Cville Housing Affordability: Good or Bad, It Hasn’t Gotten Worse

Our earliest presentation on the Charlottesville Comprehensive Plan situation took an in-depth look at the question of whether our city exhibited signs of unusually difficult housing affordability conditions. We encourage everyone to have a look at that presentation, which we believe makes a strong case that Charlottesville’s housing situation is not atypical at the median. The focus of that presentation, however, was mainly cross-sectional. That is, we compared Charlottesville mostly to other jurisdictions. We spent less time considering the time-series aspect of the question. We have attempted to put some charts together to examine the question from this angle: rather than comparing Charlottesville to other jurisdictions, we compare Charlottesville to itself over time. Why is this a useful way to look at matters? A big argument between proponents and detractors of the FLUM relates to timing. We have suggested that the inadequate analytical work undergirding the design of the FLUM requires supplementation, and therefore the adoption of the plan should be delayed. Proponents have remonstrated that we face a worsening housing crisis that demands immediate action. We believe that the unusual circumstances related to COVID and their effect on the housing market are not a good basis on which to make long-term decisions. However, we accept that if there were a longer-term trend of housing affordability deterioration, that would argue for greater urgency in adopting a new housing plan.

It is in that spirit that we examined the time series data more closely. We share below some charts and commentary arising from our analysis. Our conclusion, at a high level of generality, is that the housing situation over the last decade in Charlottesville — at least up until the middle of last year — has been stable, if not improving from an affordability perspective. The COVID effect certainly has made the housing market tighter, but does not appear to have fully reverted the improvement experienced between in the ten years prior. As always, we welcome comments and requests for further analysis. You can reach us here.

Housing affordability improves with increases in income and with decreases in housing cost. Inversely, it deteriorates with decreases in income and with increases in housing cost. Housing cost is often conflated with housing prices. Housing prices are an important variable, for sure. However, the price of a house is a stock variable. It is the price of an asset. That asset produces a flow. The flow is housing services. That is rent — either cash rent or imputed rent. We are going to look at both housing prices and housing services, but it is important to note that it is easy to be misled by the “sticker shock” related to the housing stock value. It is also the case that housing stock data is much easier to come by than flow data. Rent numbers are not as well tracked as sale numbers. There is no rent equivalent to a tax appraisal. And owner imputed-rent is unobserved. Luckily, as it turns out, Charlottesville doesn’t look too bad even when we use housing price as a metric, but we wanted to be clear about the distinction.

First we look at median rents and median household income. The census American Community Survey is our best consistent source for these two variables. ACS data only runs until 2019. The 2020 numbers should be out soon, though many people worry about COVID related distortions to 2020 survey response. Here we have numbers from 2010 to 2019.

Since 2013, median income has growth substantially faster than median rent. This suggests that at the median, Charlottesville has become more affordable over recent years. Of course, it is possible that looking at the ratio of medians can mislead us. Jensen’s inequality does hold for medians! We therefore looked at the median of the ratio over time (median of household-level rent-to-income ratio), which the ACS helpfully reports. We found a similar result. Median rent-to-income has improved slightly over time, in the context of a generally stable level (notice how compressed the y-axis is).

And finally, we looked at the distribution of this statistic. That is, the question of how many renter households are “rent-stressed.” There is no simple measure of what level of rent-to-income truly represents stress, but we used HUD’s metric of “rent burdened” and of “severely rent-burdened”, 30% rent-to-income and 50% rent-to-income respectively. We note for context, without further comment, that for the US as a whole, 48% of renter households count as “rent burdened” under HUD’s definition. Again, we see no signs of deterioration at either level in Charlottesville.

So far, we have only looked at rental costs. As we mentioned, there is no easily observable statistic for owner-equivalent rent. That said, we can look at housing costs. Census tracks this by tenure (i.e. for renters vs homeowners). We looked at housing cost for owners. Bear in mind that we would expect this number to be lower relative to income for owners than renters, as some owners have no mortgage; but possibly higher in absolute terms, because owners-occupied units tend to be larger and of higher quality. Also, owner households tend to have higher income than renter households. Since we are just looking at trends, though, this distinction is not key. We just want to see how the statistic has evolved over time.

Once again, we see the pattern of income growth exceeding cost growth. Housing supply is obviously a key concern for the FLUM. We have demonstrated in other research that Charlottesville has in fact produced housing units at a reasonable clip (1.35% growth a year for the last decade, 70% of it multi-family units). The result is evident in the graph below, which shows the relative growth of households and housing units. It is true that it would be difficult for these numbers to diverge widely for too long (it would result in either a Detroit-like situation or in zero vacancy). It is also true that the dedication of more housing units to short-term rental or second-home use could lead to housing unit production outpacing household growth while still resulting in a “supply squeeze” for residents. But again, we don’t see evidence for such a proposition in the above-described affordability trends.

