On the Structure of the Political Party System in Indian States, 1957 - 2018

Abstract

What accounts for the large and ever-changing number of political parties that contest Indian state elections? In this paper we test an equilibrium model of political parties where the number of parties depend on the average size of state constituencies, the heterogeneity of the state’s electorate, state per capita income as well as constitutional and legislative rules that directly affect party numbers. Extending the analysis to consider the effects on entry and exits highlights those factors that have affected party turnover instead of or in addition to changing aggregate party numbers. Comparison of the effects on numbers with those on the effective number of parties (ENP) suggests that the model explains variations in the number of smaller sized parties that are a more transitory part of the electoral process. Whether this implies that the model is more applicable to the lesser developed Indian states is examined by re-testing the model against a subdivision of states that differ by their stage of development. While most variables work in the same way, increases in the proportion of reserved seats have decreased both party numbers and ENP in lower income states, with the opposite occurring in higher income states. The effect of state partition has been to decrease party vote fragmentation more than party numbers.

Between Election Rivalry and the Agency Costs of Government: The Effectiveness of Party Competition Across Indian States, 1957–2018

“…The addition of election-year fixed effects dramatically reduces our degrees of freedom and in the limit allows for almost all elections to be accounted for by the combination of a state and time fixed effect. 18 To minimize the degrees of freedom issue, Jensenius (2017) and Ferris and Dash (2023) have used periodic fixed effects instead of the usual yearly ones. Assuming that the unobserved state invariant factors do not vary much over a decade, we have used ten-year periodic fixed effects.…”

“…Of the 195 state elections that took place over this time interval, we have complete data on 180 of these elections. 9 Descriptive statistics and data sources are presented in Appendix to this paper, and the data itself can be accessed at Ferris and Dash (2023).…”

“… 2. For example, in the state legislature of Uttar Pradesh, India’s largest state, 11 different parties were elected in the 2012 election, 8 in 2017 and 9 in 2022. See Chhibber and Nooruddin (2004) and Ferris and Dash (2023) for a detailed discussion of the determinants of party structure across Indian states.…”

“… 23. We follow Dash et al (2019) by defining a political party as any registered party in the top ten vote receiving parties over three successive elections or receiving more than 8 percent of the vote in any one election (all other parties and individual candidates included in a composite grouping, other). Winer et al (2019) develop this measure through 2013 and Ferris and Dash (2023) update that data for the elections through 2018. With these adjustments, the average ENPSeats (2.72) over the sample period is found to be not very different from average ENPConst (2.82).…”