Occupational mobility has widened the gap between Brahmins and Dalits rather than narrowing it, a new analysis of socioeconomic data from Uttar Pradesh reveals.
It finds that there is only marginal movement across occupations of a similar nature or rank among Dalits and no significant mobility among the Other Backward Classes despite decades of political mobilisation and welfare policies. This caste-based segregation affects labour market outcomes and contributes to persisting social and economic inequalities across castes, said the analysis.
Srinivas Goli, associate professor in demography at the International Institute for Population Sciences, Mumbai, says that as anti-reservation sentiments grow among upper castes, only data, facts, and the accurate dissemination of information can provide meaningful solutions.
“Our data from Uttar Pradesh suggest limited occupational mobility among Jatav-Chamars, which is largely lateral – meaning they have not significantly moved into higher-ranking jobs but have shifted within similar occupational classes,” Goli says, referring to the analysis with researcher Rukmi Pradeep on caste and mobility.
After a six-year delay, the Union government notified that the Census would be undertaken with the reference date of March 2027.
India was one of the few countries that had not conducted the Census due to the pandemic. Significantly, after nearly a century, India will also enumerate all castes. This exercise can significantly reduce classification complexities by incorporating pre-coded caste names and biradari (or community) classifications from official state lists and use of technology like artificial intelligence (AI) in computer/tablet/phone assisted survey interviews, said Goli.
Goli works on population dynamics, public health, and regional developmental issues in developing countries particularly India. He is co-author of the 2020 book, Backward and Dalit Muslims: Education, Employment, and Poverty.
In this interview, he highlights the issues around intra- and inter-caste mobility, the need for sub-caste level data, and the upcoming census. Edited excerpts:
Your analysis looks at occupational mobility over three generations at the sub-caste level. Why is it important in the present context? And how is it different from analyses that look at broader caste groups?
Analysis of occupational distributions across broad social groups like Scheduled Castes, Scheduled Tribes, Other Backward Classes and General castes often masks intra-group heterogeneity.
These classifications, used by the Union government for welfare administration and progress monitoring, do not capture the growing disparities within caste groups over decades in key developmental indicators – education, employment, health, and gender.
In the early post-Independence era, broader social groups exhibited greater internal homogeneity, with stark inequalities between them – particularly in critical economic resources like land and wealth.
While welfare policies and affirmative action have improved outcomes for SCs, STs, and OBCs (as evidenced by large-scale surveys), the progress has been uneven within these groups and insufficient to reasonably bridge inter-group gaps. For example, our 2015 survey in Uttar Pradesh revealed consumption-based poverty rates among OBCs ranging from 15% (Jats) to 60% (Lodhs).
Such differences also exist across employment and other indicators, and even across castes (biradaris) of other three social groups (SCs, STs and General Castes).
There is a growing complexity of challenges for welfare policy design. Today, intra-group variation contributes more to overall disparities than inter-group differences – except in land ownership, where the latter still dominates.
In diverse communities like OBCs, sub-caste-level disparities are critical. For instance, landowning groups like Jats, Patidars, Marathas, and Kapus exhibit low poverty but demand reservations due to their underrepresentation in government jobs and higher education, fueling insecurity.
Without disaggregated sub-caste data, resolving these kinds of competing demands for reservations from marginalised groups on one hand and resentful upper castes on other hand becomes difficult. When there is a distress or a voice in the society from any section of the population, we need to study and understand the problem and its roots. In particular, as anti-reservation sentiments grow among upper castes, only data, facts, and the accurate dissemination of information can provide meaningful solutions. Thus, evidence-based policymaking is the only way to navigate these emerging complexities in contemporary India.
You found that during the grandfather generation, the difference between Brahmins and Jatav-Chamars in Grade A and B service jobs was 6.4 percentage points, which in the third generation increased three-fold. Why is there such a glaring gap?
Yes, our data from Uttar Pradesh suggest limited occupational mobility among Jatav-Chamars, which is largely lateral–meaning they have not significantly moved into higher-ranking jobs but have shifted within similar occupational classes. In contrast, upper castes exhibit a divergent trend: despite a few experiencing downward mobility across generations, a majority advanced to better jobs. Relatively, there is less heterogeneity in upper castes in terms of their occupations and even in wealth status. Thus, overall they are faring better than other social groups.
This pattern is particularly evident in the service sector. Due to economic transition, urbanisation, and modernisation, upper castes have largely moved upward into higher-class service-sector jobs (from agricultural and allied occupations), while Dalits have primarily entered lower-tier service roles. Consequently, occupational mobility has widened the gap between Brahmins and Dalits rather than narrowing it.
What are some of the other crucial gaps that you observe in your analysis between various caste groups and their mobility?
The other research published by us using the same survey data reveals stark disparities in land ownership across social groups in Uttar Pradesh. While upper caste Hindus constitute just 15% of sampled households, they control nearly 30% of the state’s cultivable land – far exceeding their population share. Within this group, Thakurs own 2.17 times more land than their population proportion, while Brahmins and other Upper Castes hold 1.84 and 1.85 times higher than their share, respectively.
Among Hindu OBCs, land ownership is highly unequal: Kurmis top the list with a land-population ratio of 2.44, followed by Jats (2.20), Yadavs (1.88), and Lodhs (1.16). Other OBCs collectively fall below parity (0.84), bringing the overall Hindu OBC ratio to 1.28 – exposing sharp intra-group disparities.
In contrast, Dalits and Muslims face severe agricultural land deprivation, owning barely one-fourth the land proportionate to their population share. Hindu Dalits uniformly hover around a ratio of 0.58, with Jatav-Chamars marginally better off than others. The only exception is Muslim Upper Castes, who maintain a near-equitable land-to-population ratio.
