Grofman, Owen and Collet (1999) study the relationship between turnout and electoral outcomes in the United States for over half a decade and record a negative correlation between increased turnout and electoral prospects of the party in power. I have calculated the same for Indian elections held between 1980 and 2012 and find similar results. I find that a 1% increase in turnout is associated with reduction in incumbents vote share by 0.3%. I stress on the word correlation, as mere existence of correlation does not mean causation. It is possible that high level of dissatisfaction with the government of the day leads to increased turnout and not the other way round. Hence, if the voter dissatisfaction is widespread, then even with lesser turnout same result would have ensued. Low turnout, on the other hand, may be driven by lack of effective choice between candidates. In 1968, George Wallace, a presidential candidate in the US belonging to a party other than the two major ones, the Democratic Party and the Republican Party, famously remarked that Democrats and the Republicans were as close as Tweedledum and Tweedledee. In such cases, lacking effective choice, the voter decides to sit out rather than exercise her franchise. Thus, the lack of effective choice may be driving low turnout as well as pro-incumbency. Therefore, it is very difficult to claim, based on this evidence, that high turnout leads to anti-incumbency or the other way round.
Hanzford and Gomez (2010) adopt a clever approach to establish the causal mechanism that connects turnout with real electoral outcomes. They look for a variable that has an impact on voter turnout but has no connection with the popularity or otherwise of the incumbent government. They find one in the weather conditions existing on the voting day. Adverse weather adds to the disutility of voting. If the changes in the disutility are uniform across the voting population, then adverse weather is likely to dissuade a number of not so regular voters from voting. Pleasant weather on the voting day is likely cause the opposite. The neat part of this research design is that weather has nothing to do with popularity of the incumbent. They examine the US Presidential elections held between years 1948 and 2000 in nearly 2,000 non-Southern counties. Using turnout predicted by weather, they show that higher voter turnout is indeed anti -incumbent. A four point change in turnout costs approximately 20 Electoral College seats to the incumbent President in the US.
However, research also highlights that the impact of high voter turnout is conditional. Denardo (1986) argues that impact of increased voter turnout depends on the partisan composition of the voters of the region where there is an increase. He postulates that an increase in turnout in a Republican-dominated area is not good news for Republicans as the peripheral voters are less ideologically driven and have higher tendency to switch sides. Hanzford and Gomez (2010), empirically confirm these findings. Thus, going by these findings, the BJP should be happy if the turnout is higher in states such as Uttar Pradesh, Bihar and Odisha; the Congress, on the other hand, will be better helped in case of higher turnout in states such as Madhya Pradesh and Chhattisgarh. I deliberately do not include Gujarat as turnout may be influenced by the prospect of having a prime minister from the state. If we apply these findings to recent assembly elections held in
India, the results are striking; Rajasthan, Delhi and Chhattisgarh saw high increase in turnout and in two of the three states, incumbents lost heavily. The ruling dispensation had a narrow escape in Chattisgarsh. A closer look at the voting pattern of Bastar region of Chhattisgarh strengthens this argument further.
Madhya Pradesh, on the other hand witnessed little change in turnout and also a strong pro-incumbency wave. Another interesting point brought out by the above research is that increased voter turnout makes elections less predictable. This arises because of the malleability of the peripheral voters.
The above research findings have four implications for the current Indian context. First, as predicted by almost all opinion polls, there seems to be a strong anti-incumbency effect in the current election. Second, some political parties are committing a tactical error by raking up emotional and ideological issues at this stage, as the peripheral voter who is contributing to the increased turnout is likely to be less ideologically-inclined. Parties would do well by concentrating on day-to-day issues that bother a common voter. Third, voters are not seeing this election as a regrettable choice between Tweedledum and Twedledee, and may also be casting a positive vote. Fourth, the actual results may show much more variability than what has been anticipated. We will have to wait till May 16 for a clear answer.
The author works for Centre for Analytical Finance at the Indian School of Business