## Thursday, February 27, 2020

### Does synchronization of elections matter? Evidence from India

by Vimal Balasubramaniam, Apurav Yash Bhatiya and Sabyasachi Das.

Many countries across the world hold elections for multiple levels of the government on the same day. Examples include the United States, Brazil, Sweden, South Africa, Indonesia, among others. Importantly, there has been an increasing demand to synchronize elections across tiers of governance in both Europe and India. In India, the Law Commission, and other bodies, highlight that elections are expensive and find that "holding simultaneous elections would be ideal as well as desirable". The implicit assumption in these discussions is that the question of when voters make decisions about their national and state representatives may not affect how they make these choices and consequently, the election outcomes that emerge from them.

In our research, we examine whether synchronized elections in India lead to significant changes in voter behaviour. We refer to an election in India as synchronized if the national election (or general election, GE) and the state election (or, assembly election, AE) occur on the same day. Otherwise, we say that the elections are non-synchronized.

Specifically, we ask the question how the probability that the same political party wins a seat at the Lok Sabha and the Vidhan Sabha changes when elections are conducted on the same day as opposed to on different days. For this, we compare the same assembly constituency over time with synchronized elections against those that happened on different days. For non-synchronized elections, we pair a national election with state elections that occurred after it and before the next national election.

We find that synchronized elections increase the probability that the same political party wins a seat both at Lok Sabha and Vidhan Sabha by 0.089, which is about 21% of the base probability of 0.42. One concern about interpreting this estimate as an effect due to synchronization is that a long time gap between national and state elections for non-synchronized elections may confound our ability to pin down a plausible causal interpretation of this estimate. We vary the time gap between the elections in any given pair of national and state elections from 150 days to 270 days, and our estimates range from 0.15 (for 150 days) to 0.082 (for 270 days). The estimates are, however, not statistically significantly different from each other. Our preferred specification is the one that limits the time-gap to 180 days -- an estimate closer to the lower bound that we find -- to account for qualitative reasoning that provides plausible exogeneity in the scheduling of elections.

Figure 1 below highlights the approach we take to this study with heat maps for the probability of the same party winning both the parliamentary and state constituencies without (Case 1) and with (Case 2) synchronized elections. Synchronized elections increase the probability of a political party winning both the Lok Sabha and Rajya Sabha seats. We show this for the ten states that fall within our sample for our preferred estimate. With the exception of Odisha – which has an opposite pattern – all other states in our sample present an increase in the likelihood of electing the same party.

Case 1: Unsynchronized Elections

Case 2: Synchronized Elections
Figure 1: Prob (Same Party winning both PC and AC)

This significant consequence of synchronized elections may not occur in isolation. We characterize the voting environment and find that the winning margin in any given contest at a constituency is on average no different between synchronized and non-synchronized setup. However, there is an increase in turnout for national elections to level with the average turnout for state elections during non-synchronized elections. This suggests that the fraction that participates in state elections is in general much higher than in national polls.

We explore the potential channels that drive this significant effect of synchronization. We find that synchronized elections reduce split-ticket voting -- the Euclidean distance between the vector of vote shares of political parties in parliamentary and assembly constituencies is significantly lower in synchronized elections. This reduction in split-ticket voting could be both demand and supply-driven.

On the supply side, political parties could homogenize information sets and hold similar campaigns for the two elections when they happen on the same day. They could manage greater engagement with voters on the ground due to economies of scale with campaign resources during synchronized elections. Both these factors could align a voter to a single party. On the demand side, it may be that the cognitive demand to rationalize voting for two different parties in the two elections when they vote for them at the same time is high. This increase in decision complexity may give rise to voting for the same party when elections are synchronized. To explore these motives, we use national and state election survey data collected by CSDS for a representative sample of individuals compiled within two days after every election in India.

Panel A: Voters Decision

Panel B: Election Priorities

Panel C: Voting Consideration

Panel D: Election Issues
Figure 2: Micro-data Evidence

We present the evidence from micro-data in Figure 2. We find that voters spend more time deliberating on elections when they are synchronized (Panel A), voters are less clear about objective functions for electing their representative for the two governments (Panel B). Additionally, we find that voters reduce the dimensionality of their choice by looking at political parties more than individual candidates (Panel C). Lastly, we find that voters are more ambiguous about the issue that matter the most for their choice of the vote (Panel D). These observations strengthen our claim that the cognitive challenges of choosing two candidates at once may not be a trivial constraint, especially in a parliamentary democracy where the elected representative matters to how the citizens voice their concerns to the state. And the objective function for the voter for the two elections is compromised with synchronized elections. Thus, we observe a rise in a simple solution of voting for the same political party during synchronized elections.

A critical reason for support to synchronized elections is the cost of holding elections. Holding elections on different days does have high electoral costs both for the governments to organize the elections and for the political parties to participate in them. The most recent General Election in India in 2019 cost Rs.5,000 crore or about 700 million USD.

Recurring elections not just imply more monetary cost but also the loss of governance time as politicians focus their time on campaigning and bureaucrats remain occupied with election work as opposed to implementing policies and public projects. The deployment of security forces away from their primary objective for electoral purposes also imposes further costs on the state. Lastly, the model code of conduct, it is claimed, affects public policy-making.

Such costs may reduce by holding synchronized elections had there been no impact on voter choice and decisions or on the real economic outcomes in India. Our results imply that a direct consequence of synchronized elections is synchronized representation. A growing body of work on political alignment provides mixed effects on real economic outcomes. Political alignment could increase the transfer of resources from the national government to subnational governments in India (Rao and Singh, 2003; Khemani, 2003). However, more recent work highlights that patronage networks and rent-seeking by local politicians may strengthen in politically aligned areas, leading to inferior public service quality. The development consequences of synchronized elections, therefore, are far from straightforward.

Our paper documents that when voters choose their representatives for Lok Sabha and the Vidhan Sabha matter to election outcomes. The administrative gains from synchronized elections, therefore, need to be weighed against benefits from voters evaluating different tiers of government without any overlapping ambiguity.

Vimal Balasubramaniam is researcher at Queen Mary University, London; Apurav Yash Bhatiya is researcher at University of Warwick, UK; and Sabyasachi Das is researcher at Ashoka University. This paper was presented at the APU-NIPFP workshop Strengthening the Republic, January 11, 2020.

LaTeX mathematics works. This means that if you want to say $10 you have to say \$10.