Seeking redemption in UP 2017

Many expect the Uttar Pradesh election to go the Bihar way — a big loss for BJP. However, the same math suggests the BJP is likely to win a comfortable majority in the state

It was Bihar 2015 that started the rout — of pollsters, that is. There was near-universal expectation by politicians, pollsters, opinion and exit polls that the BJP, 18 months after its resounding Lok Sabha victory, would romp home in majority glory. It was not to be. Almost the entire class failed. Ditto for the unexpected Brexit vote; ditto squared for the US election. I have been forecasting elections since the late 1980s, often as a hobby, and sometimes as a part-time profession. I was lucky to get the Bihar election almost spot-on right, that is, I had said that the BJP+ would get 60 seats, not 160 seats as most were predicting. BJP+ obtained 58 seats.

But I did get the US election wrong, and wronger than most. I had forecast that Hillary Clinton would win by a landslide, and if it weren’t for the unexpected college-educated white women’s “support” for Trump, I, and practically the whole world, would have been right in forecasting Clinton as President. But that analysis waits another day — maybe if redemption is received in getting the UP forecast right!

UP 2017 is being discussed in the same breath as Bihar 2015, which is as it should be. They are the two largest states that are part of the Hindi-Hindu heartland, and both important for the long-term success of the party at the Centre. Amidst much fanfare and discussion, sworn enemies Nitish Kumar and Lalu Yadav joined forces in Bihar 2015, and the Congress happily pitched in for the ride. This unity party was appropriately termed MahaGathbandhan (MG) or “Grand Alliance”.

However, UP 2017 is not Bihar 2015, for the simple reason that a major third party — the BSP — has been left out of MG, so the opposition is left with just a G: SP plus Congress. Before proceeding with the analysis, a bit of background history of UP, in terms of vote shares, is relevant. The BJP won a plurality of the votes in three state elections (1991, 1993 and 1996); average vote share, with very little volatility, was 32 per cent. In the 2012 assembly elections, their vote share was a low 15 per cent — which catapulted to, in a space of two years, 43 per cent in the 2014 Lok Sabha elections. Meanwhile, for the 21 years 1993 to 2014, the SP and BSP have both averaged around 25 per cent, regardless of whether it was a state or national election.

The first conclusion that emerges is that there is a core vote for both the SP and BSP, of around 20 to 25 per cent, and a declining core vote for the Congress, possibly now in single digits. The second conclusion is that something happened in UP in 2014: That something was the emergence of Narendra Modi as the leader of the BJP. The big question for all analysts is the following: Does one take the 2012 state election results as the base, or the 2014 national elections as the base? The wide variety of forecasts we are witnessing in UP, and only in UP, is that some analysts are choosing 2012 as the base and concluding that UP will go the Bihar way, while others, who are taking 2014 as the base, are concluding that UP is for the BJP to lose.

We believe that the latest information is always preferable, unless there are strong reasons to reject this very reasonable assumption. In addition, given the sharp jump in the BJP vote share, it is a bit hard to imagine that the UP voter will, in a wholesale fashion, revert back to 2012 — though some loss in the 2014 BJP vote share is likely.

For the 2017 UP election, the SP and Congress have formed an alliance. But unlike Bihar, a major third party, the BSP, is not part of the alliance. Bihar was an “alliance math election”, as UP will likely turn out to be. This math states that if the 2017 votes are a replica of the 2014 vote, a Bihar-like MG (SP + BSP + INC) would result in a resounding victory for the MG — 263 seats. But 2017 is a three-way fight — BJP+, BSP, and G — and three-way fights should not be confused with two-way match-ups. Going from MG to G, the “alliance” is able to win only 78 seats with the BJP winning 317, and the BSP winning only eight seats.

So, our first major conclusion is that UP ain’t Bihar — and most so because of the defining math of a three-party election. How big the BJP victory is in UP will depend on how much vote share they lose relative to 2014: If the municipal elections in Odisha and Maharashtra are to be believed, the BJP is likely to gain in vote share. In addition, notebandi has turned out to be a vote-getter for the BJP.

However, there are two strong statistical factors arguing for a decline in the BJP vote share in UP, rather than the constancy assumed above. First, they scaled historic highs in 2014. Second, the ruling party at the Centre has historically lost around 5 per cent of vote share in subsequent assembly elections. If this loss is imposed (with equal 2.5 per cent gains to both SP+ and BSP), the seat shares which emerge are BJP+ 254, SP+ 127 (with INC 21 seats) and BSP at 22 seats.

The table shows that we are wide off the mark in UP, according to both conventional wisdom and the opinion polls. We have been there before, but we do realise that while we can’t get it right like Bihar, we shouldn’t get it wrong like Clinton. There is security in numbers — and forecasts. Hence, the table shows what our models, and method, suggest for the remaining four states going to the polls.

In Punjab, we get the SAD-BJP alliance losing badly; however, it is relevant to note that while SAD is on a declining trajectory, BJP actually won six more seats in 2014 than in 2012 (18 vs 12). In contrast, SAD went from 56 seats in 2012 to 33 seats in 2014. But unless the entire decline in the SAD+ vote goes to AAP, which is unlikely, the Congress, while losing everywhere else, should win.

Uttarakhand and Goa — it is difficult to not see the BJP winning. Manipur, however, might spring a surprise. The opinion polls are suggesting so, and given the notebandi surge for the BJP elsewhere in the nation, and the belief among voters across India that Modi is a man of action and vision, it would be foolish to rule out a strong BJP showing.

So, what does it all add up to? We don’t know. But what we do know is that some of the favourite explanations for how India votes are not entirely accurate: For example, elections are determined by caste voting. That is always true — except when it is not true! Above 85 per cent of blacks vote Democrat in the US, but no analyst, and not even a lazy journalist, has ever stated that the black vote is determining any election. Analogously, a rock-solid percentage of the SC vote is for the BSP, but did the SC vote cause Mayawati to win in 2007?

Another change in Indian electoral behaviour is that the assembly vote is also a referendum on the leadership at the Centre. Regardless of whether my forecasts are accurate or not, I do feel confident that the identification of these mega-trends in Indian elections is accurate.