How housing supply/demand imbalance remained for an entire generation

Chuck Marohn puzzled a bit over housing costs over at Strong Towns last week, writing that “You can’t sustain increasing demand while also sustaining increasing prices and increasing supply.”

Wiltberger St NW

Would you pay $700/sq. ft. for this 2-bedroom alley house? Somebody did, paying 170% above its 2006 tax value. Sure, valuation growth like that isn’t sustainable, but what about our cities is?

You can if (1) demand grows just a bit faster than supply, or if (2) incomes are growing, or if (3) slightly more income can go towards housing — and certainly so if all three occur. Indeed, all three of these dynamics have sustained housing price inflation in gateway cities over the past generation.

This inflation has been politically possible because many existing residents (and thus voters) are sheltered from the resulting affordability crisis. Only a minority of people are exposed to housing affordability; most current residents are sheltered from price increases, having purchased or rented their housing at yesterday’s market prices. It’s pretty much only in-migrants who have to pay today’s housing prices, and since they’re migrants, they don’t vote. In-migrants are also a surprisingly small share of Americans: in any given year, fewer than 3% of Americans move across state or national borders.

1. Between job growth, smaller households, and natural growth, housing demand is increasing faster than population (and construction) in many metro areas. This has been the case in California for decades; the LAO’s 2015 paper estimates that since 1980 (my entire lifetime!), California has built 100,000 fewer units every year than it should, and yet (a) demand to live in California continues, although definitely abated; (b) prices have skyrocketed; (c) construction has added some new supply.

2. Median incomes nationally have been flat for the past generation, but incomes in the richest gateway cities have been soaring — especially at the top of the distribution, due to rising inequality. The minority of households that are exposed to high prices may very well be able to afford those prices in these cities, explain Gyourko, Mayer, and Sinai in their paper on ‘superstar cities’: “Recent movers into superstar cities are more likely to have high incomes and less likely to be poor, than recent movers into other cities… In short, residence in superstar cities and towns has become a luxury good. The cities’ increases in housing price appear to outstrip known productivity increases and the value of any additional amenities.”

Since only a small proportion of housing units trade hands each year, cities with rising incomes at the top and relatively few houses available (e.g., the “superstar cities”) see “new money” outbidding others for those few units, pulling prices up. Because house prices are based on comps, prices for other houses also rise. As Matlack and Vigdor write, “In tight housing markets, the poor do worse when the rich get richer.”

I know this seems insane, but income inequality has gotten so far out of hand that in many cities super-luxury housing is under-supplied, with tremendous consequences all the way down the housing ladder. There are over a thousand Bay Area households with million-dollar bank accounts for every single house that came on the market last year in Atherton, the choicest of Bay Area towns. Hence, house prices in Atherton have doubled in four years.*

3. Metro economies have evolved in lots of small ways to cope with higher housing prices at the margin. At first glance, “the poor will always be with us,” but in reality metro areas differ very substantially in terms of their economic makeup. Having moved from low-cost Chicago to high-cost DC, I’ve noticed that this slowly-accumulating, giant gift to high-cost-regions’ landlords has been cobbled together by squeezing a few dollars here and there from other sectors:
– Higher labor costs: the minimum wage here is about 15% higher, and high-labor-input services (like haircuts) cost substantially more here, because the staff earn more.
– A shift towards higher-wage work and reduced labor inputs (see #2 above). There are, of course, lots of well-paid jobs in DC; nearly half of households here earn over $100K. Many dual-income “power couples” who have no problem with the local cost of living. But there are surprisingly few on-site support staff for them, and instead there’s often off-site help. Even in labor-intensive industries like restaurants, on-site prep work can be minimized by relying on commissaries and distributors based in cheaper cities. (You can forget about Jacobsean “import substitution.”) Anecdotally, I’ve heard that employers are willing to make do with thinner staffing here than elsewhere.
– People work more; DC’s female labor force participation rate is 15% higher than Chicago’s.
– Housing itself can’t be substituted (everyone needs somewhere to live), but houses can be. People downgrade their locations or living standards, living in smaller or lower-quality housing units in less desirable neighborhoods than they otherwise would. They also “pay” for housing with long commutes, often from what are technically other metro areas.
– People borrow more. DC has more mortgages and higher student-loan bills than any other metro.
– People spend more on housing, and less on other goods and services. Brookings’ Natalie Holmes notes that the 20th-percentile unit in DC costs 48% of a 20th-percentile income, vs. 38% for a 20th-percentile individual in Denver.

