Political Economy of Fiscal Reform in Indian States

Abstract

The political economy considerations for fiscal reforms in a big federal country with a multi-party political structure involve a complex process, especially at the subnational level. India, for example, is an ideal example where fiscal reforms in the states present a unique opportunity to study this complex process. There are some important political factors, with ambiguous empirical results, that will have profound effects on fiscal reforms. Fiscal populism is one such factor. It may be noted that the government is not a ‘benevolent social planner’ that maximizes the utility of the representative individual but is concerned only with winning the next election. As the ruling political party expects vote swing in its favor, it may not indulge into fiscal profligacy like providing free electricity, subsidizing power, or granting tax concessions, etc. which enables the state government to raise its own revenue as a percentage of its total expenditure. This may not necessarily be true, as, a party can become less reform oriented if there is no political uncertainty or pressure. Political alliance is another important variable. When both the federal government and the state are ruled by the same party, it is quite likely that the particular state will enjoy some fiscal benefits. As a result, own revenue as a percentage of total expenditure of the favored states might increase, and the states may be reluctant to initiate hard reforms. On the other hand, a reform-oriented federal government might influence the states with the same ruling political party to initiate bold reforms. An additional political economy impediment, namely the common-pool problem where the major interest group that benefits from the status quo (not initiating hard reforms), is well represented within the government. Practically, no serious attempt has been made so far to understand the political economy of reform process at the state level in India in a comprehensive manner. The purpose of this paper is to settle the ambiguities in the relationship between fiscal reform and the political economy determinants empirically in the context of the India states covering fourteen major states in India during the period 2001–02 and 2013–14.

This paper was prepared for the Adjunct Fellow Program of the Asian Development Bank Institute (ADBI). I thank seminar participants, especially Rana Hasan, Bruno Carrasco, and Peter Morgan, at the Asian Development Bank and ADBI for valuable suggestions. I would also like to thank Dr. Sudipto Mundle and an anonymous referee for important suggestions. The usual disclaimer applies. The views expressed in this paper are those of the author and do not necessarily reflect the views and policies of the Asian Development Bank or its Board of Governors or the governments they represent.

Appendices

Appendix 1: Regression Results with Own-Tax Revenue as an Indicator for Fiscal Reform

	Model I	Model II
	PROBIT	PROBIT
Constant	−0.41 (−1.28)	−0.39 (−1.48)
VSCH: change in vote share	−0.12 (−1.95)	−0.11 (−1.89)
VSD: Changes in vote share adjusted for distance from elections	0.18 (2.24)
VSDGOV: Changes in vote share adjusted for distance from elections and strength of the ruling party in the assembly		0.31 (2.17)
Marginal effect for 1 year before elections (compare with the marginal effect in Table 3)	0.047

1. Source Author’s estimates
2. Note Figures in parentheses are t-statistics

Appendix 2: The Test of Endogeneity

Test of endogeneity follows the procedure suggested in Wooldridge (2002, p. 122) and Bun and Harrison (2014). We use Model II with VSCH and VSDGOV as the explanatory variables. Note that VSDGOV = VSCH * (percentage of total seats in the assembly by the ruling party/Time). Time takes three values 3, 2, and 1. Thus, if VSCH is suspected to be endogenous, so does VSDGOV. Let us denote (percentage of total seats in the assembly by the ruling party/time) by GOVTIME. We use OLS to test the endogeneity of VSCH and VSDGOV. The procedure is as follows:

Step 1: Run two OLS regressions with dependent variables VSCH and VSDGOV. The explanatory variables (instruments) are COMMITOWNREV, GOVTIME, Alliance, and COMMITOWNREV multiplied by GOVTIME.²¹ Save two residuals—V1 and V2.

Step 2. Run the original liner probability model (OLS) and PROBIT in Model II with the binary dependent variable as explained in the paper. The explanatory variables are VSCH, VSDGOV, V1, and V2. Notice that under the null hypothesis that VSCH and VSDGOV are exogenous, the coefficient of V1 = the coefficient of V2 = 0.

The estimated equations are as follows:

OLS:

$$ \begin{aligned} {\text{Binary}}\,{\text{Dependent}}\,{\text{Variable}} &= 0.32 - 0.07\,{\text{VSCH}} + 0.18\,{\text{VSDGOV}} + 0.05\,V1 - 0.13\,V2 \\ &\quad \left( {2.88} \right)\quad\left( { - 2.70} \right)\quad \quad\left( {2.61} \right)\quad\quad\quad \quad\left( {1.63} \right)\quad\quad \left( { - 1.57} \right) \\ \end{aligned} $$

R² = 0.21

PROBIT:

$$ \begin{aligned} {\text{Binary}}\,{\text{Dependent}}\,{\text{Variable}} & = - 0.47 - 0.21\,{\text{VSCH}} + 0.64\,{\text{VSDGOV}} + 0.15\,V1 - 0.40\,V2 \\ &\quad \left( { - 1.37} \right) \quad\left( { - 2.59} \right)\quad\quad \left( {2.43} \right) \quad\quad\quad\quad \left( {1.63} \right)\quad\quad \left( { - 1.56} \right); \\ \end{aligned} $$

Following this procedure, we derived the F statistic of 1.40 with 2 and 37 degrees of freedom for OLS and the likelihood ratio of 2.91 with 2 degrees of freedom for PROBIT. Both are not significant at 5% level. Thus, we do not reject exogeneity of VSCH and VSDGOV.