In the year since the unveiling of the first draft of the Charlottesville Comprehensive Plan and its Future Land Use Map, CFRP has worked hard to test the assumptions, both latent and explicit, underlying the plan.  We’ve set out our findings in extensive research available on this website.  We have found that many of the FLUM’s justifications rely on “just-so stories” rather than data:  that Charlottesville is a particularly unaffordable real estate market (it’s not ), that Charlottesville is growing like gangbusters in population (it’s not), that Charlottesville has not produced new housing units (it has, at a rate above VA average), and that Charlottesville’s existing zoning, as practiced, is highly restrictive (again, it’s not).  We have questioned in turn some of the philosophical underpinnings of the FLUM, namely that density necessarily contributes to affordablility (it doesn’t) and that highly local supply results in more affordable housing in the highly local market (it may, but it is highly uncertain). 

The one area where we found pro-upzoning arguments more persuasive relates to environmental impact.  There is certain conjuncture of circumstances where we see a plausible connection between upzoning and environmental preservation.  Consider a city that is experiencing very high job growth.  The job growth drives demand for housing and ability to pay up for it.  The natural economic response is for housing production at the urban boundary – so-called “growing out” rather than “growing up.”  In response the city chooses to enforce an urban growth boundary to preserve open space and to forestall a massive increase in aggregate commuted distance.  This will require housing to be built on limited land inside the growth boundary and will tend to drive up land prices.  In such a situation, increasing use intensity through upzoning may be an effective response.  

The question is: does Charlottesville meet any of those conditions?  The “just-so” story we get from FLUM advocates asserts the following propositions: 1) Charlottesville has an unusually high ratio of jobs to employed population; 2) Charlottesville is experiencing extremely strong job growth; 3) The holders of new jobs are either themselves commuting from ever-greater distances or driving existing jobholders out of town to commute ever-greater distances.  We collected data on jobs, workers and commutes in the Charlottesville area over time to test these propositions.  The short answer is that (1) is true but a mere artifact of spatial geography; (2) is not true, as Cville job growth is around the US average; and (3) is false and provable so to a surprising degree – aggregate commuted distance to city jobs has barely grown at all even as job growth has moved in line with US averages.  Further, we have demonstrated elsewhere that Cville land prices are actually comparatively low. The urban growth boundary is outside of the city and therefore outside of its control.

Before we get deeper into our results, a word on the main data sources we used.  The Longitudinal Employer-Household Dynamics dataset is collected by the Census Bureau from administrative data.  The dataset captures the location of jobholders and their primary jobs down to the Census Block level.  The latest year available is 2019.  We used 2010 and 2019 as our main comparison years.  Our wider dataset included all records where the jobholder location or the job location lay within the Charlottesville metro area.  We used Census American Community Survey data for household characteristics and housing prices.  Finally, we used 2010 and 2020 decennial census redistricting population and housing unit counts for those variables.  Commuting distances were calculated using linear distance extracted from the Census’s TIGERline geographic shapefiles.  The LEHD data has some limitations.  As it is administrative data, confidence intervals are not available as with survey-based data.  The administrative data may miss some employees (such as off-the-books workers).  In addition, we find that some worker/workplace pairs are implausible (i.e. workplace is Washington D.C, home is southern Albemarle County) and likely a result of remote work or situations where the employer of record may be a home-office address, while the worker is at a branch office.  Some analysis, therefore, included only worker-workplace pairs where the residence side of the relationship was within an “augmented” MSA that included the formal MSA plus somewhat further outlying areas such as Staunton, Waynesboro, Buckingham County, Louisa County, and Augusta County.  We established a cut off at the county boundary where housing prices appeared to start rising again (usually due to proximity to Richmond or DC).  We felt that people who lived in more expensive distant area were not doing so because of housing affordability challenges closer to Cville.

So, on to the first pro-FLUM proposition: Charlottesville hosts an unusually high ratio of jobs to employed population.  The idea here is that such a ratio implies that housing costs are forcing people to live away from their jobs.   Charlottesville has, according to the 2019 LEHD data, just over 40,000 jobs and just over 20,000 employed residents.  Now, obviously, if we compare that to the United States as a whole, the ratio looks high.  For the US, the ratio is just about 1 (almost by construction).  But the problem here is that the ratio is highly dependent on the physical size of the jurisdiction under consideration.  Jobs and residences tend to cluster, and to cluster in distinct areas.  Therefore, if you zoom in enough, you will find the ratio go to either zero or infinity.  We have a more technical discussion of this effect here, for readers who are interested.  Charlottesville is a physically small jurisdiction that hosts institutions that serve the entire region.  It would be odd if it didn’t have more jobs than population.  It is instructive to compare the Cville jobs-to-employed population ratio to some other cities in the region. 