That these coping mechanisms exist by no means implies that high prices are benign. From a local economic development standpoint, high housing prices don’t just deter potential employers, but also vacuum up dollars that could be more useful elsewhere in the local economy. Rent checks, unlike haircuts or restaurant meals, don’t have big job multipliers. As a Global Cities Business Alliance report puts it:

Citizens are spending money on accommodation that they would readily divert to goods and services if their housing costs were lower… the money ‘trapped’ in the housing market runs to billions… Unleashing this spending would in turn boost business revenues and create more jobs. Assuming that businesses were to channel all additional revenue into employment, we estimate that Beijing could generate more than 400,000 new jobs, Mexico City more than 200,000, São Paulo more than 143,000, and Hong Kong nearly 148,000.

* Chuck’s follow-up post posits that property owners are speculating on upzoning. This line of reasoning is beloved by so-called “SF progressives,” who relish pinning the blame for everything upon evil, greedy developers and the obnoxious “kids these days” who inevitably fill their apartments. Yet this densification/speculation theory cannot explain the skyrocketing housing prices that are at the very epicenter of America’s metro affordable housing crisis — in places that have zero multifamily growth and zero transit investment, but LOTS of high-wage jobs, like Atherton, Menlo Park, and Palo Alto in Silicon Valley, or Chevy Chase in Maryland, or the Hamptons. Atherton is the most extreme example: the town banned all multifamily housing and sued to stop transit, and yet house prices have doubled in four years.

Perhaps, instead of transit-oriented speculation, exclusionary, single-family-only snob zoning has left supply and demand imbalanced. Believe it or not, the demand for $3M houses in Atherton vastly exceeds the supply of $3M houses, so the $3M houses have been bid up to become $6M houses. I know this seems insane, but there are over a thousand Bay Area households with million-dollar bank accounts for every single Atherton house that came on the market last year.

There are also many fashionable urban neighborhoods where housing prices have spiraled even while housing unit density is declining: the demand for mansions is so high that humble apartment buildings get demolished for glamorous single-family houses. (Once again, life imitates the Onion.) This was even the case in my onetime home of Bucktown in Chicago, where the ward boss infamously handed out spot rezonings upon “request”; in theory, these could have been used to add units, but in practice the McMansions just got fatter.

Some guesses as to implications of autonomous vehicles

Autonomous vehicles, driverless cars: ask two people what they think, and it seems like you’ll get three opinions. Here are my reactions to four recent publications on the topic — keeping in mind that previous reports of distance’s death were an exaggeration. (As CBRE’s Revathi Greenwood notes, vehicle speeds won’t change, and so Marchetti’s Wall still remains. Even if the drudgework of driving is taken away, travel time still has a cost, and we’d rather be at our destinations already — e.g., “are we there yet?”)

WSJ (columnist Christopher Mims):

  • AVs will be limited to small areas for the foreseeable future. “We’re likely to see vehicles that don’t require drivers but can only operate on a fixed, well-mapped route in cities with fair weather… the consensus of those I interviewed is that it will be many years before we get cars that can truly go anywhere.”
  • Existing trials (Singapore, Pittsburgh, Babcock Ranch), which are limited to relatively small, intensively researched areas that are frequently remapped. Level 2/3 autonomy will remain limited to expressways, which have a protected ROW.
  • Echoes some of Recode’s timeline (perhaps similar sources were interviewed).
  • Takeaway: Autonomous shuttles will appear within campuses, urban districts, and planned communities, initially as “walk extenders.” “Robot valets” will enable more remote parking and reduced parking footprints. Freeway driving may shift to autonomy, but uptake is limited by consumer acceptance (see next).

