/ US, RACE, WEALTH

The effect of universal basic income on a novel measure of the racial wealth gap

The large racial wealth gap traces its roots to slavery, redlining, and other discriminatory policies, and persists largely due to racial income gaps. In honor of Black History month, we explore how closing part of this income gap with a universal basic income would affect the racial wealth gap, using novel measurements that consider how Black and White families differ across the full wealth distribution.

The two most common measures of the racial wealth gap simply compare mean and median wealth between White and Black families.1 Based on the 2019 Survey of Consumer Finances2, White families have mean wealth 5.7 times that of Black families, and median wealth 6.4 times that of Black families.

As Matt Bruenig pointed out in his June 2020 piece, the mean wealth gap is largely shaped by overall wealth inequality, with over 70 percent of both Black and White families’ wealth held by the top decile.

While the median tells us where the middle of the distributions lay, it can obscure what is going on throughout the rest of the distribution. To visualize a variable’s distribution in full, we can turn to the cumulative distribution function (CDF), which shows the population share with less than a specified value.

The following graph shows how to interpret a CDF graph of of wealth per adult for all families in the U.S. (Note that while we have logarithmically transformed the x-axis, the y-axis remains linear.)

Now, how can we compare the distributions of wealth between Black and White families? The two-sample Kolmogorov-Smirnov (KS) statistical test offers a way to test the difference between two distributions. The KS statistic, or D-statistic (for “distance”), is the maximum absolute vertical distance between two CDFs.

One advantage to this statistical method is that it is quite simple to explain visually. The below chart plots the CDF for White and Black families. The dotted line drawn between the two curves shows where the vertical distance between them is at its greatest.

This shows us that approximately 65.9% of Black families have a wealth of less than $50,000 per adult, whereas only 33.3% of White families fall below that line. As we will see, the threshold can be at any point along these curves; the KS statistic is the value measuring the size of the gap where the gap is the largest, at 0.326 (the actual wealth value corresponding to the D-statistic was $50,888).

In the following model, we simulate how saving a total of one year’s UBI payments could change the racial wealth gap. We apply a flat tax on all income to fund the UBI program, and subtract each family’s new tax payment from their wealth.

In the bottom graph, you can see the vertical distance between the two curves at each point along the x-axis from $0 to $2,000 per month.

It is notable that while the KS D-statistics declines overall, it fluctuates until we get to $500 payments, at which point the D-statistic begins a steady decline to minimum of 0.300 under the $2,000 monthly benefit scenario. This is because the distances between the two curves are much larger around the center of the distributions than around the tails, thus the changes in the KS test are not picking up changes at lower incomes, where the curves were much closer together to begin with.

In our simulations, the wealth point at which the curves are furthest apart increases with the UBI amount. The wealth at which the gap is the biggest rises to just above $121,000, from $50,000 at baseline.

Further, without UBI the median White family has $6.45 for every $1 held by the median Black family. Under a $1,000 monthly payment, this falls to $3.49 for every $1, and further down to $2.64 with a $2,000 monthly payment.

The median wealth for White families rises from $115,103 without a UBI to $119,425 under a $1,000 monthly payment, and rises to $127,505 with a $2,000 monthly payment. For Black families, a $1,000 UBI nearly doubles the median wealth, from $17,853 to $34,183, and a $2,000 monthly benefit grows it further to $48,310.

It should be noted that the median wealth per adult3 rises among White families, it merely rises by less than that of Black families.

Unlike the rising median wealth for White families under the same UBI, mean wealth for White families declines from $557,216 without any UBI to 555,135 under a $1,000 monthly payment, and falls further to $553,054 with a $2,000 monthly payment. For Black families, the mean wealth rises from $97,167 to $104,768 with a $1,000 benefit, and further to $112,369 with a $2,000 benefit.

Finally, we look at the share of families that fall below $50,000 per adult, the point at which those two curves were further apart in our baseline scenario.

The share of Black families with a per adult networth of over $50,000 rises from 65.9 percent to 61.2 percent when we raise the monthly UBI to $1,000 monthly, and falls further to 51.7 percent when we raise the benefit to $2,000 monthly.

In the scenarios we explore, the outcomes for Black families improve in the share of families with wealth per adult below $50,000 declines, median family wealth increases. In the final plot below, you can explore the relative changes in all of the metrics we’ve chosen.

While these metrics aren’t truly comparable, we can see that the KS statistic is more stubborn than other measures, especially the gap in median wealth. Median Black wealth is so low that even modest redistribution can close that gap quickly, but larger gaps in the middle of the distribution will persist.

We will continue this line of research, especially longer-term dynamic analysis to reflect the accumulative nature of wealth. Our simplifying assumption that the entire tax and transfer system flows straight to wealth, rather than consumption, may exaggerate these trends given propensities to consume decrease with wealth. But recurring redistribution would likely have a significantly larger effect in closing the gap, even accounting for consumption. UBI isn’t the only policy that can reduce racial wealth disparities, but it certainly helps.

  1. White and Black families are families headed by a White and Black person, respectively. ↩

  2. We used the Policy Simulation Library’s scf and microdf Python packages for extracting and processing this data in this script. ↩

  3. Family wealth conventionally adjusts for number of adults to avoid distorting statistics due to marriage patterns; for example, see Credit Suisse’s Global Net Worth Report. ↩