You can see that it is not in fact that high for a college town.  It is above Richmond, but Richmond is a physically larger city (60 sq miles as opposed to 10 square miles; Richmond represents almost three times the physical share of its MSA as Charlottesville does of its MSA).  

A latent assumption of the proposition that Charlottesville’s jobs-to-population ratio suggests housing problems is that commuting is mostly “one-way” from residences outside the city to workplaces within it.  In fact, that is not true.  There are more employed residents of Charlottesville working outside the city than insideaccording to LEHD data.  Now, we think that some of this may be a result of the assignment issues we alluded to above.  But even if we consider only jobs in the MSA, there are almost as many Cville residents who work in the MSA-ex-city as work in the city (87%).   If we compare this metric of directionality to the same cities as above, we find that Cville does not exhibit the high stay/commute-out ratio of an undisputedly expensive real estate market like DC.  We added in an acknowledged high-cost/low affordability college town, Bozeman, as another comparator, and predictably the stay/commute-out ratio is far above Charlottesville’s. Basically, Charlottesville is a cheap enough real estate market that many people are willing to live inside it even if their job is elsewhere.

Charlottesville, then, does have a high ratio of jobs to employed population.  This ratio, however, is a fatally biased metric and reveals very little about the housing market.  Even if we choose to consider this metric in spite of its shortcomings, we find that Charlottesville’s high metric is very similar to the ratio prevailing in other small cities with regional institutions and to other college towns.  Finally, the prevalence of outbound commutes undermines the narrative of an expensive, jobs-rich city surrounded by cheap, jobs-poor hinterland.  It also raises question about whether resident inflows driven by upzoning (should they materialize) would necessarily turn into environmentally friendly, short within-city commutes.  We will see more on this point when we look at the question of job growth. That the numbers belie the FLUM narrative perhaps explains why it appears the consultant conveniently conflated city and MSA numbers in a way that made job growth look falsely high relative to housing production.

So, what about job growth in Charlottesville?  To hear the pro-FLUM rhetoric, employment in town is growing as inexorably as our English Ivy.  In fact, Charlottesville job growth has been almost precisely in line with the national average.  But more interesting is that job growth in the city has fallen substantially behind that in Albemarle County.  Those who have experienced the quality of services, policy-making and political culture in the city can probably understand why employers might prefer to locate over the county line.

Interestingly, the growth of working population in Charlottesville with a job in the city has growth faster than either jobs or working population in total: another data point that flies in the face of the idea that the pattern of economic development consists of job growth in the city, accommodated by workers making ever-longer commutes.  It would, on the other hand, potentially support the idea that new resident-workers might disproportionately work close to home.   We don’t want to overstate the case, though: the increase in city residents working in the city only represented about 60% of the increase of city working population over the time period we examined. 

While FLUM advocates expressed particular concern about public sector workers (e.g. teachers and first-responders) no longer being able to afford to live in town, they adduced no data in support of it. It ought to have been quite easy for the city to consult its payroll provider to see what percentage of W-2 addresses on the payroll fell in various geographies. Then again, given the level of decay of the city’s administrative apparatus and analytical capability, maybe “easy” wasn’t easy enough. The LEHD data, while not capable of giving as precise an answer, does offer a breakdown of public sector vs private sector jobs and jobholders. We find support for the exact opposite conclusion. While private sector job growth far outstripped public sector job growth, the count of in-city-working residents with public sector jobs grew faster than the count of in-city-working residents with private sector jobs.

The bottom line on job growth in Charlottesville is that the pace is very average and seems to be below that of Albemarle County.   Of the job growth in the city where the identifiable residence was within the “augmented” metropolitan area, substantially all of it was accommodated by increase in workers resident within the city itself or Albemarle County. 

As an aside, we note what a truly “hot” job market looks like. Over the same period that saw 17% job growth in Charlottesville, Austin experienced 27% job growth at the city level and 38% at the MSA level. Bozeman posted 38% job growth at the city level and 45% at the MSA level. We appreciate the spirit of civic boosterism that inspires the many Cville-lifers among supporters of the FLUM to attach superlatives to their hometown, but, as is the case with real estate values, job growth qualifies for no such distinction.