Kelley Blue Book consumer survey:

  • Americans are still broadly uncomfortable with the idea of Level 5 autonomy.
  • Level 4 autonomy is most popular with current US consumers, who still want to be able to take the wheel. Level 3 seems less comfortable than Level 2.
  • However, key early-adopter groups feel more comfortable with complete autonomy: luxury car buyers, consumers with experience with Level 2 AVs, and people used to the backseat: ride-hailing customers and teenagers.
  • Takeaway: The transition to AVs is dependent upon social acceptance, and currently many Americans want to maintain the status quo. The transition might take a while (more Americans will have to try AVs), but may be steep once it happens.

Rocky Mountain Institute forecast:

  • Mobility services in major US metros are a potential $120 billion annual market by 2025, including $60 billion just in large Sunbelt metros.
  • Because AV and EV technologies reduce operating costs and increase capital costs, they will find broad acceptance in high-utilization fleets first, where their low costs will subvert the individual-car-ownership paradigm. (2017’s EVs will be cheaper for fleets than gas cars.)
  • AVs will cut the cost of rides by 60% to be cost-competitive with car ownership by 2018, with another 60% decline in costs as economies of scale are realized. The switch from personal cars to AV fleets will occur between 2020-2025, with long-term demand for cars falling to ~6 million.
  • Lower mobility costs will result in a $1 trillion annual consumer surplus to be spent on other sectors. (Keep in mind that spending on autos has a low multiplier effect.)
  • Even if VMT doubles and more power plants are built, these two technologies will result in sharply lower CO2 emissions (nearly -1 GT CO2E by 2040 = ~13% cut in today’s emissions).
  • Takeaway: Parking demand may sharply decline, but what parking is left will need significant EV infrastructure. Loading/valet zones will quickly need to be implemented. Consumer spending on cars could be pivoted to other spending, like higher-quality real estate.

City Observatory (Joe Cortright) [part 1] [part 2]:

  • RMI’s cost estimates of <$0.50/mile are roughly in line with other published estimates, with lower costs associated with smaller/lighter vehicles. This is lower than the per-mile cost of not just driving, but even short transit trips.
  • However, $0.50/mile is much higher than the perceived $0.15-$0.20/mile marginal cost that most Americans assume for private-auto trips. (Most Americans only consider the cost of gas when driving; costs such as depreciation/wear, insurance, repairs, monthly parking, and wasted time are all considered sunk.)
  • “Pay by the slice” mobility, like car-sharing, tends to encourage shorter trips. Pricing will probably be more, not less complex, with various “surge” surcharges that use information to optimize the balance between travel demand and supply.
  • Rush-hour capacity will still be an issue, especially in high-density downtowns. Rail transit, walking, and cycling will still move more people in less space.
  • Takeaway: Mobility won’t be “too cheap to meter,” as optimists once said of nuclear electricity. As such, central locations will still matter, even if price differentials flatten somewhat. (TNCs are already “filling in the lines” between transit corridors and increasing the value of secondary urban locations.) Whether dense downtowns built around rail/walking remain useful is an open question.

What everyone agrees upon is that this is the first huge shift in metropolitan mobility since the 1940s-1980s shift towards mass car ownership. It’s important to remember that American suburbia is a political and social construct, not a fact of life, and that policies put into place immense structural supports for American suburbs.

Long-timers vs. newcomers: Census-reported rents are meaningless in some places

Earlier, I wrote about how the CNT/HUD Housing + Transportation Affordability index (HTA), while a very worthy addition to discussions around whether households can affordably live in suburban vs. urban areas, has a denominator problem at metropolitan boundaries, or even within metros which have severe income inequality. Yet LocationAffordability also has a numerator problem: For housing costs, it relies upon another Census data point which has One Weird Quirk.

Daniel Kay Hertz wrote an excellent “Definitive Field Guide to Median Rent Statistics” for City Observatory recently, noting that ACS-reported rents “Can’t answer: What is the median rental price facing people on the rental market today?”


DeKalb Market, Long Island University

The tract on the left (in downtown Brooklyn) is public housing, with a reported median monthly rent of $768; the tract on the right $2,467. But for someone just moving to town, the only vacancies are in that fancy new building. 


The rents that people report paying to the Census ACS are probably true, but in a few cities they have very little relationship to what vacant apartments are renting for. In particular, as Michael Lewyn flagged (and as we wrote up in Streetsblog a while ago) cities with rent control can look amazingly cheap:

[B]y looking at average rents, which in some cities include many rent-stabilized units, the calculation doesn’t necessarily capture what someone searching for shelter is likely to pay. If you’re trying to find an apartment in New York now, getting a place for the average rent would probably be extremely difficult.

