Another look at those Indian numbers

The Indian election is done and dusted (see previous report here). The electoral commission has declared the 543 members of the new lower house of parliament, or Lok Sabha (the complete list is here – scroll halfway down to get to the English version), and Narendra Modi, leader of the Bharatiya Janata Party, will be sworn in today as the new prime minister.

But it hasn’t entirely escaped notice that the Indian electoral system, as is its wont, did some peculiar things. So I think it’s worth having another look at the figures to see just how misleading the notion of a BJP landslide is.

The analysis is a bit tricky because the official website doesn’t seem to report an overall vote total, and there are obviously a lot of votes for small parties that it doesn’t list (of the order of 20 million). There are also about six million votes for “none of the above”, which I’ve basically ignored. But anyway, making a few assumptions, and supplementing the official numbers in a few places from Wikipedia, here’s what I get:

The last two columns are the seats that, by my calculation, would have been won under a proportional system, allocated by either a Sainte-Laguë or a D’Hondt method. (This is a good demonstration of how similar the two are, but how D’Hondt slightly favors the larger parties.) In any actual proportional election, voting patterns would look different, since both parties and voters would change their behavior, but the experiment is still instructive.

The difference between the proportional version and the actual result can fairly be described as dramatic. The BJP won a majority with just 31% of the vote; Congress won less than half the seats to which it was proportionally entitled. The pattern of representation of smaller parties is quite different, with some greatly under-represented and others over-represented – based mostly on how much their vote is concentrated, with a decent admixture of luck.

Interestingly, though, proportional representation wouldn’t mean a big difference in the number of parties that were represented. In fact D’Hondt would give exactly the same number, 35, while Sainte-Laguë would give a modest increase to 40. But by dramatically scaling back both the BJP and the regional parties it would make it much harder to put together a majority.

Opponents of proportional representation would take that as an argument for their side, saying that it would make India ungovernable. But as the above table demonstrates, the single-member voting system does much more than artificially construct a majority so as to provide (in theory) stable government. It also introduces quite arbitrary unfairness to the system, particularly disadvantaging parties whose support is spread across different regions – just the sort of thing you might think India needs more of.

Of course, in politics as elsewhere, perception tends to become reality. Modi won a landslide not because that’s what the numbers say (since they don’t), but because that’s what the media and other observers say. He can’t be blamed for playing by the rules as they actually are, not how they might more ideally be.

But it would be good if it were more widely appreciated just how much those rules determine the result.