Thanks to Stephen Smith, I recently ran across an interesting database: HUD data on building permits by municipality. So I decided to find the number of permits per 1,000 for a wide variety of cities, focusing on (1) multifamily permits (because rising rent is a bigger problem in most places than rising home costs) and (2) during 2015 and 2016 (because isn’t two years of data always better than one?). Here’s what I found for the places I bothered to look up:
Growing high cost cities
permits per 1000/mean price for units with 5 or more structures in thousands
San Francisco 8.8
San Diego 7.6
Los Angeles 6.7
Growing low cost cities
Austin 13.2 (not sure whether this counts as a low-cost city- really its kind of borderline)
Low demand (i.e. declining) cities
St. Louis 3.6
Another fun fact: suburbs of expensive cities lagged behind even the low-demand cities. Nassau and Suffolk Counties had 0.3 and 0.2 multifamily permits per 1000 respectively, Marin County outside San Francisco 0.7, Orange County outside Los Angeles 4.7.
Some takeaways: 1. Low-demand cities generally had less building than even the most restrictive cities.
2. Within the high-cost city group, it seems to me that there is a strong correlation between permits and prices: Seattle and Denver are certainly cheaper than San Francisco or Los Angeles.
3. On the other hand, there were some low-cost cities that didn’t have a lot of new construction, like Houston and Indianapolis. But this weirdness can be explained away by looking at new single-family construction: Houston had more than half as many new single-family permits as multifamily (9273 to 15475) as did Indianapolis (1515 single, 2017 multi) while Atlanta had eight times as many multi-family (1615 to 13,113) and Dallas had more than six times as many multifamily (3080 to 20,429). Where land is plentiful enough, some people buy instead of rent, creating demand for single-family housing.
4. All of this is subject to one big caveat: this data gives us permits, not actual construction. In 2015, New York City had four times as many permits as certificates of occupancy. I do not know whether this is typical.