In an assessment of the quality of India’s implementation of anti-poverty programs in 1985, then Prime Minister Rajiv Gandhi famously said: “For every rupee spent by the government for the welfare of the common man, only seventeen paise reached him.” That state of affairs was one of the motivations for the 1993 passage of the 73rd amendment, which decentralized the implementation of government anti-poverty programs to local governments. By mandating elections at the local level and providing real, but limited, discretion over the selection of beneficiaries to elected village council presidents (sarpanch), the Indian state entered into a massive experiment to test the hope that bringing government closer to the people would increase the chance that policy benefits intended for rural poor reach them in practice. Two decades later, can we say that decentralization has encouraged pro-poor targeting?
Despite improvements in recent years, research shows that targeting of anti-poverty program benefits are much less pro-poor than we should expect from the policies on the books. For example, consider the allocation of Below Poverty Line (BPL) cards, which are a requirement for eligibility to a wide range of subsidies for the poor from grain to healthcare. A 2009 EPW article found that two-fifths of Below Poverty Line (BPL) cards, a requirement for eligibility to a wide range of subsidies for the poor, were in the hands of the non-poor while majorities of those in abject poverty did not receive a BPL card. While these results cannot tell us whether we should blame local politicians, bureaucrats, or the system for identifying the poor, the broad conclusion of recent research is that gram panchayat (GP) presidents, or sarpanch, use their discretion over the allocation of BPL cards and policy benefits to enrich themselves, reward their supporters, or favor their co-ethnics or family members. This provides evidence for the view among academics and policy-makers that anti-poverty programs in India tend to be captured by local politicians or influential rural economic elites who do not prioritize the welfare of the poor.
This pessimistic view of local politicians has serious policy consequences that may rest on an inappropriate diagnosis of the problem. First, central governments in developing countries like India lack the resources needed to provide benefits to all poor families or the capacity to ensure (through government monitoring, for example) that only the poor receive anti-poverty benefits. This means a substantial degree of local discretion by bureaucrats or politicians is inevitable. Second, we cannot infer local politicians’ preferences or targeting strategies from policy outcomes such as BPL card allocation because sarpanch have limited powers over allocation along with politicians and bureaucrats at various levels of government. As a result, policy-makers would be wise to throw in their hats with local leaders who hold pro-poor preferences in the first place.
Consistent with a more optimistic view of local democracy in India, Neelanjan Sircar and I demonstrate, in an August 2015 CASI working paper, that voters in poor rural settings use free and fair local elections to select politicians with preferences for targeting the poor. In addition, we argue that local elections in an ethnically-diverse setting like rural India favor those with broad-based targeting preferences rather than those known to favor members of the same caste or narrower kin group.
It is also important to note that what we call distributive preferences refers to the types of voters sarpanch favor when it comes to limited benefits such as cash or housing benefits.Unlike analysis of policy outcomes, we are concerned with the intrinsic preferences leaders have over potential beneficiaries. To measure this, we included a laboratory game in a 2013 survey across eighty-four low-income GPs. In this game, sarpanch were asked to divide five tokens in any amount they wished, which would affect the winner of a 200 RS prize. Since each voter received one lottery ticket at minimum, and allocation of the lottery prize was kept secret, our measure is crucially unaffected by social pressure or electoral strategy.
Now we can consider why local elections in poor villages lead to the election of pro-poor local leaders. First, unlike state and national elections, this is a very high-information setting. Voters and candidates for sarpanch know each other quite well by the time the latter is on the ballot. In our data, sarpanch reported to know randomly sampled voters from their GPs personally 95 percent of the time. Second, since sarpanch or their close family members often functioned as low-level middlemen in the past, voters can determine whether sarpanch have anti-poor or pro-poor preferences before entering the polling booth. Third, village life in India is characterized by dense social ties between villagers. In the poor, subsistence-based villages that we studied, the median voter was not only likely to be low-income, but is also socially tied to the poorest villagers. This means that the median voter has a pro-poor preference. One of the key findings in political science research is that parties or candidates must locate themselves near the median voter to win. In our setting, this concerns pro-poor preferences. According to similar logic, voters will screen out politicians with preferences toward targeting exclusively members of one’s caste or kin group. Broad-based (i.e., multi-ethnic) distributive preferences are needed to win in a diverse context as is the case in rural India. Our statistical results strongly support these expectations.
While these results should be cause for optimism, local elections cannot screen out targeting bias toward those the sarpanch sees as his supporters. We find that those perceived as supporters were 3.6 times more likely to have been targeted than non-supporters. When we take socio-economic status into account, sarpanch target poor supporters twice as much as poor non-supporters. In fact, they target rich supporters at a slightly higher rate than poor non-supporters.
Local democracy in poor rural settings can powerfully encourage pro-poor targeting by making it possible for voters to elect pro-poor candidates. This suggests that increasing the power that elected politicians have over allocation relative to unelected local agents (e.g., bureaucrats), while improving the capacity of local elected bodies to withstand external pressure to target the non-poor is warranted. In a context of discretion and weak state capacity, we should expect to see the exclusion of poor non-supporters. Designing effective and feasible ways to limit discretion or incentivize more inclusive pro-poor targeting would help to address this problem.
Mark Schneider is a Visiting Professor of Political Science at Swarthmore College and a CASI Non-Resident Visiting Scholar. He can be reached at email@example.com.
India in Transition (IiT) is published by the Center for the Advanced Study of India (CASI) of the University of Pennsylvania. All viewpoints, positions, and conclusions expressed in IiT are solely those of the author(s) and not specifically those of CASI. IiT articles are re-published in the op-ed pages of The Hindu: Business Line. This article can be read here.
© 2015 Center for the Advanced Study of India and the Trustees of the University of Pennsylvania. All rights reserved.