Gentrification and Housing Affordability Challenges: Yelling at a Wave and Conjuring a Tsunami

CFRP has done extensive research to try to provide an accurate picture of the contours of the housing market and the nature of housing affordability challenges in Charlottesville. Despite the irrefragable commitment of local activists to the narrative of an acute local housing affordability crisis stemming from overly restrictive land use policies, competent analysis shows a market that is neither highly expensive, highly unaffordable, highly restrictive in land use, or low in housing production. The consultant hired by the city (the only one to respond to the RFP) made an absolute hash of the economic analysis, though one wonders whether anyone’s data would loosened the Narrative’s grip on local minds.

In this piece, we look at some new data sources that cover ground we have been over before. This time, rather than looking at it through the lens of whether the Narrative is right or wrong, we want to focus on the question of whether there is a lot of room for “gentrification” to run in Charlottesville. Why is this an important question for zoning? The literature on the effect of land use regulation on housing prices and affordability is very clear about the existence of two competing effects: the supply effect and the amenity effect. The supply effect tends to drive down prices, though it’s not quite the “simple supply and demand” that people sometimes cite as though repetition of the phrase were a dispositive argument (for reasons we have discussed elsewhere). But, yes, if there is a supply shock that creates more housing units that are largely similar to housing that already exists, then ceteris paribus, housing prices should be lower than otherwise (again, no one is going to build just to drive prices below replacement cost, but in places where housing prices are far above replacement cost, the supply effect should be meaningful). The amenity effect, though, runs the other way. If new supply brings in new residents, that shifts demographics. The amenity effect kicks in when the new residents are wealthier than existing residents. The shifted demographics change the attractiveness of the jurisdiction, both directly (like attracts like, when it comes to incomes and education levels) and indirectly (through the arrival of amenities catering to the new residents and through the attraction of new lower-wage workers to meet the services demand of new, wealthier residents).

But, obviously, the risk that the amenity effect dominates the supply effect depends on whether the jurisdiction is already gentrified. It seems quite unlikely that allowing more building in Palo Alto could possibly make Palo Alto any richer or amenity-dense than it already is. Supply effect will dominate the amenity effect. On the other hand, to use an extreme example for the sake of clarity, building condos in Big Sky, MT meant tons more rich people, more restaurants for them, and all kinds of services businesses, with attendant employees, serving the new residents. Prices kept going up for both luxury and commodity housing — amenity effect dominates supply effect. Therefore, it is easy to see why a jurisdiction that is already heavily gentrified might see less risk from the amenity effect than would a jurisdiction that could still see significant gentrification.

So where does Charlottesville fall on this scale? To go by the Narrative, Charlottesville is already heavily gentrified: super-expensive cost-of-living, overall and in terms of housing, unaffordable rents for workers, massive influx of high-earners pushing out lower-earning households, and a tsunami of remote workers. Once again, the Narrative turns out to be nonsense.

First, let’s look at some new data on cost-of-living (you can look at our earlier research for more on this question). We got access to the Council for Community and Economic Research’s (C2ER) cost-of-living index. C2ER’s index is heavily used by the private sector (for example in setting wage scales when moving employees geographically) and has impressive depth and coverage. So how does Charlottesville stack up?

Charlottesville stands a whisker above the national average for overall cost-of-living and only slightly higher for housing costs (around 2% and 7% above the national average respectively). The city has sat at around the same level since 2017, meaning the overall and housing cost-of-living have just tracked the national average. Recall, C2ER’s series covers the vast majority of markets in the country, some of which are “failing jurisdictions” with extremely low (deservedly) cost-of-living. See, for instance, St. Louis, in the above chart. Some highly desirable areas that have seen large inflows of high-earners come out much more expensive than Charlottesville. For all the talk of some sort of mass-flight “over the mountain” (and again, we have looked at this in other research and found very little evidence to support it), the Staunton-Augusta market’s cost-of-living is not dramatically lower than Charlottesville’s (8% lower overall, 15% on housing). In addition, prime-age, non-student residents of outer-lying areas who work in the Cville MIGPUMA (MIGPUMA is a Census area designation for place-of-work and source-of-migration) don’t seem to have sharply lower incomes than those who live and work in the MIGPUMA (a place like Jackson Hole sees incomes fall by roughly 50% for workers living “over the pass” in Idaho, as a point of comparison).

What about people moving over the mountain? We mentioned above that one cannot simply compare those who move from Charlottesville over the mountain to those who live in Charlottesville, because movers (wherever they are moving to) are systematically different from non-movers — the most salient difference is that movers are younger than non-movers. Therefore we compare prime-age non-student households that moved from the Charlottesville MIGPUMA to the “over the mountain” PUMAs to (1) households that moved to some other MIGPUMA and (2) households than moved from within the Charlottesville MIGPUMA to the Cville PUMA (most of these will be within-the-city movers). Once again, we see very slight differences, undercutting the idea of a low-earner exodus over the mountain.

