Some revision of the thesis the other day is in order, now that I have had a chance to look at the district-level results (as part of a project with Royce Carroll). Again, these are preliminary results, but final results are unlikely to be substantially different.
A key point of my earlier thesis was that this was less a smashing electoral mandate for Modi than it was a product of strategic alliance-building by the BJP and fragmentation of the rest of the field. The BJP, after all, won a majority of seats on a mere 31% of the vote. I think most of that thesis remains accurate, but should be modified in one crucial respect now that I have seen the district-level patterns.
First of all, the BJP’s biggest advantage was not that it faced fragmented opposition, but simply that it had a highly efficient distribution of its votes. Had fragmentation been the key to its success, we would expect it to have won many seats with under 45% of the vote, maybe under 40%. In fact, its mean winning percentage was 48.8%. If anyone benefited from fragmentation, it was the Indian National Congress, whose mean winning percentage was only 42.8%. (The difference is statistically significant–easily.) The BJP averaged 32.6% where it came second, so it is not as if it lost many close races. In fact, out of the 54 seats where it came in second, there were just ten seats in which the BJP had over 40% of the vote. (Congress averaged 33.0% where it came second; that was 223 districts, including 44 where it had over 40%. In percentage terms, that’s about the same failure rate in 40%+ districts.) It seems that the BJP put its resources into winnable seats and thus wasted few votes outside of places where it could harvest seats.
[Paragraph that contained an incorrect statement about the 1999 national result removed.]
Moreover, as I said in the earlier post, alliances were indeed critical. It is impossible to know how many fewer seats the BJP might have won without stand-down agreements with parties like Shiv Sena (Maharashtra) and Telugu Desam Party (Seemandhra), but these and other parties’ voters gave votes to BJP candidates that could have been decisive in many seats in their respective states.
The efficiency of both the BJP and its ally is really on display in Maharashtra. There, the BJP contested 24 districts, and averaged 54.6% of the votes; it won 23. Its ally, Shivsena contested 20, averaging 50.4%, and won 18. In those 18 seats won by Shivsena, Congress was the runner-up in 8 (averaging 34.4%) and its ally Nationalist Congress Party was runner-up in 7 (averaging 33.1%). The results are remarkably stable across the alliance pairings.
All in all, the victory is a combination of a very favorable swing (the BJP had won less than 19% in 2009), an organization that allowed it to target seats where it could win close to a majority of votes and not spread itself thinly around the country, and assistance from key regional allies.