Now, we will take a look at some metrics that do use the stock variable of housing cost. This allows us to take advantage of higher frequency data that extends beyond the end of the ACS’s data availability period. But before we dive into that, we want to offer an illustration of the importance of thinking of stock and flow as different quantities. We build a time series of the “monthly cost” of owning a just-purchased house over time. Our inputs were the median house price, the average 30-year mortgage rate in Virginia, and estimated real estate tax (using the house price, the tax rate, and the historical assessment-to-sale-price ratio). We compare that monthly cost flow to the stock variable of house price. We use the ZHVI index for price.

It is fascinating to see how very little of the startling COVID-related run-up in prices over the last 15 months has actually made its way into the monthly cost of homeownership in Charlottesville. That number is barely back to the pre-COVID peak. This is almost entirely a function of a monetary-policy-induced drop in mortgage rates. Now let’s compare household income to cost of ownership.

We don’t have 2020 or 2021 median household income data, but national level data suggests relative stability (in spite of COVID) from 2019 numbers. Household income has comfortably outpaced the change in ownership cost. As an aside, this is entirely compatible with the felt frustration of home-buyers. The housing affordability metric we have here depends on the assumption that a buyer has access to credit. We also assumed a 10% downpayment. That certainly can be a barrier to purchasers. And in a tight inventory market, sellers may favor cash buyers. All this tends to disfavor first-time homebuyers. But on a pure affordability metric, Charlottesville has not seen signficant deterioration, even recently. From a policy perspective, we believe this suggests responses other than merely encouraging more housing production. Credit assistance and downpayment assistance might well more directly address the source of this first-time homebuyer frustration. The Housing Chapter indeed has some plans in this regard, which get less attention than the zoning policies therein.

So what’s going on here then? We think it is as we have described in our other work. Charlottesville does not have a fast growing population. Nor does the surrounding MSA. The city, while small and relatively built-out, has not thrown up insurmoutable barriers to housing production — unit growth has run well ahead of population growth. The situation isn’t changing rapidly. It’s plausible that the situation has been in need of change for some time (our cross-sectional analysis tells us that if it is, most other places need the same change, too, and to a greater extent). But a claim of urgency based on a deteriorating situation is not well grounded in the data. We want to point out one other reason why the housing situation here may not have gotten worse in the way it has in other desirable cities. Our earlier research (see our work on the Wharton Residential Land Use Restriction Index and home prices/rents) suggests that income inequality is a powerful driver of housing affordability problems. Charlottesville does suffer from high income inequality. We are bad on income inequality, but fortunately, we have not gotten worse in recent years.

We flag this because it relates to one of our worries about the FLUM. An upzoning that generates more luxury housing additions may also result in inflows of new, wealthier-than-average residents from outside the metro. That can be good for the fisc, and good for people attracted here by employment opportunities created by businesses these folks bring the city, but it can be disastrous for people already here if it catalyzes further intensification of income inequality.

Some final thoughts and caveats. Time-series household income statistics suffer from one important potential problem. They are, in a sense, a form of ecological inference. We don’t know if the household income number has changed because the income of households has changed; or if household income has changed because the identity of the households has changed. It could be asserted that household income growth is not independent of housing costs, but may be driven by it, because the housing costs drive out poorer households. We think that is not what is going on Charlottesville, for two reasons, one direct and one indirect. In our earlier presentation, we provided analysis of mobility data that showed that there was almost no relationship between the probability of a move out of Charlottesville and income. Lower-income people were only very slightly more likely to have departed the city in any given year than higher-income people. The indirect evidence is that Charlottesville’s household income growth has not differed meaningfully from US household income growth. If households in the US as a whole — the composition of which we know has not changed dramatically in a decade — managed to notch a particular amount of gains, we do not have a strong reason to doubt that Charlottesville households could manage the same income gains.

Bottom line: however good or bad you think Charlottesville is on housing affordability, it is difficult to find grounds to say that the situation has become meaningfully worse in recent years. There is some evidence that it has gotten better. So while there may be reasons to advocate for changes in the city’s housing and zoning policies, there is a striking absence of evidence of a need to act in haste. We believe this supports CFRP’s call for more time for the city to pursue a more data-driven and inclusive process for reworking land use policy.

October 2021