We have just come out of a series of state elections, where it is fair to say that the opinion polls as a group got it wrong. Over the last few years, there have been numerous instances both in India and abroad, where the opinion polls have got it wrong. The just concluded elections was only the most recent instance of a growing list of inaccuracies. Well, in that case is there a better way of understanding the election process and its outcome? Perhaps the financial markets, which as the saying goes is adept at discounting the future, might have something to offer.
In 1998, University of Iowa devised an unique futures market called the Iowa Electronics Market. The experimental market, was designed to provide insights into the behavior of traders and also the processes in play in naturally occurring markets. The design gave it another practical use: it could be used to predict future events such as election outcomes and Federal Open Market Committee voting.
Iowa professors did a study of their exchange’s results over a 20-year period and found it to be accurate 74 percent of the time 100 days from an election, defining “accuracy” as being more precise than the polls. In other words, the electronic market prediction was more accurate than the opinion polls in almost three quarters of the cases.
In India we donot have a Electronics Prediction market, nor I reckon would something like it be permissible under existing laws. But we do have the functioning markets, stocks, commodities, gold etc to serve as a proxy. It would be interesting to take a look at the record of the stock markets in predicting election outcomes.
In order to do this, we would need to adopt a metric for interpreting the market’s message. It is almost a cliche that markets hate instability. So we could use that as a metric to fathom the markets’s message. If the market believes that the election verdict would lead to instability, or change in status quo, then the market would fall. If on the other hand, the market believes that the election verdict would result in stability, then it would rise. This would have to happen before the election results were declared and not after for it to have any predictive value. Indeed, the predictive value would be enhanced, if the prediction diverged from the opinion polls. In other words if the opinion polls predicted the victory of one party and the market differed in its opinion and accurately predicted an alternate, its predictive powers would be burnished.
So with this in the back ground lets look at the last 4 general elections in India, the ones held in 2004, 2009, 2014 and 2019. In those four elections, there were two instances where the prediction of the opinion polls differed from the market reaction and in both those instances, the markets proved to be right. Read on to know which two elections those were. In the other two the opinion polls and the markets were in lock step and in none of them did the markets get it wrong. Hence the ability to read the message of the markets might be a better way of understanding election outcomes than opinion polls. This was the case in the just concluded assembly elections as well. While the opinion polls were giving the opposition the advantage in certain states, the markets remained firm and in fact started rising in the run up to the elections signalling perhaps that the market did not foresee any change in status quo.
General Elections 2004
The general elections of 2004, were held in four phases between 20 April and 10 May. The counting began on 13 May. The opinion polls coming into the elections were quite sanguine about a BJP (NDA) victory. The exit polls did show a slight swing towards the Congress (UPA) but still went with a NDA victory.
Well what was the verdict of the markets. We will take the Nifty 50, one of the leading benchmarks as a proxy for the markets. We looked at the Nifty over a 45 day period, starting from April 1, 2004 upto May 12, 2004, or the day before the results were announced and the BJP conceded defeat. We have divided the period into two, the first from April 1, 2004 to April 19, 2004, the day before the commencement of the first phase of the elections and the second period from April 20, 2004 to May 12, 2004 the day before the election results became know.
| Period | Nifty 50 Returns |
| April 1 2004 – April 19 2004 | +1.3% |
| April 20 2004 – May 12 2004 | – 7.2% |
So in the first phase leading upto the elections, the markets were flat, as if brooding and unsure. Then as the elections commenced, the markets started selling off, in anticipation of an adverse out come. Markets don’t sell off 7% if it is anticipating any form of continuity. Recall that the opinion polls at this time were projecting a resounding victory for the BJP.
The market view diverged from the opinion polls and the market view proved to be correct.
General Elections 2009
In the general elections of 2009, elections were held in five phased from April 16 2009, to May 13, 2009. The counting began on May 16. The opinion polls coming into the elections by major agencies gave the UPA an edge over the NDA, but none were predicted to get absolute majority. The opinion polls reckoned that regional parties would get a significant number of seats and would have a decisive say in government formation.
Now lets take a look at what the market was doing during this period. Here again we are looking at the period from April 1, 2009 to May 15, 2009. We will divide the period into two parts, the first from April 1 2009 to April 15, 2009, the day before the commencement of the elections and then from April 16 2009 to May 15, 2009.
| Period | Nifty 50 Returns |
| April 1 2009 – April 15 2009 | +15.3% |
| April 16 2009 – May 15 2009 | + 5.4% |
In the first phase, the markets rose 15% in 15 days, which is rising at the rate of 1% a day. The markets totally discounted the notion of a fractured verdict that the opinion polls were indicating. In the second phase, the markets rose another 5% during the month that the polls were underway. No sign of any nervousness. The markets returned 20% in a month and a half, something that is not very common. It would be fair to say that there was a blazing bull market at a time when the country was apprehensive of a indecisive election verdict.
As it turned out, the markets were again right and the UPA went on to form the UPA II government which was a stable government which served out its 5 years. In this case as well, there was a divergence between the markets and the opinion polls and the markets once again trumped the polls.
General Elections 2014
The 2014 general elections were held in nine phases form 7 April 2014 to 12 May 2014. The counting began on May 16. In these elections, the consensus of the opinion polls was that the NDA would get a majority and form the government after a ten year spell pf the UPA government.
This was the most polled elections till date and the trend in the polls from a year and a half preceding the polls is in the table below. From Jan 2013, the NDA had a lead which increased over time.
The exit polls published after the last phase of polling was concluded, confirmed the trend picked up by the opinion polls and gave clear lead to the NDA.
The actual results were declared on May 16, 2014 and the NDA got 336 seats beating almost all the polls. The opinion polls got the winner right but underestimated the magnitude of victory.
So how did the markets perform during this period. Since the polls were conducted over a period greater than one month, we will look at the period from April 1, 2014 to May 15, 2014 as one composite period.
| Period | Nifty 50 Returns |
| April 1 2014 – May 15 2014 | +10.5% |
The markets posted solid gains in the month and a half during which the 9 phase of polling was conducted. In this instance, and unlike the last two times, both the markets and the opinion polls got it right. Though the market gain was half that achieved during the same period in 2009, it nevertheless portended the coming of a stable regime.
In the elections of 2014, both the markets and the opinion polls got it right.
General Elections 2019
Like the elections of 2014, the elections of 2019 was long and drawn out. It was held in seven phases form 11 April 2019 to 19 May 2019. The counting was on 23 May 2019.
The extended period of polling was not the only similarity this election had with the earlier one. The opinion polls were overwhelmingly one sided with the only matter of contention being the extent of BJP’s majority. There were only a sprinkling of polls that were predciting a hung parliament.
Opinion and Exit Polls
In the backdrop of the overwhelming consensus about election results, lets take a look at what the markets did. As was the case with the 2014 elections, we are not bifurcating the period under consideration. We will be looking at the period from April 1, 2019 to May 22, 2019 which was the day before the election results were announced.
| Period | Nifty 50 Returns |
| April 1 2019 – May 22 2019 | +1% |
The market reaction in this case was interesting, The markets essentially did nothing. There could be differing interpretations on this but the most cogent one is that since the results were essentially a continuation of the past and this was widely known, the markets did not consider it to be a significant event. But in any event the market was correct in its assessment that there was nothing untoward on the cards.
It would therefore appear that we, including pollsters, would be better off looking at the markets instead of the polls to get a better view of the electoral outcome. It would be useful to bear this in mind while considering the outcome of the 2024 polls.
Author: Soumitra Sengupta
(Director: Lamron Analysts)