A few months ago, I blogged about the impact of Airbnb on rents for traditional month-to-month or year-to-year tenancies. I suggested that this impact was pretty minimal, reasoning as follows: even in a large city such as Los Angeles, Airbnb units are less than 1 percent of all rental units. So even if every single Airbnb unit would (in the absence of Airbnb) otherwise be part of the traditional rental market, Airbnb is unlikely to increase rents in that market. 

The comments (and a recent San Francisco Chronicle story) raised an interesting response to my theory: what matters isn't the percentage of all rental units, but the percentage of all rental vacancies or all new housing units. In the words of the Chronicle story: "where a typical year sees just 2,000 new units added, a few hundred units off the market makes a significant dent."

But as I thought about the argument, I was less and less persuaded by it. Here's why: first, the number of vacancies is limited to new housing units. San Francisco has just over 236,000 rental housing units. The units other than the new units are not owned by their current owners or occupants forever: rather, they shift around from occupied to unoccupied as tenants move, and as owner-occupants become landlords or vice versa. So the number of units vacant at any given point in time is a bit higher than the 2000 figure, and the number of units that become vacant at some point over the next year or two will be higher still. 

Second, it seems to me that a few hundred units will have little effect upon overall vacancy rates, which in turn means that they will have little effect upon rents. A recent report by the Furman Center [pdf] (affiliated with New York University) lists rental vacancy rates of eleven cities (p.8). San Francisco has the lowest vacancy rate (2.5 percent) and the highest rent ($1491). Boston, New York and Los Angeles are in a virtual three-way tie for second lowest vacancy rate (between 3.4 and 3.5 percent). These three cities are numbers 3-5 in rents (p. 10). (Washington is no. 5 in vacancies but no. 2 in rents, perhaps because Washington is a more affluent city).* At the other end of the spectrum, the two cheapest cities, Houston and Dallas, were no. 10 and no. 9 in rental vacancies. In sum, there seems to be a pretty strong correlation between vacancy rates and rental rates. Since the law of supply and demand suggests that a small supply normally leads to high prices for any commodity, I suspect that this correlation indicates a causal relationship.