California’s big cities, from San Francisco to Los Angeles, have been struggling to contain spiraling housing costs, leading to an economic squeeze on the middle class, gentrification and displacement of low-income areas, and overall income inequality.
The debates that have raged in response, however, have been largely untethered from common and accepted data about what is actually causing the increases. Affordable housing advocates want limits on new development and more subsidies for low-income housing, arguing that adding to the supply will simply encourage more displacement. Renter advocates meanwhile want more construction overall to boost the supply and stabilize prices for everyone.
In an impressive display of independent scholarship, Eric Fischer at the blog Experimental Geography stepped into this data void. After pouring over apartment ads in the City of San Francisco since the 1940s, as well as other economic data, he figured out how to predict rents in the city in any given year.
The bottom line: the three big factors are jobs, salaries, and housing supply:
[I]t is possible to model year-over-year change in rent in terms of year-over-year change in employment, wages, and housing construction. In this case the best fit says that a 1% increase in employment means a 0.95% increase in rent, a 1% increase in wages means a 1.74% increase in rent, and a 1% increase in the housing stock means a 1.7% decrease in rent. It’s the same basic idea, but the magnitudes are different. I don’t know if it is any more correct than the first model, or if they are both bouncing around within the same uncertainty.
Basically, it boils down to the economy and available housing stock. So if we want to dramatically decrease housing prices to make the city more affordable to a wider range of income groups, here is what Fischer suggests:
Can we roll the clock back 35 years, to when the CPI-adjusted median rent was approximately one third what it is now?
It will be very hard. If the (first) model is correct, it would take a 53% increase in the housing supply (200,000 new units), or an 44% drop in CPI-adjusted salaries, or an 51% drop in employment, to cut prices by two thirds. A steep drop in salaries or employment would also be devastating to the ability of people to afford the new lower prices. It is enough to make you believe Randal O’Toole that affordability can only be achieved by continued outgrowth, as San Francisco could do in the early 1950s.
200,000 new units is an incredibly tall order, given the land constraints and certainly the political opposition to densifying the city at that scale (Fischer describes this concern as relating to “visual stability” of the city). As a result, achieving rent decreases in San Francisco is probably off the table. But rent stability is still achievable. As Fischer assesses:
Therefore, if price stability is the goal, the city and its citizens should try to increase the housing supply by an average of 1.5% per year (which is about 3.75 times the general rate since 1975, and with the current inventory would mean 5700 units per year). If visual stability is the goal instead, prices will probably continue to rise uncontrollably.
Fischer’s work is a big contribution to the debate and ultimately reinforces the notion that improving housing supply is our best bet for stabilizing rents. But his study does not address the regional forces at work. San Francisco is a kind of island, stuck at the top of a peninsula. So what would the impact be of increasing housing supply across the bay in Oakland, Berkeley, and San Rafael, and then down the peninsula in Silicon Valley? If those areas boosted supply at a significant clip, would that have a similar effect in San Francisco of building those 200,000 additional units within the city?
It would be a helpful next step for this research. But in the meantime, we now finally have a good sense of how much price increases are due to the economy and how much due to supply. For that, we can thank Fischer — and hope that he continues on this research path.