What Abhijit Banerjee and Thomas Piketty missed in their study of Indian voter behaviour

The celebrated economists have argued Indian voters vote for caste and religious issues, and not for social spending.

Representational image | Voters stand in line at a polling station in Varanasi, Uttar Pradesh | Photo: Kanishka Sonthalia | Bloomberg

Economists Abhijit Banerjee, Thomas Piketty and Amory Gethin published a paper on Indian electoral behaviour in March this year, just weeks before the Lok Sabha elections. In the paper, they argued that Indian voters vote on account of caste and religion, and not on the government’s social spending or their own social class.

They wrote, “voters seem to be less driven by straightforward economic interests than by sectarian interests and cultural priorities. In India, as in many Western democracies, political conflicts have become increasingly focused on identity and religious-ethnic conflicts rather than on tangible material benefits and class-based redistribution.”

Banerjee and Piketty advised the Congress on devising a poverty alleviation scheme that the party then promised in the Lok Sabha elections a few weeks after the paper was published. The scheme was called the Nyunatam Aay Yojana (NYAY), and it proposed to make sure no Indian family earns less than Rs 72,000 a year. If they did, the Congress, if voted to power, would give them enough cash by direct transfer to make up for it.

The idea did not help the Congress win the election—bearing out the hypothesis laid out in the paper the three economists had published in March.

Ignoring the swing voter

In their paper in March, titled ‘Growing Cleavages in India? Evidence from the Changing Structure of Electorates, 1962-2014’, they use two sets of data to make their point. They use the CSDS-Lokniti post-poll surveys to measure shifts in voter behaviour by caste, and find that the shifts have been minor. Second, they use social spending data of governments over the years to see if greater spending on “education, sports, art and culture, medical and public health, water supply and sanitation, housing, welfare of scheduled castes, scheduled tribes and other backward castes, social security and welfare, and labor and labor welfare” results in re-election, and find that it has not. The only exception is caste-based reservations.

The first problem with their hypothesis is that they used very limited data to make broad, sweeping generalisations.

The economists argue that the political cleavage is now so well-defined that parties speak to different electorates. This is not true: there are enough swing voters within caste and religious groups, otherwise, no party would ever lose an election.

There is no doubt that different social and religious groups are attached to different political parties. So, for example, upper castes are more likely to vote for the Bhartiya Janata Party while Dalits are more like to vote for the Congress. What the economists miss is that even small shifts are huge determinants of electoral outcome in a first-past-the-post system. A vote-share change as small as two per cent can translate into a vast number of seats. It is thus unfair to ignore or dismiss these seemingly minor shifts in the voting preferences of caste groups.

There was, for instance, a significant shift in Dalit and Adivasi votes towards the BJP in the 2014 Lok Sabha elections, one that has stayed if not increased since then. They dismiss this as an exception: “The 2014 election was an exception: for the first time, nearly one third of SCs and STs supported the BJP and other right-wing parties. However, support for the right among other caste groups increased in similar proportions, leaving the voting gaps between upper castes and lower castes essentially unchanged.”

But it is not as mechanical as that. Just because more upper castes start voting for one party, it does not necessarily mean more lower castes will also start voting for that party by default. There has to be something common for these inimical groups to be voting together. In 2014, that commonality was Narendra Modi’s promise of the good days—low inflation, reduction in corruption, development and prosperity.

The data used by Banerjee, Piketty and Gethin clearly shows, for instance, that Brahmin support for the BJP declined marginally between 1998 and 2009, and went up sharply in 2014. In 2009, it was below 50 per cent. Isn’t that strange, considering that L.K. Advani ran a national security campaign in 2009, around Afzal Guru and the 26/11 Mumbai attacks that had taken place just months before the election? According to the paper’s hypothesis, the BJP should have won the 2009 Lok Sabha election.

Why parties lose

The Congress, on the other hand, campaigned in the 2009 election around rural employment guarantee (NREGA), the Right to Information and a farm loan waiver. It won handsomely, winning over 200 seats. According to the paper’s hypothesis, the Congress should have lost that election, since voters don’t reward social spending.

They look only at social spending by incumbent governments, not at what the opposition promises to win an election. They explain why parties win (caste and religion), but not why they lose.

The Manmohan Singh government’s rights-based approach increased social spending between 2004 and 2014. This focus on social spending took place on the understanding that the BJP had lost the 2004 Lok Sabha elections on the “India Shining” slogan, and the Congress had won on the slogan, “Aam Aadmi ko kya mila? (What did the common man get?)” It was the first Lok Sabha election since the 2002 Gujarat pogrom, whose memory was still fresh. If religious polarisation helps parties win elections, Atal Bihari Vajpayee should have been back in power.

Beyond datasets

There is no dearth of such examples, which may be dismissed as mere anecdotal evidence. Yet, data cannot explain everything. Data often hides more than it reveals. What this paper hides is the role of campaigning in election results. They study voter behaviour according to social spending, thus presuming that people vote only on past performance. Yet, an election is a vote for the next five years. What if people think the alternative is even better, and might spend even more on development? What if the social spends don’t reach people, or the government is too mired in corruption scandals, or the social spends selectively target some social groups or geographical regions?

There are any number of variables that may determine an election result, and these variables can only be understood by studying campaigns. How parties present their performance and promises to voters is key to understanding voter behaviour. Unfortunately, the study of campaigns does not fit neatly in Excel sheets.

Why NYAY failed

This is what explains the failure of NYAY to make the Congress win even a handful of seats in the poorest parts of mainland India. The NYAY campaign was launched just a few days before polling began. The campaign was so poorly executed many voters hadn’t even heard of it when they voted. Many posters and hoardings were in English, for a scheme that targeted the poorest voters. Those who knew of NYAY either did not understand how it would work, or did not believe it was possible, or did not think the Congress was going to win even if they voted for it.

The failure of NYAY was thus not a sign that voters don’t vote for economic well-being. Anyone who travelled out of Delhi in the Lok Sabha elections met voters who said they were voting for Modi because he gave them toilets and pucca houses, just as there were many who said they were happy about the Balakot airstrikes on Pakistan. We don’t need data to know that it was the social spending that won the BJP the votes of the lower castes who are not its core voters.

Views are personal.