Field experiment on the behavioural foundations of inter-group discrimination and its effects on public good provision in India

The first step on the way to measure, combat and ultimately eradicate poverty is to understand its causes. Often, the poorest have little or no access to key public goods like health, water or education.

In countries like India, which suffers from widespread poverty, issues like low public good provision exist in the context of a longstanding social divide along caste and religious lines. Previous research has shown that there is lower public good provision when such goods are shared across social divides, whether racial, religious or linguistic in nature.

The purpose of this research project is to understand why this is the case. There are two main reasons why two (or more) groups may not want to share a particular public good with each other. To illustrate our point we will consider the case of education, although the analogy extends to many other cases.

The first explanation is that different social groups may prefer different varieties of the same public good: for example, it may be traditional among certain social groups to send their children to single-sex schools. As such they may be reluctant to send their daughters to co-educational schools. The second explanation is that they do not wish to share the same group with those who do not belong to their social group. In other words, members of a particular group would discriminate against outsiders.

From a policy-maker's point of view, understanding which of the two causes is driving attitudes and behaviour is essential in order to formulate the right policy response. While in the former case, it may be best to invest single-gender schools to boost school attendance among children of a particular group, in the latter case, it may be optimal for the government to encourage inter-group dialogue.

Understanding the behavioural foundations of inter-group social interaction may also provide clues as to why certain social groups tend to be socially insular, and why others are not. Is this to do with close ties of affiliation or is this the result of distrust towards outsiders? Is this a function of how large the group is relatively to the whole population? This has important consequences in terms of social mobility, and poverty alleviation. The answer to this question is important to political scientists and economists, as well as policy makers.

We will study this question by combining extensive household surveys with field experiments in West Bengal, India. This way, we hope to uncover the behavioural motivations for inter-group discrimination, while studying them with a highly relevant population. West Bengal, India has one of the largest proportions of Muslim citizens in India (about 24% of the West Bengal population). As such, it is an ideal case study to understand the underlying causes of the impact social and religious fragmentation have on behaviour.

The beneficiaries of this research project will be academics, Indian, and international government agencies, NGOs as well as anyone in the wider civil society who is interested in issues of poverty, racial and/or religious discrimination. We will concentrate our attention in this section to non-academic beneficiaries.

In terms of policy users, we plan to produce a non-technical impact report, which will be generated at the end of the grant period. This report will contain not only a summary of the findings of the project, but also their policy implications to the West Bengal Government and India-based NGOs. We will organise a workshop in which we will present our findings to representatives from the local and state governments, NGOs and other parties such as USAID or the World Bank - the latter not only has an interest in issues related to caste-based discrimination, but has engaged in collaborative research with the CSSSC.

We envisage that our research will impact on poverty alleviation efforts and policy indirectly. It will do so by identifying the channels (in particular inter-group discrimination) through which social fragmentation along religious lines affects public good provision. Understanding why the presence of multiple social groups (typically ethnic and/or religious) in a given society leads to lower provision of public goods allows governments and NGOs to create the appropriate policy response and maximise the return to aid investment.

A case in point is the provision of a local public good, such as education. Two social groups like Muslims and Hindus may prefer not to share a local school for two very different reasons. On the one hand, they may prefer two different types of public good: madrassas typically segregate students along gender lines, while state schools do not, which may be important to more orthodox Muslim families. On the other hand, members of one group may not want to share the public good with people of the other group. The optimal response is quite different depending on the cause. In the former case, subsidising madrassas is optimal, as it will lead to higher rates of school attendance by women from Muslim backgrounds. In the latter case, encouraging attendance into the state school system may be more beneficial in order to reduce prejudice across religious lines.

Finally, we hope to be able to introduce experimental economics as a valid new tool to policy makers. Policy designers increasingly resort to experimental methods to test the applicability of particular market mechanisms. Examples include the design of 3G bandwidth auctions in the UK in 2000 (Binmore and Klemperer, 2001) and the sale of CO2 emission permits in Australia (Cason et al, 2003). We hope to demonstrate that economics experiments can also be useful tools to uncover underlying behavioural mechanisms around which one can tailor policy.

The timescale for impact to be realised is likely to exceed the duration of the grant. The main means to achieve impact (the workshop and the report) will be realised at the end of the grant period. As such, we expect that impact of the research in terms of policy and wider community will be felt in the 2 years following the completion of the grant. We hope this will happen through the recognition of the effect of discrimination in the design of poverty alleviation policies.