This brings us to the final proposition, namely the claim that city workers are moving ever farther from the city due to housing affordability issues.   The job growth numbers already strongly suggest the weakness of this claim.  It is instructive to look at a map of the area, with each Census Tract colored according the change in the number of city workers resident in that Tract – that is, commuters.  It is immediately evident that in absolute terms, the very strongest growth comes from just outside the city limits.  This should not surprise anyone who has seen the vibrant residential construction environment in the Rio corridor.  But what reader will perhaps find surprising is that many outlying areas have seen absolute decreases in commuters.  Parts of southern Albemarle, Greene County, and Nelson County fall in this category.  Staunton and Waynesboro, on the other hand, are two more distant areas that have seen modest increase.  But overall, the “heat map” is glowing right around the city, not at the exurban fringe.

Another way to map the data is to plot on the map the ratio of commuters to Charlottesville City from each tract as a fraction of housing units in the tract. For example, a score of 0.4 would mean 40 commuters per 100 housing units. On the map below you can select 2010 or 2019 numbers. You can see that the “heat” on the map tends to move closer to the city from 2010 to 2019. The closest parts of Albemarle County have become more intensive feeders to city jobs, while outlying areas have changed very little or become less intensive.

If indeed there were an overwhelming trend of workers getting pushed further and further from city jobs, we would expect to see that areas that added commuters to the city would have longer average commute distances to the city than average, and likely longer commutes than areas that lost commuters.  We looked at the block group level and divided blocks groups into those that added commuters to the city and those that lost commuters to the city.  We then calculated a change-weighted average commute distance for “increase” block groups vs “decrease” block groups.  We found that “increase” block groups showed almost no difference relative to “decrease” block groups.  Increase block groups had on average only very slightly longer commutes than the overall average for the augmented MSA (including the city) and for the MSA-ex-city.  This is at variance with what we would expect to see if the FLUM “conventional wisdom” were true.

We looked at the same analysis using tract-level housing prices.  Again, what the FLUM narrative would predict is that “increase” areas should have lower home and rental prices than the overall average.  We found a similar pattern.  The median home price for “increase” areas fell below that of “decrease” blocks.  Both were slightly below the overall average. None of the magnitudes of difference were large.  For rents, there was no significant difference between categories. This pattern undercuts the claim of large movements of city workers to cheaper areas (as opposed to just more distant areas).

One way to put the two questions – of commute distance and of housing cost – together is to chart the median home price, median rent, and change in commute flow to city jobs for various distance buckets within the augmented MSA. 

While it is true that the most distant “bucket” is considerably cheaper than the closest “bucket”, the detailed pattern is more complicated and sheds doubt on the simple narrative of flight to distant exurbs for affordable housing. 

First, the price of housing increases as commuting distance increase from zero to 15km.  It then decreases from 15km to 30km, but after that flattens out.  A trough a somewhere around 30km does not comport with the story of large numbers of workers relegated to 75km commutes by housing markets.  And the change in commuter flows from 2010 to 2019 also contradicts this narrative.   The largest increases, both absolutely and proportionally, took place in the zero to 10km range.  There were net flows from the more expensive 10-15km range to the less expensive 15-20km range, and likewise from the 20-30k range to the 30-40km range, but these were modest.  Bottom line: no evidence of a massive flight to the cheap exurban fringe.

The Internal Revenue Service makes available a data set of tax returns by county, including returns of filers who moved from one county to another. While this dataset does not get at commuters, it can give us a check on the commute data in considering the question of economic migration. We looked at the data on the number of filers who moved from Charlottesville to one of the surrounding counties in the 5-tax-year period of 2016-2020. In addition, we compared the mean Adjusted Gross Income (AGI) of filers moving to vs moving from Cville for each of these counties. If the “economic flight” narrative is true, we would expect to see large net flows out of Charlottesville and also systematically lower incomes for those who moved out of Charlottesville versus those who moved in. By now, you can probably guess whether the data supported the narrative.

Net flows to all the surrounding counties besides Albemarle were tiny, amounting to only about 100 filers per year. To put that in context, Charlottesville has nearly 30,000 filers. And even Albemarle represented a net flow of only 200 filers per year. What about the composition of flows? Were those filers who left Charlottesville showing lower income than the filers who came to Charlottesville? That’s what one would expect if, first off, people with low earning power were being forced by economic circumstances to look for cheaper housing outside the city and, secondly, if one could only move to the city if it meant earning more money to meet the higher cost of living.

