Cross-Cutting Cleavages and Electoral Stability in India

• In Event: The Political Economy of Voting

Abstract

How do social and economic divisions impact electoral volatility? The main finding of this paper is that cross-cutting cleavages boosts electoral volatility.

There is a well-established notion that social and economic cleavages stabilize electoral politics. Applying this intuition to ethnic contexts, one would expect to find lower electoral volatility in multi-ethnic democracies. However, empirically we often observe the opposite: relatively higher electoral volatility in many ethnically heterogeneous societies. India is a notable example. Prior studies explain this incongruity either by emphasizing the role and configuration of social and economic cleavages or by attributing volatility to the scope of the developmental state and weak parties. This paper combines the two approaches. It argues that high cross-cuttingness boosts electoral volatility. By cross-cuttingness, we mean the extent to which social cleavages reinforce class divisions. The paper demonstrates that cross-cuttingness influences electoral volatility in two ways: First, cross-cuttingness has a direct effect on voting behavior. High cross-cuttingness boosts electoral volatility by making vote-choice less rigid, less dependent on identity of incumbents and voters, and more sensitive to economic performance of incumbents. Second, cross-cuttingness has an indirect effect on electoral volatility mediated through party institutionalization. The paper demonstrates that when cross-cuttingness is high, party—candidate linkages are weak, and party is more likely to switch candidates (or a candidate is more likely to switch parties) between two consecutive elections.

The paper measures cross-cuttingness between identity and class groups in three different ways. These include the extent to which identity groups overlap with class groups; where class groups are configured on the basis of i) occupation-types, ii) consumption, and iii) occupational-status. The first type treats each occupation as a class group in and of itself. The second type is agnostic of occupations and configures class groups on the basis of household consumption expenditure. The third type uses a hybrid model, wherein it first ranks occupations on status and skill; and then treats occupations with similar status and skill as a class group. In addition to cross-cuttingness, the paper also uses several other measures of social and economic divisions. These are i) vertical inequality or the Gini Coefficient of consumption inequality, ii) horizontal inequality or the Group Theil decomposition measuring consumption inequality between identity groups, iii) job fractionalization, measuring heterogeneity in occupations, and iv) ethnic fractionalization. To estimate social and economic divisions, the paper uses data from the National Sample Survey. These surveys, the most comprehensive of its kind, reports each respondent's religion, caste category, occupation, and household cash and non-cash consumption.

The paper uses data from elections to sub-national state legislatures, which collectively constitute more than 4,000 constituencies across 27 Indian states. Use of constituency-level data across India allows for comparison of units with the same institutional framework that face similar external shocks. There are also fairly stable cleavages and campaign issues within each state. These similarities allow us to rule out a number of alternative explanations and focus the analysis in on cross-cuttingness and electoral volatility, which are the main variables of interest in this article. The analysis covers all state assembly elections held in India between 1987 and 2007.

Empirically, the paper demonstrates that around 1/4th of all observed electoral volatility in India is mediates through party-switching, while vote-switching accounts for 3/4th of the observed volatility. More importantly, party-switching is more likely the cause of volatility in poor and less developed contexts with weakly organized parties; while shifts in vote choice of the electorate is more likely to cause volatility in relatively developed contexts with organized party systems.

Authors

• Francesca R. Jensenius, University of Oslo

• Neeraj Vimal Prasad, University of Amsterdam

• Pavithra Suryanarayan, Johns Hopkins University