Residential Segregation and Unequal Access to Local Public Services in India: Evidence from 1.5m Neighborhoods

Sam Asher ⓡ, Kritarth Jha ⓡ, Paul Novosad ⓡ, Anjali Adukia ⓡ, and Brandon Tan ⓡ

June 2023

Download the Segregation Paper Media fact sheet View all figures

The Segregation of Muslims and Scheduled Castes in India

India’s villages have historically been segregated by caste. Marginalized groups have lived together at the outskirts, excluded from village social circles and amenities.

We set out to examine whether cities were different. Does the market bridge social divides, or do rural inequities get reproduced?

We used administrative data on 1.5 million urban and rural neighborhoods to calculate the Dissimilarity Index, a standard measure of segregation. The Dissimilarity Index describes how evenly groups are distributed across the neighborhoods of a city. It tells us what fraction of marginalized group members would have to move in order to achieve an equal distribution across neighborhoods. High values indicate high levels of segregation.

We find that urban areas are just as segregated as rural areas for SCs, and even more so for Muslims. India's Muslims and Scheduled Castes are about as segregated as Black people in the United States today.

Access to Public Services

What does access to public services look like in segregated neighborhoods?

This graph shows secondary school access as a function of the Muslim share of a neighborhood. Each point shows the mean secondary school availability across about 20,000 urban neighborhoods.

On average, when a neighborhood is 10% Muslim, there are about 5.6 secondary schools per 100,000 people.

In neighborhoods that are 75% Muslim, there are only about 42 secondary schools per 100,000.

Neighborhoods with moderate numbers of SCs are doing well, but the most segregated SC neighborhoods also have far fewer secondary schools.

Granularity Matters

The extent of public service inequality depends a lot on the geographic scale at which you are looking.

Let’s look again at urban secondary schools that serve members of Scheduled Castes.

States with more SCs have more secondary schools

A state with 10% higher Scheduled Caste share has an additional 0.3 secondary schools per million people.

Within those states, districts with more SCs have more secondary schools...
...and within those districts, towns with more SCs have more secondary schools.

Holding the district fixed, a city with 10% higher Scheduled Caste share has an additional 0.7 secondary schools per million people. These effects add up to a substantial advantage in schooling access in towns with many SCs.

But when we go to the neighborhood level, nearly all of that advantage disappears.

In the same city, a neighborhood with 10% higher SC share has 1.1 fewer secondary schools per million people. Combining all geographic levels, SC neighborhoods are slightly more likely to have secondary schools. The neighborhood disadvantage erases most of the cross-district and cross-town advantage.

For Muslims, the story is a little bit different.

The state, district, and town Muslim shares are not substantially correlated with school availability.

But Muslim neighborhoods face a substantial disadvantage.

A 100% Muslim neighborhood is only half as likely to have a public secondary schools than a non-marginalized neighborhood.

Adding up across all geographic levels, Muslim neighborhoods have considerably fewer secondary schools.
Our findings are similar for health clinics, water and drainage infrastructure, and for electricity access, in both rural and urban areas.

You can see more results like this in Figures 5–7 of the full paper, and in our interactive paper appendix webpage.

Child Outcomes

Unequal neighborhoods can entrench inequality.

Young people living in poor neighborhoods have worse opportunities.

We looked at child school completion in neighborhoods where marginalized groups live. In the average urban neighborhood with a low Muslim and SC share, young people on average get 9.2 years of schooling.

In the same city, in a neighborhood with a 100% SC share, the average person gets 1.6 fewer years of education.

Children in Muslim neighborhoods fare even worse, getting 2.2 fewer years of schooling than children in non-marginalized neighborhoods.

This isn’t just about social groups, it’s about neighborhoods.

In neighborhoods that are majority SC and majority Muslim, all social groups are doing worse in terms of education, not just SCs and Muslims.

The graph to the right compares outcomes (by social group) between non-marginalized neighborhoods and neighborhoods that are majority Muslim or majority SC.

A non-Muslim, non-SC child growing up in a majority SC neighborhood gets 1.4 fewer years of schooling than the same child in a better-off neighborhood.

The same child in a majority Muslim neighborhood would get 2.2 fewer years of schooling.

For more results and context, please download the full paper .

© Development Data Lab, 2023

About DDL