On April 21, 2010, a macabre incident of caste violence in Mirchpur, Haryana shocked the Indian public at large. A mob of 97 people, primarily members of the dominant Jat community, torched several Dalit houses, killing a physically challenged girl and her septuagenarian father. The violence itself escalated over an innocuous issue – that of a Dalit boy protesting to young Jats about hurling stones at his pet dog.

Not many would disagree that Manu’s four-fold occupational segregation (popularly known as the caste system) continues to define the way Indian society functions at large, matrimonial alliances being a case in point. Incidents like Mirchpur, though scattered, also continue to bring to public consciousness the fact that caste is not an issue that can be brushed aside. On the other hand, there is increasing talk of the system itself diluting: to pressures of urbanization, migration and more generally, economic liberalization. How often would you and I for instance enquire about the caste of a cook while being served food in a city restaurant? Or as Dalit scholar Chandrabhan Prasad would argue, think about the identity of women cleaning toilets in malls behind the modern veneer of gloves and a uniform?

Evidence – both in the nature of national surveys and small area studies– seems to suggest that there has been a transformation in the situation of the Dalits. In a recently concluded review of how groups traditionally excluded in India (Dalits, Adivasis and women) have fared over the 1980s and 1990s, we at the World Bank find progress on several fronts. Data from different rounds of the National Sample Survey (NSS) indicate impressive gains in post primary educational attainment among Dalit men, at a pace similar to gains registered by non-Dalit/non-Adivasi men. Village level studies reveal much greater group confidence and Dalit ability to assert oneself in public spaces (Jodhka 2008). Social movements asserting the power of Dalits seem to have swept through several states, with political ramifications. And Dalit aspirations seem to be finding a new voice in pop culture, not least in pop songs (sample for instance Punjabi music albums that sell by the name of The Fighter Chamar and Hummer Chamar) (Outlook, 2011).

Yet, and notwithstanding such gains, stories abound of upper caste parents not allowing Dalit students to sit with their children in schools or to eat together (Nambissan 2007). Dalit students continue to perform poorly in schools, perhaps on account of their own belief systems shaped by a history of prejudicial treatment (Hoff and Pandey 2004). And despite gains among men, only slightly over 10 percent of Dalit women in the 2004-05 NSS, report themselves as being educated beyond Class V.

The greatest divide between Dalits and others occurs in the labour market. National Sample Survey data suggest that for the most part, Dalit men continue to be in casual labour. Over the 20 years or so, beginning in 1983, while the proportion of Dalit men in casual labour declined slightly (from 44.6 to 41.7 percent) and in nonfarm self-employment increased slightly (from 11.0 to 15.6 percent), these changes were small to say the least. This is largely to do with the fact that historically Dalits have not owned land. Their landless status therefore excludes them from the large category of farm based employment, driving them to casual daily wage labour, more often than not in the fields of the landed upper castes.

While there has been a slight increase in the proportion of Dalit men engaged in nonfarm employment, they face huge handicaps in starting their own businesses –because they lack the social links to access credit, cheap raw material and markets to sell their produce (Jodhka and Gautam 2008). Not surprisingly, about two-fifths of Dalit men in self-employment are found in low end manual trade.

What about the salaried job market? Here we find that the real differences arise not in the proportion of Dalits and non-Dalits in regular salaried jobs (13 percent as opposed to be about 17 percent non-Dalits), but the kinds of jobs the former land even within salaried work.  In central government services, for example, despite the benefit of reservations, Dalits are vastly overrepresented in the least skilled occupational categories. To give one illustration, in 2006, almost 60 percent of the sweepers in central government ministries were Dalits, indicating that they remain as likely to undertake ritually unclean, manual work as perhaps they were centuries ago. This is what we call a glass wall i.e. a system of horizontal segregation with Dalits bound to certain occupations. Even within an occupational hierarchy, Dalits find it difficult to traverse a glass ceiling i.e. they are more likely to hold the low paying jobs.  This may be partly attributable to their low educational levels. But this is only part of the story. Upon conducting decompositions of earning differentials, we find that about 60 percent of the earnings gap between Dalits and upper caste salaried workers is on account of discrimination.

So what can help Dalits break this glass wall and/or ceiling?

One of the oft cited policy responses is higher education. However education is not necessarily a panacea. In fact, we find that in urban areas, where the bulk of salaried jobs are located; the effects of higher education on Dalit men are more muted in comparison with those on other men. While post primary education almost halves the probability of Dalit men doing casual labour as compared to if they are illiterate (from 35 to 16 percent), the drop is more marked among upper caste men (from 28 to 10 percent). In other words, the returns to higher education do not appear to be significant for Dalits. In contrast, what seemingly takes them forward is their own caste network, which serves the twin purpose of being a source of mutual insurance and of helping them respond to new opportunities (e.g. metal workers in Uttar Pradesh organizing production along caste lines). Together this history of poor returns to education combined with a self-inflicted choice to remain in one’s own social circle, mean that while caste weakens and mutates on one hand, it also solidifies and enhances on the other.

In sum, the picture on caste is mixed. We do find evidence of subtle changes looking through the lens of educational attainment, the labour market and Dalit voice and agency. These lead us to believe that caste is far from the immutable frame that Weber suggested it to be. But these changes appear as mere cracks in the glass wall. Considering their abysmally low starting points, Dalits in India still have a long way to go before they match step with the upper castes, both in education and labour market outcomes. As for Dalit voice, incidents like Mirchpur serve a grim reminder of how political assertion and empowerment through songs are yet to make a dent on the social superstructure of rural India.

Detailed findings can be found in the World Bank’s recent report on Poverty and Social Exclusion in India, prepared by a team led by Maitreyi Bordia Das.