Probit regression model can almost accurately predict election outcomes (Comment)

This year -- the year before the 2019 Lok Sabha elections -- was to witness elections in a total of eight states.

Three are already done, and it will be interesting to see how voters' opinions shape electoral outcomes in the remaining five states.

In deciding who to vote for, citizens take into account certain factors which they think of as important in choosing their leaders.

In most democracies, a common belief dominant among politicians is that of all the factors, it is good economics that makes for good politics.

But to attribute an incumbent's successful re-election to just economic growth might be too naive in a complex democracy like India.

There have been numerous studies conducted for the developed nations (especially the US) that primarily seek to prove the importance of economic policies on the popularity of the incumbent party.

In India, however, there is a paucity of such forecasting models for electoral outcomes. This is somewhat explained by the fact that because of the disparities that exist within our regions, and hence among the citizens living in those regions, it indeed becomes difficult to capture the factors that have a role to play in affecting their voting behavior.

Going forward then, an attempt to develop a comprehensive prediction model of this kind must address two empirical questions.

First, whether growth rates based on the economic theory of voting can explain state-level election results in India and, second, what are the other measures that voters might be considering in the process of rewarding/penalising the incumbent party in their state.

The economic theory of voting rests upon the popular notion that the chances of an incumbent party winning elections rise if the region experiences positive economic growth during its term.

However, in India, in 2006, the ruling Dravida Munnetra Kazhagam (DMK) in Tamil Nadu lost the elections with a growth number as high as 13 per cent.

Even more interestingly, Arunachal Pradesh in 2014 chose to re-elect the Congress party despite its poor show in the state in terms of economic aspects.

Moreover, a simple correlation analysis shows that there is a negative relationship between the growth rate of GSDP per capita of a state and the re-election of the incumbent party, implying that in most of the cases higher growth does not lead to the incumbent getting re-elected.

This observation then means that our voters, while ranking the governing capabilities of different parties, also consider a whole lot of factors other than just the economic performance achieved.

The other factors that might contribute in shaping the voting patterns of our citizens can be (a) the level of social progress in a region, (b) the level of consumer and business confidence in the government at the Centre and (c) the level of social media leveraging by political parties.

Using a probit regression model for the period 2006-2017 with the re-election of the incumbent party as the dependent variable and the rest of the factors as explanatory variables, the success rate of the predictions turns out to be 78 per cent.

Perhaps the biggest prediction by way of this model was that Tripura would not be ruled by the Left any more.

For the upcoming elections in Karnataka, the model predicts that the incumbent Congress has only an 18 per cent chance of coming back to power given the conditions that exist today.

Similarly, in the case of Mizoram, it is predicted that the ruling Congress has almost negligible chances of re-election.

However, in Chhattisgarh the ruling BJP is expected to make a comeback. Likewise, in Madhya Pradesh, the BJP has a 90 per cent chance of coming back to power. In Rajasthan, however, it is going to be a close battle since the chances of the BJP again gaining power lies at around 60 per cent.

In the model, the social progress in any state is mapped through using the social progress scores from the Social Progress Index, India, which is positively related to the prospects of getting re-elected.

The index is an integrated measure of the capacity of a society to meet the basic human needs of its citizens, establish the building blocks to allow the citizens to enhance and sustain the quality of their lives and create the conditions for all individuals to reach their full potential.

Furthermore, two different indicators are taken to capture the confidence on the central government -- the Business Confidence Index by the CII and the Nielson Global Consumer Confidence Index.

Lastly, the use of Google trends data (through newspaper articles, television news, or social media platforms) is made to portray the level of interest enjoyed by political parties on the internet.

One of the most striking features of this research is the noteworthy role of media in shaping elections in the country.

Over the years, there has been no less than a revolution wherein the media platforms are not only just a carrier of information but also an influence over the perceptions of the voters about their political leaders.

With the advancement of technology, the reach has only amplified. Clearly, since the voters may primarily be guided by the social development measures and not the economic indicators in their area, it is then for the political parties to focus on garnering interest in their favour by working on the real development issues that people face in today's times.

(Amit Kapoor is chair, Institute for Competitiveness and can be contacted at .

The views expressed are personal. He can be contacted at and tweets @kautiliya.

Manisha Kapoor is a researcher at Institute for Competitiveness, India)



Source: IANS