The chart clearly shows that for most of the counties (bars in green), the mean AGI of those moving out of Charlottesville to that county was higher than the AGI of those moving to Charlottesville from that county. Staunton had such small flows that we did not get a usable mean AGI for inflows. Only Greene County and Waynseboro city had ratios above 1, and they were barely above 1 .

Finally, we take a quick look at the summary statistics for average commuting distance and aggregate distance commuted for intra-augmented-MSA flows to city jobs in 2019 vs 2010.  The average commute distance for city jobs held by ASMA residents has in fact declined from 15.5km to 14.2km from 2010 to 2019.   When we tally up the aggregate commuted distance we find that it has lagged substantially behind the growth in jobs.  In fact, the increase is surprisingly small.  Hardly the picture of environmentally catastrophic commuting sprawl that FLUM proponents would have us believe exists.

As a check on the LEHD-based numbers, we compared the Virginia Department of Transportation’s “Vehicle Miles Traveled” (VMT) statistics for 2010 and 2019. We wanted to make sure that the pattern looked similar, namely that vehicle miles traveled lagged behind population and working population growth. We especially wanted to see the overall VMT for Albemarle and Charlottesville. Any new commuters to the city from outlying areas would necessarily have to transit Albemarle and part of the city, so if there were a large increase in long-distance commuters, we would expect to see not only VMT increases for the source jurisdictions, but even larger increases for Albemarle and Charlottesville.

The VDOT numbers coincide with our LEHD-based analysis. VMT lags substantially behind in-jurisdiction population growth and working-population growth in Charlottesville and Albemarle. There is no evidence of large VMT increases in Cville+Albemarle coming from workers driving in from outlying areas. Outlying counties that saw large percentage increases in population working in Cville MSA (the numbers in the chart above, to be be clear, are for workers with job location in the MSA counties, not in Charlottesville City specifically) did not see concomitant large increases in VMT in county either. The large percentage worker increases amount to very small absolute increases because of the extremely low 2010 base. As we saw above, there just aren’t that many workers in these outlying counties coming into the MSA (and even fewer into the city specifically).

So, are we saying that our whole area is an environmental and commuting paradise?  Absolutely not.  Public transit leaves much to be desired, for one thing.  For any amount of commuting distance, we can certainly try to have more of it covered by public transit.  And the story for Albemarle County is quite different from that of Charlottesville City.  While aggregate commuted distance has not increased quite as fast as overall jobs in the county, it has come a lot closer (15% commuted distance growth vs 21% job growth) than in the city.  In addition, “increase” areas have substantially longer commutes than then average for county jobs.  But the city’s zoning cannot solve Albemarle’s issues, such as they are; nor should city residents be asked to accept the risks of gentrification and undesired transformation to bear the environmental burdens of Albemarle’s economic development strategy.

We want to close with some brief thoughts at one level higher of abstraction — not on the result of our analysis, but rather the question of what analysis ought to have been done as part of a Comprehensive Plan design process. We believe that whatever one might think of the outcome and conclusions of our various analyses, it is difficult to argue that the designers of a Comprehensive Plan should dispense with these sorts of studies. A CP should be informed by an accurate picture of the housing market, of income distribution, of the actual operation of the zoning ordinance in real-world conditions, and the pattern of commuting in the region. Charlottesville’s CP process proceeded with almost none of this foundational work. It was not for lack of spending. Was it because the prime contractor, RHI, doesn’t know this sort of work is important? Doesn’t know how to do it? That RHI was selected from a list of candidates responding to the RFP of precisely one — RHI itself submitted the only proposal — means their “victory” tells us almost nothing of their capability.

Still, RHI is an established firm, and it is difficult to believe they lack elementary understanding and competence. That suggests to us that our Planning Commission did not ask RHI to do the work; or worse yet, asked them not to do it. Perhaps a better result could have been expected had the RFP contained a well-thought-out, detailed list of technical requirements and deliverables rather than leading with a polemical gust of copypasta cribbed from the ideologues at the Legal Aid Justice Center and their fellow-travelling astroturf organizations. Luckily, a CP is not ordinance. There is still time to rescue the city from throwing itself down an ill-advised policy path. We don’t expect the city to simply use our analysis. After all, we didn’t get paid a million dollars to do it. We ask the city to go back and hire the right advisors to do the work before the City Council passes a zoning ordinance and other legislation that will turn the CP into operational reality.

May 2022