But if we talk about gentrification potential, there is so much more room for cost-of-living increases than decreases. Think about the cost-of-living arbitrage for someone living in Arlington (overall cost-of-living 36% above Charlottesville, housing 120% above Charlottesville)!

Another aspect of this data that speaks to the housing situation here relates to the gap between overall cost-of-living and housing cost. In general, housing cost should filter into the cost of other categories, and the reverse can happen as well. One potential indicator of a housing market that is not generating supply commensurate with price signals is a housing COL index that runs far ahead of the overall COL index. And what do you know? Our housing index sits pretty much in line with the overall COL index. A “failing city”, where housing prices plunge below cost of production, like St. Louis, will have a housing index well below the overall index. But look at booming Zoomtowns like Bozeman and you will see a housing index that has shot ahead of the overall index.

We also wanted to check wage data to see how that mapped to the cost-of-living index. As we have mentioned elsewhere, the presence of student households skews Charlottesville’s household income data. To deal with that distortion, we look at BLS wage data. That way, we only capture people involved in the labor force. It helps screen out students, but we still get some distortion from students in the labor-force part-time, which makes the average wage look lower than would if only non-student workers were included.

Charlottesville’s mean wage relative to the COL index comes in slightly below the national average, but again far above places like Bozeman. But the conclusion we would take from this chart is that Charlottesville has more a problem of low wages than one of high housing prices. That a place with a near-monopsony employer would experience low prevailing wages is neither a shocking outcome nor a consequence of land use policy. Bottom line cost-of-living: we are nowhere near as high as “gentrified” areas. Things could get much worse, and Cville will appear quite cheap to prospective residents from NoVA, for example.

Now we will take a quick look at rental affordability and rent-stress. We have pointed out many times that the Affordable Housing Plan drafted for the city by consultants (HR&A did most of the work) did not propertly account for the distortion from student households. We know that the typical UVA family is wealthier than the typical local family (2017 data showed nearly double the median family income); at the same time, most students show minimal income. Recall that nearly all Pell Grants go to students from families with less than $60K annual household income. Very few UVA students receive Pell Grants.

For US students overall, in excess of 30% of first-year undergrads receive Pell Grants. We took a look at the data directly ourselves, to compare UVA families to those of peer institutions (Virginia Tech, VCU, and JMU) and to Charlottesville non-student households. For the schools, we used National Center for Education Statistics for 2019-20, for Charlottesville we used 2021 ACS 5y PUMS data. One point to note is that for public institutions NCES only tracks family income data for in-state students getting some sort of Title IV assistance or loans, so we only looked at this population, but we think it is fair to assume that students without any Title IV participation are relatively wealthy.

Seems legit that poorer families in Cville should have their neighborhoods remade to accommodate housing for the gilded offspring of the state’s wealthier families, no? Considering the wealth of UVA families, perhaps it should not surprise you to learn that the biggest residential projects in town currently under consideration are clearly aimed at students. Rhetoric that screams solidarité, benefits that accrue to jeunesse dorée. But, hey, activists, keep telling yourself new construction will be for “teachers and nurses” — you’re all-in on the idea that repeating a statement over and over makes👋 it👋 true👋!

The Census Bureau has published research on the importance of considering the student distortion when looking at poverty statistics in college towns. Apparently, the highly paid “experts” at HR&A missed it. Regular Census cross-tabulation did not allow any easy way to exclude student households. We finally “bit the bullet” and built an analysis using Census Public Use Microdata (PUMS). We could look at data in a much more granular way, with the one downside being that PUM areas (PUMAs) do not match up exactly with cities or countries. However, the PUMA that contains Cville City is a pretty good proxy — it covers the city and a few census tracts to the north. We were able to go from the “headline” rent-stress numbers to numbers reflecting subpopulations that might be of greater interest when considering whether we have a rent affordability problem. That is, market rent will always appear unaffordable to students showing no income and be unaffordable to people disconnected from the labor market. Rent-stress caused by high rents will show up in high levels of rent-stress among the employed population. Here’s what we found in Charlottesville, with the Commonwealth as a comparison. The data is PUMS data from the 2021 5-year ACS data set.

The headline number for rent-stress at the 30% level is what we often hear cited, uncritically. For example, from a WVIB report: “The councilor [Michael Payne] notes that 60% of people living in Charlottesville are renters, with half being cost-burdened, meaning at least 30% of their income every month goes towards rent.” We’re not picking on Mr. Payne. He is not a demographer or economist, nor has he ever claimed to be one. He relies on the city and its consultants for this sort of data; HR&A and RHI lazily peddled unfiltered statistics and set loose this ScareFact on the community. Even a smart person like Mr. Payne will have trouble designing policy if he is supplied with inaccurate information. Luckily, we can drill down. Student households create a huge distortion, with over 70% showing “rent-stress” according to the HUD 30%-of-income threshold. For non-student households, the number is below 40%. But that’s not all. Charlottesville has many households disconnected from the labor market, and their income-to-rent ratio will by construction look terrible. But that is not a market rent problem. For non-student households with at least one member working 35+ hours per week, the rate of rent-stress (at the 30% income-to-gross-rent level) is below 30%. If one person in the household has an Associate’s Degree or higher, the rate drops to just over 20%. At a household income equivalent to having two full-time (2000hrs/year) employees earning the UVA minimum wage, the rate is less than 10%! For those households earnings less (<$60K), the level of rent-stress is almost 80%. Again, is this a land-use problem? The median income of households with less than $60K median income in Cville is under $30K. No market rent will be affordable to households earning so little. From the standpoint of gentrification risk, note that the rent-stress levels for non-students, the employed and the credentialed fall below the rates prevailing in the Commonwealth. Should wealthy outsiders come in, there is certainly room for rents to squeeze more from working households’ budgets.

