On a recent 5-day trip to rural Bihar (full disclosure: I am a card-carrying member of the Limousine Liberal group that has been making these fact-finding treks since 1998) caste-based voting is all I heard from political leaders, journalists and self-proclaimed savvy analysts. This concept is currently in vogue amongst all those trying to decipher the outcome of the all-important Bihar Assembly election.
I have never understood the concept of caste-based voting, primarily because it contains zero information about election trends or outcomes. That some people vote on the basis of caste is a cast-iron truth. In the US, almost 90 % of the black population votes Democratic; and it is very likely that 90% of the “rednecks” (politically reactionary, working-class Americans) vote Republican. How does this “fact” affect one’s analysis or forecast? It does not.
Equivalently, assume for a moment that X% of the Yadavs have voted for former Bihar chief minister Lalu Prasad in the past. [You can put any value to the X]. How do we conclude that people are voting according to their caste? Because Lalu is a Yadav? Or because the political candidate is a Yadav? If the latter, then what stops the BJP coalition from putting up a candidate with the Yadav surname? If the former, then, by that logic, all Dalits should vote for the NDA, since Jitan Ram Manjhi is a senior Dalit leader , but neither Nitish Kumar nor Lalu nor Sonia Gandhi is a Dalit.
And all the OBCs (except for Yadavs and Kurmis) should vote for the NDA because PM Narendra Modi belongs to a caste under the OBC category. So, regardless of what happens in the Bihar election, can we please not pursue lazy journalism, and even lazier analysis, by pursuing this theme of caste-based voting?
Interestingly enough, very few of the voters interviewed in my Bihar trip said they would vote on a caste-basis. What do they say they are voting for? Either the development (vikas) agenda of the BJP or the development agenda of Nitish (roads, rural electrification and girls’ education). So, who is going to win given that caste-based voting provides zero information?
Unfortunately, there is very little hard information on which to make a forecast. There are no polls, opinion or otherwise, because our Supreme Court believes that opinion/exit polls can materially affect one’s voting behaviour. (Will internet polls be banned next? Just asking.)
A possible forecast method is to analyse opinion polls conducted before voting started on October 12. There were 11 such polls and they went both ways—4 said BJP would win, and 2 said JD(U)+ would win and 5 predicted a hung assembly. Another forecast can be based on the odds in the satta bazaar (bookies). The odds have been shifting towards a too close to call result with a slight edge to JD(U)+. The break-even betting point today is 110 seats for the NDA, a marginal loss for the NDA (winning number of seats is 122). Finally, an “ankhon dekha haal” forecast can be made, but for most, and especially a wannabe statistician, that is not a reliable method.
From all the methods mentioned, caste cannot be used to make a forecast, opinion polls suggest a tie, and satta bazaar suggests a close fight. The only remaining determinant of forecasts is historical trends. The first historical trend is that any time a political race is too close to call, it usually ends up as a convincing victory for one party. Two recent examples are the elections in UK and Canada. The second historical trend is that of the ruling party losing vote-shares/seats in the first two years post a national election. This is a standard result in the US where the mid-term elections are generally a loss for the national election winner. The trend in India (since 1980) is that in the first two years post a Lok Sabha election, the winning party loses about 6% of the vote-share.
But elections are not determined on the basis of vote-shares alone; it also matters how many major parties are contesting the election. There is a radical transformation in the Bihar election 2015. In Lok Sabha (LS) 2014, there was a multi-party fight, with the BJP united, and the opposition—the JD(U), RJD and INC, equal to JD(U)+—hopelessly split. Compiling the results on an assembly constituency level, the result was: the BJP/NDA won 172 of the 243 assembly segments. If the opposition had been united in 2014 as they are united today, the NDA would have only won 92 seats, and JD(U)+ would have won 145 seats.
In other words, the opposition united effect is to subtract 80 (172 minus 92) seats from the NDA. Each 1% swing against the NDA adds about 8 seats to the JD(U)+ kitty. If the historical average (negative swing for the ruling party) occurs, the NDA tally would likely be reduced to less than 50 seats.
What has to happen for the NDA to win this election? A ‘little over 4%’ swing in its favour, though the intolerant atmosphere is not conducive for this to happen. A large fraction of the voters who voted against the NDA in the Lok Sabha election will now have to vote for it in the assembly election. Lalu’s party won 20.5% of the vote in 2014, Congress obtained 8.6%. Neither is likely to vote for the NDA. A disgruntled JD(U) voter (and disgruntled not because of Nitish’s bad performance, but because of his association with Lalu!) may vote for the NDA, but this number is unlikely to account for much. All in all, the NDA is likely to lose.
One final point on a two-party versus a three-plus-party election. If our forecast of a reasonable loss for the NDA is correct, it is likely that the media buzz will be “Bihar looks like Delhi”. Just look at the data in the accompanying table for the Delhi assembly elections in December 2013 and January 2015. Note, the BJP retained its 32-33 % vote share on both occasions. In 2013, it was a three-way fight with the incumbent Congress pulling in a respectable 24.6 % vote share. In 2015, a three-party fight became effectively a two-party affair. The NDA, which had won 31 seats in 2013, won only 3 seats with the same vote share!
Here comes the forecast: First, the caveats, no information except historical analysis and a visit to Bihar. Both (data and visit) indicate that the people of Bihar seem to be happy about Nitish on at least three counts—the building of roads, electrification of villages, and enhancement and encouragement of gender equality (free bicycles for girls entering high school, etc.). But good governance these days is in precious short supply on the NDA side. All things considered, the NDA should find it difficult to top 80 seats with a reasonable probability of obtaining seats in the 50 to 70 zone. If forced to make a point estimate, I would say 175 seats for JD(U)+ and 60 seats for the NDA.
The author is chairman, Oxus Investments, and contributing editor, The Financial Express