Unfair Shares

Allocations and Race

Allocations and Race 

FOR THE CURRENT RHNA CYCLE, cities with higher percentages of non-Hispanic white residents clearly received lower RHNA allocations of moderate and lower income housing units; in fact, city demographics were more strongly associated with allocations than city median income. Chart 4 shows the relationship between the percent of non-Hispanic whites in a city and the number of units allocated for the moderate income and lower categories.13,14 The orange colored trend line in the chart shows that as the percentage of white residents goes up, the number of housing units allocated to that city goes down.

This pattern persists even after adjusting for the existing population size of each city, as seen in Appendix Chart 1.

Chart 5 shows the relationship between cities’ median income and the number of housing units each city was allocated. The relationship between income and housing allocations is weak. As previously stated, race was more strongly associated with the number of units allocated to a city than income.15

In addition to analyzing these bivariate relationships, we also utilized a multiple regression test to control for multiple variables at once. Our results from this test also demonstrated that there was a robust relationship between race and city allocation size when simultaneously controlling for income and population size, while income was consistently a statistically insignificant variable.16 These results are displayed in full in Appendix Tables 1 and 2. 

Cities with larger non-white populations do tend to have larger overall populations, but if population size were the only driver of RHNA allocations, our regression tests would show that race were no longer an important variable – but this is not the case. Additionally, in at least one previous RHNA cycle, the 4th Cycle, the relationship between city race and the number of units allocated relative to cities’ existing population size disappeared. This is despite the fact that there was still a relationship between percentage non-Hispanic white residents and the number of overall units allocated for the lowest three income categories. Chart 6 shows the relationship between race and the number of units allocated in the moderate and below income categories in the 4th Cycle, and Appendix Chart 2 demonstrates that there is no correlation between race and per capita allocations for moderate and lower income units during the 4th Cycle.17

Chart 4 showcases the 5th cycle allocations by race

Chart 5 showcases the 5th cycle allocations by city median income

Chart 6 showcases the 4th cycle allocations and race

These results strongly suggest that there was an inequitable allocation methodology in the 5th Cycle that minimized the obligation of jurisdictions with higher non-Hispanic white populations to produce housing, and that this outcome does not need to be a foregone conclusion.

Taken together, these results highlight the key role that whiteness plays in maintaining exclusive communities in the region. Differences in RHNA allocations across jurisdictions of similar population sizes cannot be explained away by the argument that more diverse cities tend to have lower incomes, or that more diverse cities are more populous. Furthermore, despite the fact that the existing income distribution is a factor that was explicitly taken into account in the calculations for the 5th Cycle methodology, income does not predict allocation size for moderate and lower income units, while race does. 

  • 13. Note that San José, San Francisco, and Oakland were outliers for the purposes of this analysis with 20,849, 16,333, and 6,949 moderate, low, and very low income units allocated to them, respectively. Including these three cities skews the line of best fit considerably, meaning that these three cities alone disproportionately influence the line of best fit and the r values, and they were therefore removed from the Pearson’s correlation tests. When San José, San Francisco, and Oakland were included in the statistical test, the r value was -.2649 (p = .0074), meaning that the strength of the associated between race and allocations was weaker when including the two outliers.
  • 14. When testing the correlation between each of the unique income categories and the percent of non-Hispanic white residents, a greater percent of non-white residents consistently is associated with higher allocations for each income category. However, the strength of the association is weaker for higher income categories than it is for lower categories, meaning that more diverse cities receive greater allocations for low and very low income units. Above moderate (r = -.3806), moderate (r = - .3781), low income (r = - .3958), and very low income (r = -.4447).
  • 15. Three outliers, San Francisco, San José, and Oakland were dropped for the income analysis as well. With these three cities included in the analysis, r = -.1426 (p = .1550).
  • 16. Depending on the variables included, the percent of non-Hispanic white residents was variably a stronger or weaker predictor of allocation size than other racial group, such as the percent of Latino residents, percent of African American residents, or the combined percent of Latinos and African Americans.
  • 17. Note that San Francisco, San José, and Oakland were again outliers, with 19,271, 18,878, and 7,140 moderate income units and below, respectively, and they were therefore excluded from this analysis. When they were included, the results were r = -.2694 (p = .0065).