Active Stocks
Thu Mar 28 2024 15:59:33
  1. Tata Steel share price
  2. 155.90 2.00%
  1. ICICI Bank share price
  2. 1,095.75 1.08%
  1. HDFC Bank share price
  2. 1,448.20 0.52%
  1. ITC share price
  2. 428.55 0.13%
  1. Power Grid Corporation Of India share price
  2. 277.05 2.21%
Business News/ Politics / Policy/  Bihar polls: Straddling a two-party contest
BackBack

Bihar polls: Straddling a two-party contest

Pollsters are collectively overestimating chances of a close contest between the NDA and the Janata Alliance

Voters queue to cast their ballots at a voting centre in the village of Mahmoodpur in Samstipur district on 12 October. Photo: AFPPremium
Voters queue to cast their ballots at a voting centre in the village of Mahmoodpur in Samstipur district on 12 October. Photo: AFP

As Bihar starts voting in a protracted six-phase election, it’s a good time to take stock of all the opinion polls so far and see if there’s any collective wisdom that can be gleaned.

For example, we had seen back in 2013 that while most exit polls in Karnataka had vastly differing forecasts in terms of seat shares, they pretty much agreed on the vote shares of different parties.

Taking these vote share forecasts as “consensus forecasts", Election Metrics was able to predict correctly that the Congress was going to get a comfortable majority, which it did.

A pseudonymous blogger who calls himself “politicalbaaba" has compiled results of all opinion polls on Bihar published so far, and one thing that is clear from these polls is that the contest is a two-horse race between the National Democratic Alliance (comprising the Bharatiya Janata Party, Ram Vilas Paswan’s Lok Janshakti Party and Jitan Ram Manjhi’s Hindustan Awam Morcha) and the Janata Alliance (comprising the Janata Dal (United), the Rashtriya Janata Dal and the Congress).

Yet, despite these polls being rather recent, we see that we still don’t have too much of a consensus on forecasts, though several of them predict that both alliances should poll around the 40% mark (notable exceptions being the polls conducted by Zee News and CNN-IBN, with the latter’s poll having a basic flaw as exposed by political insights consultant Subhash Chandra.

These numbers are not very different from what pollsters forecasted when opinion polls started around August, indicating that either there has been no significant shift in terms of voter preferences or that polling methods have not been able to capture such a shift.

Election Metrics has mentioned repeatedly that the biggest challenge in Indian election forecasting is the conversion of vote share forecasts to seat forecasts, a task that is made difficult by local factors and India’s first-past-the-post system.

As we have mentioned before, what we need for this conversion is some kind of a model (such as using data from previous elections and constructing swings, for example; Bayesian methods can also be used for this). From the forecasts above, however, what is evident is that most pollsters have used an exceedingly simplistic model.

In its last edition, Election Metrics argued that these Bihar elections are going to be hard to forecast. There has been a massive redrawing of alliances, because of which there is no good “prior dataset" for a model to be built upon. This might explain the fact that the seat predictions above closely mirror the vote share predictions for the two major alliances.

In this regard, it is interesting to note that the four polls that predict a close contest as far as seats are concerned also show a close contest on vote share. The two polls that forecast decisive victories (one each for each alliance) do so on the basis of their polls that forecast decisive differences in vote share.

If we eliminate “others" and look at the vote share and seat share forecast for the NDA from each of the polls, what we see is that they all line up (see chart).

Election Metrics’ view based on these predictions is that pollsters are overestimating the probability of a close result. The only pollster who is showing a decisive victory is Zee News, but that is on the back of a poll where the NDA has a 14 percentage point lead over the Janata alliance.

Even CNN-IBN’s prediction of a Janata victory is not decisive enough compared to the degree of vote shares that they predict (notwithstanding the error Chandra pointed out).

This is a topic that Election Metrics has had to keep coming back to: in a two-way contest, an election that is close in terms of the vote share can be extremely one-sided when it comes to seats.

We have had several elections in the past where a party has got less than 20% of the seats despite getting over 40% of the votes. In fact, based on past elections, equitable distribution of votes in a two-way contest have seldom led to an equitable or close distribution of seats.

By predicting a close election (in terms of seats) based on opinion polls that show comparable vote shares for the two alliances, pollsters are either refusing to recognize that close contests can lead to one-sided outcomes or not showing enough confidence in their polls and so playing it safe (they can’t be faulted for doing this, for it is difficult to draw too much information from a poll that shows a two percentage point difference between leading competitors, and pollsters don’t like to give large error bounds).

If it were possible to trade in prediction markets on the outcome of these elections, Election Metrics would recommend a “long volatility" strategy.

Opinion polls are collectively overestimating the possibility of a close election, and money can potentially be made betting against this prediction, if the markets exist for making such money that is.

PS: The “straddling" in the headline refers to this betting recommendation—a good way in financial markets of going long on volatility is.

Unlock a world of Benefits! From insightful newsletters to real-time stock tracking, breaking news and a personalized newsfeed – it's all here, just a click away! Login Now!

Catch all the Politics News and Updates on Live Mint. Download The Mint News App to get Daily Market Updates & Live Business News.
More Less
Published: 13 Oct 2015, 12:52 AM IST
Next Story footLogo
Recommended For You
Switch to the Mint app for fast and personalized news - Get App