-specific Central transfers to states in a developing country like India. An annual panel data set for 22 Indian States for the period of 1980-81 to 2010-11 is used for this purpose. Considering non-availability of continuous data for the study period, States like Arunachal Pradesh, Chhattisgarh, Goa, Jharkhand, Mizoram and Uttaranchal are not considered in the study. Dynamic panel regression equations are specified to analyze the determinants of Central transfers. Various components of Central transfers are alternatively regressed on a set of economic-demographic and political variables like per-capita gross state domestic product, population, literacy rate, area, fiscal dependency, loyal voter, State election year dummy, political alignment and bargaining power. State-specific dummies are included in all models to control for the unobserved time-invariant State specific effects. Linear time trend (year) is also included in all equations, as some of time series variables used are found to be trend stationary. System generalized methods of moments (GMM) estimator with heteroskedastic corrected robust standard error is used to obtain coefficients. Post-estimation tests like the Sargan test for over-identifying restrictions and Arellano Bond autocorrelation tests at first differenced error are used for verifying the validity of model. The results suggest that all categories of Central transfers are affected by political factors, besides economic-demographic variables. The states having partisan alignment with the Centre and greater political bargaining power tend to receive larger per capita central transfers. While loyal voter variable has significant and positive effect on per capita discretionary grant, state election year dummy never emerges to be statistically significant. This indicates that the transfers in Indian case are mainly targeted to core voters than to create swing. Among the economic variables, potential influence arises from the fiscal requirement and level of service provisions in the state. Findings of the study are observed to be robust and immune to change of specifications from linear to log-linear and omission of special category states from the data set. Recommending formula-based transfers, bringing all components of transfers into formula net, objective assessment of non-plan revenue deficit grant and co-ordination among resource recommending agencies in designing weights and criteria would help to check political influence.
Read full abstract