Rethinking Caste Discrimination in Economics

In his article in the Journal of Economic Literature, Kaivan Munshi concludes, reiterating the mainstream belief, that caste discrimination in India’s urban labor markets against Dalits (individuals from the ‘lower’ castes) is “statistical.” This blog rejects this notion and highlights the larger issue of structural biases within the field of economics.

What is Statistical Discrimination?

Statistical Discrimination is arguably the most popular theory of discrimination in economics. It suggests that there is an economic rationale behind discrimination, particularly in job and credit markets (see Arrow, 1998). When employers do not have full information about all applicants, they use their beliefs attached to the productivity of their identity as a proxy to assess the productivity of all individuals belonging to that identity. This type of discrimination is based on statistical averages and not on individual characteristics or abilities.

Munshi characterizes caste discrimination in India’s urban job markets as “statistical,” implying that it occurs because Dalits are less likely to receive a good education and job skills, which are only accessible in costly schools and exclusive networks. This statistical difference between Savarnas (‘upper’ castes) and Dalits leads employers to assume, in absence of complete information, that all Dalits are less productive than their Savarna counterparts, even if this is not entirely accurate. Consequently, a Dalit individual, who may be as productive as someone from Savarna caste, still receives lower wages or fewer job opportunities. 

Based on this analysis, Munshi goes on to suggest that such discrimination will “disappear” once there is convergence between caste groups in areas such as education, health, and human capital, reiterating the mainstream belief that capitalism will eventually destroy casteism.

Caste-Based Discrimination is NOT ‘Statistical’

Munshi’s paper draws upon previous research to arrive to his conclusions. Specifically, he relies on a study conducted by Jodhka and Newman (2007), who interviewed Human Resource Managers from 25 large firms to better understand how caste factors into hiring considerations. Their qualitative study revealed that while explicit discrimination based on caste has been replaced by the language of meritocracy in the hiring process, ascriptive characteristics such as “family background” still continue to play a significant role in evaluating the efficiency of an individual.

The analysis of Jodhka and Newman’s findings by Munshi appears to have a limited scope. By characterizing caste discrimination as ‘statistical,’ he suggests that caste discrimination may not stem from animosity or prejudice towards Dalits, but rather from a desire to minimize risks or maximize efficiency. In other words, he argues that caste discrimination in urban labor markets is based on rationality and competition rather than prejudice. 

However, Jodhka and Newman took a more nuanced approach than this, acknowledging that the “language of merit” in competitive capitalism overlooks the systemic discrimination and disinvestment that prevent Dalits from competing on a level playing field. They argue that the meritocracy argument fails to fully address the various forms of institutional discrimination that exist. They also explicitly caution that “one must take the profession of deep belief in meritocracy with a heavy grain of salt.” Munshi appears to have disregarded this caution while using their findings, and thus, has mistakenly equated caste discrimination with statistical discrimination. 

This fallacy becomes clear if we consider a ‘counterfactual’ scenario where, ceteris paribus, fewer of the Savarna people are educated in good institutions and have appropriate human capital. In this scenario, we can ask whether statistical discrimination against Savarnas would still occur in labor markets. However, those familiar with caste dynamics in India would likely answer this question with a resounding “NO.” This indicates that the theory of statistical discrimination does not provide an adequate explanation for caste-based discrimination.

The Limitations of Economics in Understanding Discrimination

Munshi’s study is representative of the mainstream in economics, revealing a significant issue with the discipline: it ignores structural inequalities with historical context and is not equipped to understand them. Consequently, it faces challenges in addressing issues like misogyny, racism, and casteism. The discipline’s market reductionism seeks to account for these problems by either citing market failure, such as imperfect information leading to statistical discrimination, or by relying on Becker’s (1957) “taste-based discrimination” theory, which posits that certain employers attach a disutility to working with, say, Dalit employees, and thus are willing to pay the cost to discriminate in terms of lost efficiency. 

These mainstream theories have attracted severe criticism from heterodox schools of economics. For example, the feminist economics literature has criticized the field of economics for failing to recognize the markets as ‘gendered’ institutions rather than neutral. It argues that market ignores that the existence of ‘productive’ economies is dependent on ‘reproductive’ economies (see Federici and Cox, 1975), penalizing those working in the latter (largely women). This puts women in disadvantage in labor markets despite possessing comparable skills to their male counterparts. This approach rejects the conventional statistical techniques of measuring discrimination, such as Blinder–Oaxaca decomposition method, which treats discrimination as a ‘residual’ or ‘aberration’ after accounting for all wage-earning characteristics, as it fails to account for the discriminatory nature of the market itself.

A New Grammar of Caste: How Economics Reinforces Caste Discrimination

An issue similar to the one raised by feminist economists also resurfaces in the economic analysis of caste, where markets are assumed to be “caste-neutral” or even as a means to annihilate caste (e.g., Prasad, 2008). Caste is considered a “residual” of the cultural past that is confined to politics and society outside of modern, formal, capitalist markets. However, Moose (2020) points out that this conceals the fact that Dalits face discrimination through the very economic processes through which they seek liberation.

As such, economics has provided the neo-casteists with a new grammar of caste by isolating its analysis in markets from its dynamics outside the market. This grammar allows them to argue that caste discrimination can be eliminated without the complete annihilation of caste. Consequently, they demand, just like Munshi, that affirmative action programs meant to support Dalits should be discontinued once statistical discrimination in labor markets is eradicated, irrespective of the state of caste system outside the market.

In addition, the presence of explicitly casteist statements such as Munshi’s claim that “[caste] discrimination is not a consequence of social prejudice, but is simply the outcome of economic competition between groups” in one of the most reputable journals of the field highlights the urgent need for better representation of Dalits in economics and a rethinking of the way caste is approached in the discipline. As the saying goes, until the lions have their own historians, the history of the hunt will always glorify the hunter.