This paper presents a theory of discrimination for markets in which there are complementarities between different tasks. It is shown that, in such a setting, even when groups are a priori identical, employers will end up discriminating against certain groups. Group discrimination serves the purpose of creating a focal point in a market game. In this model, the free market, far from curbing discrimination, nurtures it, and thereby creates the need for purposive policy intervention. It is argued that with the rise of technology the problem of discrimination as focal point will get more acute and we will have to think in terms of affirmative action or a system of taxation and subsidy to support groups that get excluded.
For good or for bad, group identity matters in determining market outcomes (Akerlof and Kranton, 2010, Sen, 2006). Discrimination against certain groups of people and, by its converse, in favor of other groups has been common practice, observed in different societies and through different periods of history. India’s caste system, with its attendant practice of intolerance and effort to marginalize large groups of people, is an example, as is the history of apartheid in South Africa and racial discrimination and slavery in the United States. From an analyst’s point of view, they are often troubling because norms and laws often merge into each other. Some of these heinous practices were explicitly backed by the law as in the case of South African apartheid and US slavery. At other times, such as with India’s caste system through history and now (see Deshpande, 2010) or with racial discrimination faced by African Americans in contemporary USA, it was not backed by the law but by social norms, customs and individual beliefs and preferences. The focus in this paper will be on discrimination which does not have the backing of law.
As a mirror image of this, we often see certain groups benefiting from discrimination in their favor. This has been true of men through long stretches of history and even now in most societies. Similarly, in the United States, United Kingdom, in India during colonial times or South Africa till recently, if you could choose your skin color, I would strongly recommend white. Where do these discriminatory preferences come from and why have they been so persistent?
Without doubt there must be many explanations for this, ranging from plain, simple bigotry and prejudice, to various forms of statistical discrimination that economists have written extensively about. While not wanting to take away from those standard explanations, this paper presents a novel argument whereby discrimination has no innate origins but arises naturally in markets where there happens to be some complementarity between the different tasks we do. This kind of discrimination is a product of the free market and the beliefs of ordinary people. The upshot is the group identities of people come to matter in such settings. It is, as I will show, closely linked to the idea of ‘focal point,’ used in game theory (Schelling, 1960). In particular, race, caste and gender becomes important in equilibrium because they acquire the salience of the focal point.
In some ways, the focal point theory of discrimination developed in this paper is more alarming than other forms of discrimination where one can point to the source, be it human mendacity or the distortions of statistical information. In my analysis removing government interference and allowing competition in the market to flourish does not remove discrimination, as standard economics had suggested, because it is in fact a product of precisely the free market.
The model is based on sufficiently realistic assumptions to make me believe that, while other forms of discrimination no doubt occur, the focal point theory of discrimination does play an important role and, as such, deserves greater attention. In the last section I will argue why, given trends in contemporary labor markets, this is likely to become even more important.
- See Arrow (1973, 1998), Phelps (1972), Spence (1974), Stiglitz (1974).
- A polar case of this in a development context occurs in Kremer (1993).