Research shows dramatic caste gap in performance when caste is made conspicuous
New research surfaces a core issue related to persistent inequality: that there is a significant caste gap in performance when social identity is made salient. Experimental evidence gathered from rural Uttar Pradesh, one of the poorest regions of India, shows a pronounced difference in performance between low and high castes when social identity (caste) is announced before participation in a maze-solving exercise. When social identity is not announced, there is no caste difference in performance.
Part of the background material for the World Bank's World Development Report 2006 on Equity and Development, this experimental evidence from north Indian villages is presented in detail in a World Bank policy research working paper -- "Belief Systems and Durable Inequalities: An Experimental Investigation of Indian Caste" (full text in PDF form), by Bank researchers Karla Hoff and Priyanka Pandey.
The study suggests that a history of social and legal disabilities may have persistent effects on a group's earnings through its impact on individuals' self-confidence and expectations. "Ending discrimination is therefore not the end of the story," says Hoff, "The common knowledge of a belief system that stigmatizes the low caste holds back caste-ridden communities long after the laws that enforced discrimination were abolished."
How the evidence was gathered in rural Uttar Pradesh
Village junior high school, Uttar Pradesh, experiment site
Uttar Pradesh, where only 56 percent of the population is literate [Source: 2001 Census of India], has long been considered a puzzling case of lagging development. The impetus for Hoff and Pandey's research was an interest in understanding the factors perpetuating high illiteracy.
In visiting the schools and talking with villagers in 2002, Hoff and Pandey were struck by the passiveness of villagers in the face of teacher absenteeism in village schools, one school building in such disrepair that children could not enter the school, the absence of classroom furniture, and long delays in the payment of parateachers hired by village governments.
The authors focuses their research on cultural and institutional influences on the responses of villagers to incentives. Two experiments were conducted that required Grade Six and Seven boys to solve as many mazes as they could out of a given packet, with monetary incentives. Boys were recruited from the lowest caste, Chamar, historically 'untouchables' in Hindu society, and the three highest castes, Thakur, Brahmin and Vaishya.
321 high-caste and 321 low-caste boys from junior high school participated in the first experiment. When caste was not publicly announced, there were no caste differences in performance. However, when caste was publicly announced, the number of mazes solved by low caste boys dropped by a dramatic 25%. When caste was announced but a random draw of a name determined who in a session of six would be paid for the mazes he solved, the caste gap in performance disappeared.
Effect of announcing caste on proportion of low caste in each learning decile
(click on image for full size graph)
To test whether the low caste boys' expectation of discrimination contributed to the caste gap in performance, Hoff and Pandey undertook a second experiment that manipulated the scope for discretion in rewarding performance. When offered a gamble in which the link between performance and reward was mechanical, making caste salient did not cause a caste gap. However, when human judgment played a role in rewarding performance, making caste salient created a large caste gap in the proportion of subjects who refused the gamble.
The way the experiment was designed allowed the researchers to exclude socioeconomic differences between low and high caste boys as explanations of the caste gap in performance. Their results therefore suggest that when caste identity is salient, low-caste subjects either expect that others will judge them prejudicially, or have greater anxiety or lower self-confidence, or all three of these.
The experiment with mazes was conducted in January and March 2003. The weather was extremely cold in January, classrooms were unheated, and many children did not have warm clothing. Yet the children's performance in the experiment in these two months was similar. This suggests the remarkable resilience of the children.
Belief systems shape responses to opportunities
"We carry around a portfolio of social identities to which we refer, depending on the context, and making one or another of our social identities salient changes our behavior," says Hoff, with particular reference to the ideologies representing certain social groups as intrinsically inferior.
The authors see the value of the experiment as suggesting a more general dynamic underlying the perpetuation of inequality in many societies than has been broadly recognized in the economics literature. When social identity is salient, social groups will differ in the way they view and respond to economic opportunity. These differences in behavior are a legacy of the historical processes that categorized people and shaped their beliefs and expectations.
"Long after the legal barriers to economic and social advancement by oppressed groups have been abolished--or the conditions that gave rise to those barriers gave changed--the expectations that historical conditions created may remain and may give rise to behaviors that reproduce the effects of those historical barriers."
To the extent that the findings can be generalized to economic performance in north India, where caste distinctions remain marked, the authors say that their findings suggest that the aggregate economic effect on society of expectations associated with caste are unambiguously negative.
In the pipeline: does caste reproduce unequal structures of opportunity?
Hoff and Pandey are extending their experimentation and research to the study of whether social identities are also likely to play a role in the submission to (old) authority and in the reproduction of unequal structures of opportunity.
Their next paper, scheduled for 2006, stems from observations of the continued exclusion of large components of the population from decision maiking in a village in Uttar Pradesh even after the position of village headman (pradhan) was reserved for a low-caste individual. "Can democracy be imposed in a highly unequal society? The answer is no, unless the citizens come to demand it," says Hoff. "The low-caste pradhan does not even have access to the village account books."
The authors: Hoff and Pandey
Karla Hoff, a one-time Peace Corps volunteer in Cote d'Ivoire, is a senior research economist in the World Bank's Development Economics Research Group. Her research focuses on institutions and institutional change, particularly in the former Soviet Union and India. Profile and works
Priyanka Pandey, an Indian national, is currently a visiting research fellow in the World Bank's Development Economics Research Group. She was an assistant professor of economics at Pennsylvania State University before coming to the Bank. Her research focuses on service delivery in primary education and health, particularly in Uttar Pradesh, India.