Matt Yglesias is on the right track regarding affordable housing, but he still makes a couple important mistakes:
At the margin, adding new market-rate housing units reduces the cost of housing, making housing more affordable. In other words, given a choice between a project with 20 affordable units and 20 market-rate ones or a project with 40 market-rate units the former is more desirable from an affordability standpoint. But the 40 market-rate units is better for affordability than is letting the site remain as a vacant lot or a very low-density use. The best kind of incentives, meanwhile, do both. For example, you let developers build a taller building than the local zoning code would normally allow, but only if some of the additional square footage is devoted to affordable housing.First, it's not always true that 40 market-rate units is better for affordability than letting the site remain vacant/low-density. Whether 40 market-rate units is better for affordability depends on the area's "filtering" rate—that is, the rate at which housing units move through the quality hierarchy, usually through depreciation, but also through investments to upgrade the units. In a fast-growing metropolitan area where further construction of new housing can be expected to continue, urban economists (Steven Malpezzi in particular, if memory serves me right) have generally found that the supply of affordable housing is indeed very sensitive to the supply of more expensive housing. In that kind of metropolitan area, Matt is right that building 40 market-rate units is better for affordability than letting the site remain vacant. However, in metropolitan areas that are more hostile to new housing (e.g., San Francisco, LA, