Cast(e)ing aspersions

As long as there have been parties in India that have organised themselves around caste or religion...

As long as there have been parties in India that have organised themselves around caste or religion, there has existed the suspicion that those parties are somehow lowering the level of politics, that they care less about ethics and governance and more about patronage. Underlying much of the commentary about payoffs to MPs before this week’s vote of confidence was precisely this belief: some parties are organised along ethnic rather than ideological lines, and thus their legislators are less committed and more vulnerable to ‘bribery’. (Prakash Karat and, apparently, some members of the RSS both made this point.)

The question of whether there is evidence of this hardly ever comes up. How could such a thing be tested, in any case? Two economists at Harvard and MIT, Rohini Pande and Abhijit Banerjee, have done so in some recent work. They first made some simple assumptions: for example, that voters care both about the identity of a candidate and the candidate’s ‘quality’. They also assumed that elected legislators have some discretion, that they are not, for example ‘slaves to a common minimum programme’ of some sort, to use the hopeful recommendation of the general secretary of the CPI (M). This is important because if automatons were elected that mindlessly followed party orders. voters wouldn’t actually care about the character of the candidate.

If groups of voters then begin to care more about caste or community, such as has happened across India in the past quarter-century, Banerjee and Pande demonstrate that the relative quality of candidates that win goes down. This effect varies from party to party, however: in particular, successful candidates from parties that represent ‘majority’ communities, castes or ethnicities tend to be of lower quality than earlier, while those from parties that represent ‘minorities’ are better than before.

Story continues below this ad

This is actually quite straightforward: if people care about both quality and ethnic parochialism, each voter trades off one against the other and chooses the candidate who combines the two qualities in the proportion they most prefer. If they start to care more about ethnically targeted policies, then a party associated with a particular ethnic group will have to field a better candidate than earlier in order to receive votes from members of another ethnic group; and, conversely, that they will be able to retain votes from their own ethnic groups even with a lower-quality candidate than before. There will thus be considerable differences in overall quality between parties in terms of successful candidates, and parties representing ‘majority’ social coalitions that will be relatively more corrupt.

Is this borne out by the facts at all? Measuring the ‘quality’ of people, even of members of parliament, is hard. The way the authors of this paper have got around it is by asking people who already know: politicians and journalists. They looked at a hundred constituencies in UP, and in each asked four local experts to rank losers and winners in the 1980 and 1996 elections in terms of corruption and criminality, both subjectively and objectively; they asked questions about petrol pump allotments and personal wealth. They also asked questions about criminal records, and verified all the answers.

They discovered that in constituencies with a higher proportion of SC or OBC voters, successful candidates from SC- or OBC-based parties were more corrupt in 1996 than 1980 compared to the politicians they beat, and vice versa. In other words, as ethnic bloc-voting gained power in Uttar Pradesh, parties that catered to specific ethnicities did indeed discover that those that cared about ‘social justice’ expressed through the ballot box seemed to care less about good governance — and took advantage of that by fielding people of lower quality wherever they could. To check that this effect was coming from people voting for caste identification, and not from some other source, the authors looked at a smaller sample, made up of those seats reserved for SC/ST candidates: there was absolutely no change in relative corruption when both winners and losers were forced to be of the same caste background.

What does this imply for the future? Banerjee and Pande, quite properly, don’t take a position on it. In a world, however, where parties once associated with specific ethnic groups are creating coalitions and increasingly broadening their support base, and where many believe the BSP might well be reconstructing the invincible pre-1975 Congress coalition of Brahmins, Dalits, Muslims and Jats, we can be cautiously optimistic. However ethnically polarised the electorate, parties that are strategically changing their choice of candidate to include other groups will be forced to consider candidate quality. After all, the data do not actually back up, at all, the belief that voters cared less about quality over time: in actual fact, once all caste equations were taken out, and the data were looked at for whether candidates less corrupt found it easier to win, the answer was an unambiguous yes.