Unlike the denominator problem, which shows up at a macro level, this discrepancy only really shows up at the micro level. I just noticed it in the recent “WalkUP equity ranking” from GW’s CREUA, which (based on HTA) found suspiciously low rents reported in some very upscale neighborhoods. I suspect the typical rents found are those paid by a small number of long-term residents in subsidized or rent-controlled units (set years ago), rather than those paid by the residents of the new luxury apartments.

HUD’s new Small Area Fair Market Rents are reported down to the ZIP code level (and have the big problem of not being available everywhere), which is a much larger level of analysis than Census tract, but they’re an attempt at figuring out a systematic answer to this problem.

McMillan isn’t next to Metro, which is less of a problem than you think

McMillan Reservoir

You can see the Capitol Dome from here. Photo by Eric Fidler, via Flickr

Yes, the McMillan Sand Filtration Site is one mile (from either end of the site) to the Red Line. It’s even 0.6 miles to the nearest express bus route (Georgia Avenue’s 79), and key network improvements are still in the planning stages. Yet from the point of view of someone who wants to reduce auto dependence (and the concomitant pollution, injury, and sprawl), what matters most is that MSFS is close to downtown, rather than close to Metro.

Transportation planning research has consistently shown that location relative to downtown and to other land uses is far more closely associated with the amount of driving than location relative to transit. Ewing and Cervero’s definitive 2010 meta-analysis (cited by 679 other scholarly articles) examined over 200 other studies, then combined the correlations found by 62 different studies:

Yes, it turns out that the number of miles that people drive is four-and-a-half times as closely correlated with the distance to downtown than with the distance to a transit stop. This strong relationship between driving and distance to downtown is borne out in local survey research by MWCOG/TPB. Note that whether an area has Metro access (like Largo or White Flint, vs. the Purple Line corridor) doesn’t actually seem to impact the number of drive-alone (SOV) trips.

Some suggest that development proposed for this site should instead go elsewhere. If the development is denied, those residents and employees and shoppers won’t just disappear, they’ll just go somewhere else. They won’t go to superior locations even closer to downtown and Metro (because those are so very plentiful!), but rather to far inferior locations. For instance, the life-sciences employers might choose an alternative location within our region that has already approved a similar mix of uses — such as Viva White Oak, Inova Fairfax, Great Seneca Science Corridor, and University Center in Ashburn, all of which are much further from both downtown and Metro.

This isn’t just the suburbs’ fault. Within the District, even more intensive development than what’s proposed at MSFS has already been given the go-ahead at locations such as the Armed Forces Retirement Home, Hecht Warehouse, and Buzzard Point. All of those sites are also inferior to MSFS from the standpoint of not just transit accessibility and distance to Metro Center, but also on all of the other factors shown to reduce VMT.

If the “Reasonable Development” types truly do care about reducing driving, I must have missed their years of caterwauling over the approval of all these other sites — not to mention the countless suburban developments that together pave over 100 acres of open space every single day in the Chesapeake Bay watershed. That’s why I give more credence to the people who do actually care about paving over the region, like the Piedmont Environmental Council — a/k/a the Coalition for Smarter Growth.

Brand-new timber loft offices, now for lease

N. Vancouver Ave. frontage of One North

Last week, I published a ULI Case Study about One North in Portland, an architecturally inventive response to my previous speculation that “new-build loft offices could be popular in similar downtown-adjacent submarkets, and transformative for Sunbelt cities where sparse ‘warehouse districts’ have little competing product.”

Indeed, the anchor tenant at One North is a mid-sized tech company that had outgrown its space in Portland’s Central Eastside. As in many other growing cities, there just wasn’t a cool old loft big enough, so instead they found a cool new loft.

T3 Minneapolis under construction

Southwest corner of T3 Minneapolis, showing CLT column stacked above concrete podium

I also had a chance last week to check in on T3, Hines’ new cross-laminated timber office building in Minneapolis. Less than a year after groundbreaking, the structure is complete and the facade is almost completely hung — almost a year faster than a comparably sized concrete building takes to build. The superstructure took less than 10 weeks to build.

