The intuition that a two-party system is a natural consequence of first-past-the-post geographical voting is common, both among political observers in India and political scientists. Among the latter, it is codified as “Duverger’s Law”, which like most “laws” in social science, is an occasionally-observed regularity. Emerging from the assumption that nobody wants to “waste” their vote by voting for a party unlikely to win, it appears to hold in most constituencies across India.

In a recent paper, the political scientists Pradeep Chibber and Geetha Murali of the University of California check this intuition and discover that it actually fails about a third of the time. In a large number of constituencies across the country, more than three parties are competitive at any given time. They provide an explanation for this, arising from the simple assumption that voters care about both state and national-level politics, and thus might be willing to vote for national parties even if they are not directly likely to form a state government. Their assumptions, their three “testable hypotheses”, and how they check them are extracted alongside.

The authors then consider what their analysis implies for certain Indian states. In Tamil Nadu, for instance, there are several regional parties: they show that within most assembly constituencies, however, at most two are competitive at each time. This fits in with prediction (c). In Bihar and UP, however, almost all constituencies have been three-way contests historically: this is not entirely due to ethnic politics, say the authors, because it predates the massive ethnic mobilisation of the 1990s. Since UP in particular has always had more than two national parties, or combinations of national and regional parties competing at the state level, this fits in with prediction (a).

... contd.