The HUD definition of rent-stressed — 30% gross-rent-to-income — has very little grounding, being more of a historical accident than the product of deep analysis. It is, in fact, around the average rent burden for renter households in the US. Severe rent stress is often defined as an gross-rent-to-income ratio exceeding 50%. So how does Cville look using that standard?

Essentially, severe rent-stress is largely unknown among working households. And again, Cville does better than the Commonwealth. Student and non-working households cause an enormous distortion. Plenty of room for gentrification to wring more money from the city’s renters. A final way to look at Charlottesville’s rent-stress levels and the massive distortion caused by student households is to see what percentile (of all PUMAs in the county) Charlottesville lands at for each category of household.

When we compare properly, using non-student and/or working households, Charlottesville shows a comparatively low level of rent-stress. The gap between the overall percentage of rent-stressed households and the percentage of households with at least 1FTE that are rent-stressed is extremely wide in Charlottesville — among the widest in all the country. For the 30% threshhold, the gap is in the 83rd percentile of all PUMAs and for the 50% threshhold in the 91st percentile. The gap is not just a salient aspect of the Charlottesville market, but perhaps the single most distinctive aspect of it. And, yet, discussion or analysis is almost entirely missing from HR&A and RHI’s “reports” (we hesitate to honor them with that characterization outside of scare-quotes). Malpratice, bought with our tax money.

FInally, we look at gentrification more directly. In earlier research, we attempted to look at this using IRS and BLS data. We found very little evidence of large inflows of wealthy outsiders. However, those data sources had problems, which we describe more fully in that research (follow the above link). Using PUMS data we could better filter out the student distortion (in this case, students make it look like Cville has many high-earners moving away, because students graduate to high-paying jobs elsewhere), by looking only at “prime age” residents (From 28-68 years old). At the PUMA level, we looked at the ratio of household income for new arrivals (from another PUMA) versus the household income of current residents. The higher this ratio, the more we are seeing people wealthier than current residents moving in. We also wanted to compare newcomers to leavers. That is a little trickier, since Census tracks “former location” a little differently, using “MIGPUMAs” instead of “PUMAs.” MIGPUMAs are slightly bigger. In the case of Cville, the MIGPUMA contains not just the PUMA that covers the city, but the adjacent one that covers some of Southern Albemarle and parts of Fluvanna and Louisa. We look at the ratio of newcomer HHI to leaver HHI at the MIGPUMA level.

The Charlottesville area shows no evidence of inflows highly skewed toward the wealthy or outflows highly skewed toward lower earners. In fact, Cville’s ratios are right in line with the Commonwealth and the USA on both. Zoomtown areas like Asheville, Austin, Boulder, and Naples all show newcomer-to-leaver HHI ratios clearly above 1.0. Manhattan’s Upper East Side shows lower ratios — and this would almost certainly be more dramatic if we looked at 1y ACS data rather than 5y data. We therefore must conclude that Charlottesville has not seen intense gentrifying flows yet. Further evidence comes from looking at raw inflows and outflows by HHI band. We see rather balanced flows across income bands, with actual net outflows at the highest bands (which consist, obviously, of small numbers).

Another related story we hear is that remote work has brought in wealthy people and pushed up housing prices. Using PUMS data we could identify new arrivals and existing residents in remote work. First we looked at how remote work evolved from pre-pandemic to post-pandemic years.

For all the hoopla about remote work, Charlottesville shows up as both low on the rate of remote work post-pandemic and the increase from pre-pandemic levels. Looking for a remote work poster-city? Look elsewhere. We then tried to zero in on inflows of work-from-home workers. We measured the total number of new work-from-home workers that moved in in 2021 versus the total 2018 labor force.

Charlottesville saw an extremely modest level of arriving new remote workers. If you worry about remote workers executing the cost-of-living arbitrage by moving to town, you still have much to worry about. When it comes to WFH inflows, “we ain’t seen nothin’ yet.” Imagine what things might look like if developers create lots of cool, expensive-for-Cville live-work apartments?

Many local observers seem to regard the city’s housing situation without context. It is a bit like a person who has never been to the beach before watching a wave roll into shore. “This must be a tsunami”, the person might think. When it comes to gentrification and increases in housing affordability stress, Charlottesville is looking at a small wave and calling it a tsunami. And in reacting to the wave, Charlottesville might end up conjuring the tsunami.

August 2023