The model is so successful that Hines is now replicating the T3 building at another location that’s even hungrier for lofts: Atlantic Station in Midtown Atlanta.

Here in DC, one great location could be the PDR-2 zoned land (90′ height permitted with setback, no residential) on the west side of the Met Branch Trail along Eckington and Edgewood, one of the hottest corridors in town. Another could be around Union Market/Gallaudet, where JBG’s Andrew VanHorn says “we see the tenant base there evolving. The pre-lease opportunities we’re talking to for our office building are all private market, very young companies, as far as their employee demographics.” Or maybe this is what his firm has in mind for the “creative loft office” at RTC West.

An aside: this is another strike against “Investment Ready Places.” It sounds counter-intuitive, but it’s easier to move buildings to people than to move people to buildings. The “good bones” that economically unviable places have can have “good enough” replicas in New Urbanist settings like Atlantic Station and Reston Town Center. Not to mention that building all of the new infrastructure to overcome IRPs’ deficient locations, and then rehabilitating their buildings to code, would be much more expensive than just building anew in prime locations. It’s cheaper and easier to build new lofts in Reston than to rehab lofts in West Baltimore, and to build the new rail connection that would make West Baltimore feasible for NoVA’s growing companies.

Cities are built around people, not the other way around.

One year’s progress at the Wharf

I’ve been taking a time series of photos of construction at the Wharf from two vantage points for several months now: from the Case Bridge (under the sign for D St.) and from Banneker Overlook (by the trash can).

By complete coincidence, I seem to have snapped photos on 22 July 2015 and 22 July 2016. In July 2015, excavation was wrapping up and the tower cranes were just arriving to start building back out of the hole. In July 2016, the hotel towers appear to be almost topped-off.

Cranes at the Wharf, 22 July

Banneker Overlook, July 2015. There are still pile drivers on the site to drill the foundation, but the first tower cranes had just arrived.


Cranes at the Wharf, 2016-07-22

Banneker Overlook, July 2016. 950 Maine Avenue is front and center; below the trusses in the middle of the block will be a 6,000-seat venue, wrapped with two apartment towers.


Cranes at the Wharf, 22 July

Case Bridge, July 2015. Water-side construction of the piers was still taking place, and the excavation roads were still in use.


Cranes at the Wharf, 2016-07-22

Case Bridge, July 2016

(This isn’t really a Friday photo post, but it’s Thursday and I’ll be on vacation tomorrow.)

Friday photo: A balcony to nowhere

Twinbrook Hilton, faux John Portman

Twinbrook Hilton, Rockville, Md.

Architect John Portman was fond of putting “balconies” around the edges of his hotels’ massive atria. Perhaps this was part of the concurrent “conversation pit” trend that afflicted malls of that era, maybe people really did just pause and gawk at the tremendous volumes while walking to/fro the elevators, or maybe he liked the “columns around the Forum” look that they lend to the interior. Yet… nobody quite knows what to do with them today.

Who, after all, has need to stop and gather in a hotel hallway, the very definition of a neither-public-nor-private space? Why not go to the lobby, if it’s a public conversation, or into a room?

At the faux-Portman Twinbrook Hilton in Rockville,* the management was sick of complaints from atrium-facing rooms (one-third of the total) about nighttime noise. One option was to replace the atrium-facing windows with airport-style noise-insulating glass, but ultimately it proved cheaper to just wall off the entire atrium with a curtain-wall system – including putting these silly aluminum doors to wall off the balconies. Maybe someday they’ll put furniture out there, which surely will just gather dust. Or maybe there really are people who use these spaces, like maybe gossipy 8th graders on school trips who don’t want to keep their roommates/chaperones awake. (Isn’t that what Snapchat is for?)

Of course, that particular solution is still better than the temporary option I once spotted at the true-Portman Bonaventure, apparently aimed at LA’s exhibitionist-fitness-enthusiast crowd:

Pod people

Westin Bonaventure, Los Angeles, Calif.

* I remember once staying here as a child and getting evacuated from the building by an overnight fire alarm, in my first (but definitely not last) high-rise fire alarm experience.