Elucidating the costs of racism

Contrary to the past (though this seems to be changing in the present unfortunately), racism today is often a hidden force that is hard to overtly identify. Its impact typically plays out in subtle ways over the course of time.

Despite this fact we are beginning to see researchers isolate instances where the only obvious explanation of things is the cost of this bias against individuals from a particular background.

In particular three pieces of research around: (1) audit studies, (2) intergenerational mobility, and (3) housing policy have begun to actually quantify the large costs that this bias can have on segments of the population.

Audit studies attempt to look at the hiring prospects of individuals of different races, controlling for everything except for race. They do this by randomly assigning names that signal the race of a candidate to resumes and then using these to apply for jobs, measuring the difference in callback rates across different races as signaled by name.

The idea here is that because names are randomly assigned to resumes we are controlling for differences in education, work experience, and all other characteristics of a resume beyond name that could affect call back rates, allowing us to isolate the impact race can have.

James Heckman raises some valid concerns about this study approach in that it may not generalize to the broader market among other things, but the point is that any significant evidence of discrimination is bad because (1) it is against equal opportunity and (2) does demonstrably impose negative costs on people.

Bertrand and Mullainathan, two economists, find that call back rates are significantly lower for African American sounding names than White names and this is consistent across industries and occupations. The call back rate difference also does not seem to be fully ameliorated by differences in education, work experience, or other characteristics that are deemed positive signals to job suitability, further suggesting this difference is due to bias.

Image for post
Image for post
Table showing the differences in call back rates for the study. Note the consistently positive and significant differences between call back rates for Whites vs. Blacks.

They find that applicants with White sounding names need to send about 10 resumes to get one callback whereas applicants with Black sounding names need to send about 15 resumes. Since they vary the characteristics of resumes they send, they can also compare how valuable having a particular name is to other resume characteristics. They find that a White sounding name yields as many callbacks as an additional eight years of experience on a resume.

Obviously this study has several caveats with the most important being that they are taking name as an indirect signal of race when in fact the name might be signaling something else to employers. It is hard though to conceive of what else that could plausibly be.

Researchers at the Equality of Opportunity Project have done a large study on intergenerational mobility using individual tax records. That excellent NY Times article runs through the numerous findings in the study, but a central finding is that there is a persistent outcome gap between black men and white men that does not appear attributable to differences in parental income, family structure (single or two parent household), educational achievement, and a host of other causes. In fact the differences that appears for black men do not exist when we compare black women to white women, suggesting it cannot be a genetic difference due to race because that would require some genetic factor to only hold for black men, which is very unlikely.

Image for post
Image for post

The article highlights two things that could be driving these differences between black and white men: (1) differences in incarceration rates and (2) geographic area that a person grows up in. However, these seem to only partially explain the gaps that persist because these gaps still exist among high income families who have both lower incarceration rates and grow up in wealthy neighborhoods.

Going through all the potential controls and comparisons really reveals that some bias must be contributing to the negative outcomes for certain racial groups.

Housing and in general access to credit, which is typically used for housing, has been a large source of inequality in the US. A recent study by economists at the Federal Reserve of Chicago highlights the causal impact that federal housing policy has had on individuals and specifically African Americans.

Their study is focused on the Hole Owners’ Loan Corporation (HOLC), which was developed after the great depression to try to stabilize housing markets. The agency created residential security maps in the 1930s for over 200 cities to quantify the risk of lending in those areas using data on housing age, occupancy, and prices, as well as the racial and ethnic makeup of the neighborhoods within. These maps graded residential areas from A (least risky) to D (most risky).

Image for post
Image for post
Cities for which the authors have residential neighborhood grade maps.
Image for post
Image for post
Example residential grade map of Chicago.

The economists quantify the causal effect of the HOLC grades on the evolution of these neighborhoods by comparing outcomes of people living on either side of a HOLC boundary i.e. what is the attributable impact to living in a lower graded area (the treatment group) vs. a higher graded area (control group).

The key is that spatially close areas should be similar except for the difference in HOLC grade, so comparing metrics like level of racial segregation, home ownership rates, and home values in these two nearby areas over time can reveal the impact that a difference in HOLC grade can have. This technique has been used to study countless economic issues such as the minimum wage.

