The relationship between economics and politics

As election outcomes are apparently random, elected politicians have no incentive to deliver good economic performance on their watch

ecent political developments in India, in particular the choice of Yogi Adityanath as the chief minister of Uttar Pradesh after the Bharatiya Janata Party’s (BJP’s) thundering victory in that state’s recent election, has again cast light on an evergreen question in the Indian political economy: What is the relationship of economic performance and a party’s electoral prospects? Does good economics help assure, if not guarantee, an incumbent party’s re-election chances, or do these hinge rather on non-economic considerations, such as the pursuit of a social or cultural agenda?

Unfortunately, fables, morality tales and armchair narratives have typically substituted for evidence-based analysis of these questions. Thus, it is frequently asserted that the BJP-led government lost the general election in 2004 because of a backlash from poor and rural voters against its triumphalist “India Shining” campaign. Likewise, it is asserted that it was the Mahatma Gandhi National Rural Employment Guarantee Act (MGNREGA), appealing to the selfsame rural poor, which helped win re-election for the Congress-led government in 2009. As it happens, neither of these narratives can be confirmed by data.

Thus, on the former, as Rupa Subramanya and I argued in our 2012 book, Indianomix: Making Sense Of Modern India, the most parsimonious explanation for the BJP’s defeat in 2004 is that the close election was a coin-toss and they came up the losers. Any specific hypothesis to explain their defeat—whether India Shining, middle-class apathy, poor alliance choices, or old-fashioned anti-incumbency—could well hold some truth at the margin but none of them can be conclusively established by data as the sole, or even the most important, explanation. Likewise, a data analysis of the 2009 election by political economist Praveen Chakravarty shows that neutralizing for all other effects, most crucially for pre-poll alliances, there is no statistically meaningful relationship between the Congress vote share and the level of MGNREGA spending across constituencies. Similarly, there is no statistically significant evidence that the Congress vote share went up between 2004 and 2009 in constituencies with higher MGNREGA spending. This data analysis thus busts yet another morality tale.

The reality is that when hundreds of millions of people vote in a complex election with many issues at play, there is an ineluctable element of randomness in the final outcome, which aggregates across the preferences of all these individuals. The attempt to anthropomorphize all these hundreds of millions of differently motivated individuals into one representative voter with a specific motivation is doomed to failure.

Debunking simplistic narratives is one thing, but does a rigorous, data-based analysis show any relationship, in the aggregate, between good economic management by incumbent governments and their subsequent re-election prospects? Here, too, evidence is mixed. For instance, a recent Mint analysis by Pramit Bhattacharya and Tadit Kundu (“Does Good Economics Make For Good Politics In India?”, 24 March) analyses all the assembly elections after 2001 in 18 major states, dividing the states into three buckets: moderate incumbency, strong incumbency, and anti-incumbency. In the former two categories, there is some correlation between incumbency and higher growth rates during the incumbency period, while the relationship is weaker in the anti-incumbency states. Of course, correlation does not prove the existence of a causal relationship, suggestive though it may be.

Other recent research, cited in the Mint analysis referred to above, also appears to suggest India’s fabled anti-incumbency may be weakening, with incumbents more likely to be rewarded for good economic stewardship in the past decade than was the case in previous years. Yet such findings are very fragile in a statistical sense.

Thus, the more basic point, which I argued some time back (“Do Indian Voters Really Care For Economic Growth?”, Mint On Sunday, 29 November 2015), is that selection of data, time period and methodology can yield very different and even confounding results.

Thus, Chakravarty analyses assembly elections from 2004-14 in the 12 largest states—a different data set than in the recent Mint analysis—and finds no statistically meaningful relationship between economic performance (as measured by economic growth during time in office) and an incumbent government’s re-election prospects.

It is tempting but ultimately unhelpful to draw the nihilistic conclusion that as election outcomes are apparently random, elected politicians have no incentive to deliver good economic management (which hopefully translates into good economic performance) on their watch. For, if it is difficult to prove that good economics is good politics, it is equally difficult to prove that bad economics is good politics. In other words, once elected, a politician of conviction, freed from the notion that there is some strong provable relationship between what they do while in office and their re-election prospects, may be guided to do what they believe is right, simply because it is the right thing to do.

The relationship between good economics and good politics may, ironically, come down to morality in the end.