The 2015 Uttar Pradesh survey also reveals stark educational disparities across socio-religious groups, mirroring patterns of inequality seen in other indicators. Among individuals aged 16 and above, over 50% of Hindu Upper Castes had completed secondary education, dwarfing the attainment rates of Dalit Muslims (9%), Upper Caste Muslims (15.9%), Hindu OBCs (20.1%), and Hindu Dalits (16.9%).
Yet, intergenerational progress offers glimmers of change: Jatav-Chamars surged from 2.3% (oldest generation) to 31.1% (youngest), while Yadavs, Ansaris, and other Hindu Dalits also showed notable gains.
Despite this, upper caste Hindus – especially Brahmins and Thakurs – advanced disproportionately between generations, widening the gap. By the youngest generation, only Yadavs and Jats neared the educational level of the oldest Upper Caste Hindus, while most groups lagged two generations behind.
At higher education levels, disparities grew starker: the oldest Upper Caste cohort matched the youngest Dalit Muslim, Hindu Dalit, and Muslim OBC cohorts. Though Jatav-Chamars, Ansaris, and Jats are narrowing gaps with dominant castes, structural hierarchies persist, with Upper Castes maintaining a multi-generational lead at every educational tier.
Further, the survey data reveals that a substantial proportion of Dalit respondents continue to experience untouchability practices in various spheres – including dining, housing, social interactions, and access to religious spaces. Interestingly, an even higher percentage of non-Dalit Hindus and Muslims acknowledge the practice and persistence of these discriminatory practices, corroborating the Dalit community's experiences.
There has been a perspective change on Pasmanda (backward) Muslims. But the government does not recognise Dalit Muslims (or Dalit Christians) as SCs, although states include them under OBC categories. Your analyses have noted the experience of untouchability among and the vulnerability of Dalit Muslims “who are equally or more deprived than Hindu Dalits”. What are the welfare alternatives in this structure?
Untouchability and social discrimination one of the key criteria for providing affirmative benefits to Hindu Dalits. When Muslim and Christian Dalits face similar untouchability and social discrimination within their own religion and outside, they deserve the place in affirmative actions of the Government.
I hope the state will address these issues by collecting comprehensive data on practices like untouchability and social discrimination across all religious communities. This evidence-based approach can inform inclusive policies and foster a more equitable society.
You mention that the gap in occupational mobility between Muslim general castes and Muslim Dalits rose from 0.8 percentage points in the first generation to 5.6 percentage points in the third generation. What data did you use to overcome the limitation of data on caste within Muslim communities and what were the challenges in finding reliable data in this regard?
We have used primary data collected by the Giri Institute of Development Studies, Lucknow, through its 2014-’15 project examining the social and educational status of OBC and Dalit Muslims in Uttar Pradesh. The household survey, conducted from October 2014 to April 2015, employed a robust multi-stage stratified systematic random sampling design to select a representative sample of 7,195 households across 14 districts spanning UP's four regions.
Caste, religious, and social group data were collected through three distinct question sets and subsequently cross-verified with social and religious group affiliations. Thus, we implemented rigorous validation procedures, cross-checking biradari (community) reporting with both social group designations and religious affiliations. We have made consultations with sociologists, anthropologists, and political scientists, complemented by thorough literature reviews–particularly crucial for identification and classifications of communities like Dalit Muslims.
For the upcoming caste census in India, state governments could adopt a similar kind of design and approach by preparing pre-coded lists of castes and submitting them to the central government. However, a remaining caste list emerging in the field can be collected as a biradari name which can be coded back into the list at the stage of data compilation.
It’s important to note that while we carefully documented reported caste identities, not all respondents possessed official caste certificates.
How do you see the impact of the Covid on mobility and possibilities of upward mobility particularly for marginalised and oppressed caste groups?
Regrettably, no comparable dataset with sub-caste level information has been collected in the post-Covid period. To my knowledge, this remains the only comprehensive survey of its kind conducted in India. While I hesitate to speculate without empirical evidence, existing socioeconomic patterns suggest the pandemic has likely exacerbated inequalities, disproportionately affecting lower caste communities across all religions.
The government has notified the Census and intends to enumerate all caste groups for the first time since 1931. While some states like Bihar, Telangana, and Karnataka (where the 2015 exercise was recently scrapped) have enumerated castes, how do you look at the government’s initiative given that the previous attempt in 2011 (Socio Economic Caste Census) was dropped due to errors? How can your analysis and its conclusion inform caste enumeration exercises?
As a demographer, I consider access to reliable data – particularly Census data on sensitive social indicators – to be critically important. When utilised responsibly with right intentions, such data serves as a powerful tool for problem-solving rather than exacerbating tensions. These empirical foundations enable evidence-based design and targeted implementation of welfare policies and social programmes.
Some scholars argue that now is the time to abolish the caste system, claiming that collecting caste data will only reinforce caste-based divisions.
However, at a time when caste inequalities have peaked across social, economic, and human development indicators and inter-caste couples becoming victim of honour killings – and when affirmative action is being undermined by the privatisation of education, employment, and healthcare – not collecting caste data would do far more harm than good. The right moment to abolish the caste system will come only when caste-based inequalities themselves have been eradicated.
Furthermore, gathering caste-related data is not an insurmountable challenge nor beyond human capability.
The technological solutions now exist to effectively address enumeration challenges in census data collection. With the upcoming digital census, we can leverage computer-assisted and tablet-based interviews combined with AI-powered identification systems.
By incorporating pre-coded caste names and biradari classifications from official state lists, we can significantly reduce classification complexities.
Modern machine learning algorithms, python and other programming tools have advanced to the point where they can accurately process and categorise even non-numeric qualitative data with high precision.
This article first appeared on IndiaSpend, a data-driven and public-interest journalism non-profit.