In the context of this study, however, comparing two areas on either side of an HOLC border is not perfect because it is likely those areas have other differences as well, obscuring the actual effect of HOLC grade on neighborhood outcomes. For example, prior to the creation of these maps, there was already a difference in the racial and economic composition of neighborhoods on either side of a border and differences in the trends of these characteristics.

The authors attempt to correct for this problem by creating control neighborhoods that are a weighted combination of other neighborhoods on the higher graded side of a border in their data set to best match the treatment neighborhoods on the lower graded side of a border.

This allows them to address the confounding issues raised by straight comparison across borders because the created control groups are designed to be similar to the treatment groups on every characteristic except for the HOLC grade. It turns out that the results from this method (and others the authors try) are similar to the results from the simpler comparison method above, providing a robustness check and level of confidence in the results.

They find that HOLC maps affected the degree of racial segregation, resulting in rising differences between the fraction of black residents along C-B and D-C borders. Lower HOLC grades also led to negative economic outcomes as well as home ownership rates and home values were lower on the lower graded sides of borders.

The plots below show the expected gap between neighborhoods on either side of an HOLC boundary (either D-C or C-B borders) in the absence of HOLC grades (control boundaries in orange) to the actual gap in the data (treated boundaries in blue). The difference between the lines is the causal impact of a low HOLC grade.

We see that the share of African Americans rose significantly in neighborhoods on the lower graded side of D-C borders. The authors note that this appears to be due to “White flight” i.e. White residents leaving the neighborhood as opposed to Black families moving in.

Image for post
Image for post

In addition to the level of segregation rising along the lower graded side of a boundary compared to the higher graded side, we see a rise in the gap of ownership and home values. People living on the lower graded sides of a boundary, who were more likely to be African American, saw lower rates of homeownership and lower house values.

Image for post
Image for post
Comparison of difference in home ownership rates in a neighborhood in treated and control groups over time. Note the widening over time after 1930 when HOLC maps were drawn.
Image for post
Image for post
Comparison of difference in home values in a neighborhood in treated and control groups over time. Note the widening over time after 1930 when HOLC maps were drawn. Because of data limitations the authors only have house value information post 1970 and prior to 1940.

Given that for most individuals and families a home is the largest asset they own and determines a significant portion of their wealth, these factors had huge negative economic consequences, adversely affecting the development of these neighborhoods over decades.

These three separate strands of research have documented negative consequences for African Americans in the labor and housing markets as well as how these impacts seem to repeat themselves over generations, passing from parents to children.

More work is needed to understand the interaction of the effects of bias in these three areas (labor markets, housing, and intergenerational mobility), but we are starting to get a complete picture of the ways in which certain segments of the population are harmed through this bias.

As we accumulate evidence like the above that points to these persistent and significant costs of racism, we need to accept that racism still exists and can have large, negative effects.

The solution is to expand policies around things we know can promote equal opportunity and reduce the cost of racism such as training on bias, increasing the supply of affordable housing, expanding educational opportunities through local investment and quality charter schools, and fostering relationships through mentors and social networks that tackle the underlying issues raised by the research above.

Vinod Bakthavachalam

Written by

I am interested in politics, economics, & policy. I work as a data scientist and am passionate about using technology to solve structural economic problems.

Vinod B

Vinod B

Politics, Policy, Economics, & Data

Vinod Bakthavachalam

Written by

I am interested in politics, economics, & policy. I work as a data scientist and am passionate about using technology to solve structural economic problems.

Vinod B

Vinod B

Politics, Policy, Economics, & Data

Medium is an open platform where 170 million readers come to find insightful and dynamic thinking. Here, expert and undiscovered voices alike dive into the heart of any topic and bring new ideas to the surface. Learn more

Follow the writers, publications, and topics that matter to you, and you’ll see them on your homepage and in your inbox. Explore

If you have a story to tell, knowledge to share, or a perspective to offer — welcome home. It’s easy and free to post your thinking on any topic. Write on Medium

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store