Are voters delivering more decisive mandates?

Contrary to conventional wisdom, a comparative study of elections since 1980 indicates that verdicts are, if anything, getting less decisive and election wins are becoming narrower

Are Indian voters empowering governments with more decisive mandates? This is the emerging wisdom among many experts and observers of Indian politics. For instance, pollster Pradeep Gupta, in his book, How India Votes: And What It Means, has argued that new forms of technology are enabling increased communication between voters during the rough and tumble of the campaign period. This, in turn, is enabling voters to consolidate their opinions about a given candidate or party’s odds of winning, leading to a shift of support away from likely losers. Voters, the argument goes, are learning how to rally behind election winners, giving them supersized mandates.

This thesis seems eminently plausible. A cursory glance at recent assembly elections suggests that landslides abound. The Aam Aadmi Party’s sweep in 2015 and 2020 in Delhi, the Bharatiya Janata Party (BJP)’S 2017 victory in Uttar Pradesh, and the Left Front’s and Trinamool Congress’s decisive wins in Kerala and West Bengal, respectively, in 2021 — all point in the same direction. Yet anecdotes can be misleading. In fact, our analysis of historical data on Indian elections suggests that this popular wisdom may well be misguided.

We looked at data from 249 assembly elections, taking care to code all parties and pre-poll coalitions in state polls going back to 1980. Because some parties that claimed to be in alliance did not actually adopt robust seat-sharing arrangements, we considered parties to be in a “true” coalition only if they did not run against each other in more than a handful of races.

Additionally, because an election can be “decisive” in different ways, we calculated three different measures of “election decisiveness” — the seat-share margin between the winner(s) and runner(s)-up, the vote-share margin between the winner(s) and runner(s)-up, and whether the election resulted in a hung assembly.

Using each of these three measures —across a range of specifications, including limiting our sample to only major states with larger populations — our results are strikingly consistent. Contrary to received wisdom, elections have largely been getting less decisive over time, although the effect is generally not statistically significant.

After notching up larger mandates in the early 1980s, winning parties (or coalitions) exhibited an average margin of victory— in terms of seat-share difference between the winner and runner-up — of between 35 and 39% until the end of the 1990s. Margins of victory narrowed to roughly 32% in the first decade of the 21st century, rebounded to their earlier levels between 2011 and 2015 (39%), and finally fell back to their previous lows (33%).

The data on vote-share shows the same broad trends, while the data on hung elections was more inconclusive. 18% of assembly elections between 1980-1985 failed to produce a clear majority winner. This figure stood at 30% in the late 1990s before dipping to 21% between 2016 and 2021.

In other words, there is little evidence to suggest that elections have been getting more decisive over time. If anything, the data suggests the opposite — by some measures, election victories have been getting slightly narrower.

This aggregate analysis looks at the average outcomes of state elections across India. However, while some changes, such as the rise of social media, might affect voters across India more or less equally, changes within particular states may be just as or even more important.

Indeed, when we disaggregate election decisiveness by state, we see a great deal of regional variation. In states such as Gujarat, there is a relatively clear trend toward narrower election victories. Yet, in Uttar Pradesh, a shift toward narrower election victories in the 1980s and 1990s has since given way to a strong trend toward more decisive victories in the 2000s. And in other states, such as Andhra Pradesh or Bihar, there seems to be no clear trend in either direction.

What might explain these findings, which seem to contradict the conventional narrative?

One possibility is that new avenues for coordination and consensus among voters are being counteracted by forces that discourage communication and shared opinion. While social media might enable voters to converge on a shared body of information, for example, it might also separate voters into multiple, insular communities that increasingly diverge in their opinions and beliefs about the world— including which parties or candidates are likely to win.

Additionally, while new technologies might enable certain forms of voter cooperation, they also provide political parties new avenues to make their case to the public and influence voters. This can allow more marginal parties to cling to a narrow support base even when they have little chance of winning outright.

Another possibility is that voters are indeed voting more strategically but, given the inaccuracy of pre-election polling and the general uncertainty of election outcomes, there are limits to their ability to predict winners and losers in advance. Indeed, research on Muslim voting patterns suggests that coordinated bloc voting is regularly stymied by the powerful influence of local factors, which often weigh heavily on voters’ minds.

While independent candidates and candidates from more marginal third parties may be easier to write off as unlikely to win — indeed, we find that this group is winning significantly fewer seats over time — if runner(s)-up continue to appear to be serious contenders, the typical margins of victory the winners enjoy will remain unchanged.

Finally, even if Indian voters are voting more strategically, this effect could be offset by other factors known to influence election outcomes, from anti-incumbency to a “honeymoon effect”, whereby the electoral fortunes of the national ruling party contesting state elections will depend on whether or not the assembly election closely follows a general election. In this case, simply examining trends in “decisiveness” over time may be glossing over important variations in voter behaviour.

Rather than seeking out universal explanations of how Indians vote, it may be more productive for election analysts to take a narrower, more context-specific approach.

If we have learned one thing from decades of scholarship on Indian voters, it is that they have an endless capacity to defy simple narratives.