The Future of Housing in San Francisco
Predicting 5 years of new housing construction
I’ve only lived in San Francisco 5 years, but since I’ve moved here I keep telling myself “the housing situation is pretty bad now, but it couldn’t possibly get any worse.” And then another year passes and it has somehow gotten worse, and I tell myself the same thing. Relief could come from the over 50,000 units somewhere in the housing pipeline, but a lot of that is going to be built over the next couple of decades, which isn’t very helpful for someone struggling to find an apartment now.
What I wanted to know is how much of that housing will be built over the next few years; a timeframe that’s a lot more relevant for someone living in the city. While it’s impossible to predict how much demand will change, will all the construction lead to a meaningful increase in supply? That would at least give us a chance at some breathing room.
The reason we can answer this question is the fact that years before housing gets built, building permits need to be filed and approved. If we look at all the outstanding permits in the city we can get a pretty good estimate of how much housing will be added in the next 5 years. If you want more detail on how I came up with the numbers in this post, check out this spreadsheet where I did all my analysis.
New housing in 2014
In order to predict what will happen in the future, you first have to look at what has already happened. For this analysis, I started with data the city published on all housing built in 2014. The first thing that jumped out was that virtually all housing added is in buildings with 20 units or more. In 2014, 90% of all new units came from the 27 buildings with 20 or more units. Because new construction is so dependent on large projects, I decided to base all the predictions on the data for large projects and then adjust that number assuming that they’ll account for ~90% of new units added. A quick sanity check confirms that large projects consist of 88% of the new units that have filed a building permit.
It took on average 4.4 years from when the permit was filed until it was completed
Once I imported the 2014 data I cross referenced it with building permit data to figure out when the permit was filed and when it was issued for each building with 20 units or more. This lets us calculate averages of how long a building takes to be completed after it’s permit is filed or approved. For large projects that were completed in 2014, it took on average 4.4 years from when the building permit was filed until it was completed. Once the permit was approved by the city it takes on average 3 years until the building is completed. This does not include the time projects spend under review by the planning department, which has to occur before building permits can be filed. Planning permit data is a little harder to come by but the data I could find suggests it takes at least 3 months and can easily stretch a year or more.
Projecting forward
The next step is to apply these times to buildings that have filed for building permits to get estimates for when they’ll come online. San Francisco publishes reports about pending projects, which can also be cross referenced with permit data to get the filing and approval dates. Applying the average time to completion based on either the filing or approval date gives us a reasonable estimate for what year it will be completed in. These totals need to be adjusted to account for buildings with less than 20 units and for projects that haven’t even filed a building permit yet. After accounting for those factors we get this prediction for how many units will be built each year:
Projected Net Number of Units Built in SF
The model predicts that San Francisco is going to have just under 3,000 units built in 2015, and then jump to a higher level for the next 4 years. To help put those numbers in context, here’s some historical data on new housing in SF:
Net Number of Units Built in SF by Year
2015 will actually be a drop from the previous year, but from then on it will rise to a higher level. The big dip after 2009 is also a good way to see how far the 2008 financial crisis set back housing construction in the city.
What does it mean?
When I first looked at these predictions, it was dispiriting to know that so few units would be built. Fewer units will be built this year than last! It seems unlikely that prices will go down significantly unless there’s a major change in demand.
A more optimistic take, assuming new housing has the same number of people per unit as existing housing (2.25), is that enough room will be created in the next 5 years for almost 50,000 new people to live in the city. This would put the population of San Francisco at over 900,000 by 2020. Despite how hard it is to build new housing, the city is growing faster than it has since the 1940s. A couple other points stand out:
- Most of the new housing being built currently comes from large projects in formerly industrial areas like Mission Bay and SoMa.
- Building housing in San Francisco takes a really long time, and that can contribute to higher prices since supply takes a while to adjust to changes in demand. Some of the buildings built this year filed their building permits in 2010, when demand for housing was high, but a lot less than it is today.
Looking into this issue has made me appreciate how much longer the timescale is for housing is compared to the tech industry. Maybe the most effective way to increase supply in the near term is to find a way to speed up the planning and permitting process, especially for larger projects that will add a majority of new units. Given the long lead time and relative predictability of the housing supply, maybe tech companies who are thinking of opening large offices in the city will start to take housing into account in their hiring plans. At the very least, housing isn’t a problem that’s going to be solved overnight and will take the sustained effort of a lot of different people to fix.
There’s still a lot to be said about this issue, so if you have any feedback or other areas you’d like to see investigated please send it my way.
The folks at Numeracy and I will be publishing more about data we find interesting. If you’d like to find out about new posts you can follow @numeracyco on Twitter.
