Affordability Metrics: An Update
- We update our analysis from last fall of the housing affordability trend in Charlottesville over time.
- New data from Bureau of Labor Statistics allowed us to extend analysis to 2021 and to deal with some of the shortcomings of census income data.
- We find similar results: housing affordability as measured by averages of housing prices and wages has actually improved over time in Charlottesville, unlike in many other places.
The principal narrative used last year by the Planning Commission to justify massive increases in density under the new Comprehensive Plan was that housing in Charlottesville has become increasingly unaffordable. Research we shared last fall showed that the claim of trend deterioration was not true. By nearly every metric, affordability at the median had improved over the previous 10 years. You can read that research here. The data, though, had a few weaknesses. First, census data only extended to 2019. Second, much of the census data is “smoothed” (5-year ACS data is based on the 5-year average of responses) causing changes to “lag” reality. Third, as we have discussed elsewhere, household-based metrics in Charlottesville can be seriously distorted by the presence of student households. Finally, we have provided much evidence in our research that housing affordability stress in Charlottesville, where it exists, may have more to do with labor market disconnection than with housing market production. Labor force disconnection, if it is a problem, will depress median household income metrics, obscuring the picture of affordability for labor-market participants. While housing affordability stress due to labor market disconnection is no less painful for those experiencing it than affordability stress due to housing market problems, it represents a distinct problem with distinct causes and solutions.
We decided to revisit the theme of the previous analysis but to use a different data set for the income side of the equation. The Bureau of Labor Statistics publishes county-level wage data. This data set addresses many of the shortcomings of the census data. First, it is both timely and up-to-date. We have numbers through the end of 1Q2021. Second, as it focuses on labor market participants, it will screen out non-working students (the source of large distortions in household metrics). Third, it measures wages for active workers. People who face affordability problems because they cannot access the labor market will not factor into these results. One shortcoming of the data is that it is mean and not median data. However, wage data (as opposed to income data, which includes capital gains and pass-through business profits) does not suffer from upward outliers skewing the mean in the same way income data does.
Based on BLS wage data (and Zillow home price data for the same period), we see no meaningful deterioration in housing affordability since 2009.
We wanted to make sure that the data set showed home prices outpacing income in places we are pretty sure have highly stressed housing markets. We have in the past used Bozeman as our “control” for this purpose. And sure enough, we see the opposite story there (in fact, these numbers are for Gallatin County; for Bozeman city itself, they would likely be a good deal worse as housing prices are higher than countywide; unfortunately BLS data only goes down to the county level).
One other worry we had — which we mentioned in the earlier time-series piece — is that income could keep up with home prices because the identity of the residents is changing in response to higher home prices. That is, home prices are affordable even when they rise only because rich people are replacing poor people. Now, we have looked at both census and IRS data elsewhere and found little evidence of this phenomenon. However, we thought it would instructive to look at a specific industry as a check. Presumably, “richening” of a city would involve largely change in industry composition. Since we have heard a lot of concerns about “nurses and teachers” specifically (which we did address in an earlier presentation), we used BLS category 1025, Education and Healthcare. Somewhat surprisingly, we found that affordability had improved even more for this sector.
Again, Bozeman showed a different story. Bozeman is also a college town, so we figured that if there is a “higher education” distortion here, we would see it Bozeman, too. But we didn’t. Bozeman showed a huge deterioration.
Finally, we present the data in the form of the ratio of median home price to average annualized weekly wage. We did this for “all sectors” and for “education and healthcare.” We also threw in Minneapolis as an additional comparison. Recall that this ratio is of median home price to the average weekly wage of an individual wage earner. The ratio to median income of a household with employed persons will certainly be much lower, as most homeowner households will have more than one wage-earner and some non-wage income. You can see that Charlottesville’s ratio has been stable and is between (low) Minneapolis and (high) Bozeman, but closer to the the former.
The results for Education and Healthcare are even more dramatic. Here, Charlottesville has the lowest ratio as well as the best trend.
Finally, we show the ratio of Zillow median home price to BLS mean wages (2020-2021) for a range of counties, as well as the population-weighted median for all US counties with population above 25,000. Charlottesville is very slightly above the US median (57th percentile), but substantially below the Virginia median (37th precentile).
At this point, we feel quite certain that the narrative of a housing-supply-oriented housing affordability crisis in Charlottesville is obviously and demonstrably false. The question is whether city leadership is so willfully impervious to evidence and argument that it will insist on a risky treatment for a disease that is not present. When the Comp Plan process started, the city leadership’s error consisted only in failing to secure the right kind of foundational analysis for making good policy. Now, as the City prepares to codify massive use and form intensity increases into the zoning ordinance, it is studiously ignoring available analysis and evidence, in favor of repeating facile slogans. We would hope that even at this late date city leadership might come to its senses and embrace data-driven policy making. But we fear, rather, that it will be up to voters to embrace different leadership.
Appendix: Public Administration
We also ran the numbers for the Public Administration (1028) classification. What we found might explain why staff buys into the housing crisis narrative. While Charlottesville’s house-price-to-wage ratio for the public administration category is not unusually high in the national context, the trend raises questions.
Above we showed that for the overall local employed population, average wage growth has outstripped home price growth. However, according to the BLS stats for public administration workers, affordability has deteriorated. In fact, according to these numbers, it is now worse for public administration workers than for the general workforce.
But this only gets back to our point that housing affordability stress can stem either from housing supply inelasticity or from labor market failures (or both, or other factors!). That the local public sector may have failed to adjust its wage scale appropriately (perhaps in part due to prioritizing over employee compensation ill-advised spending on single-bidder consulting contracts like the one with RHI?) is not the fault of zoning. In fact, it is largely the fault of the same leadership that wants to blame every social ill it can think of on zoning.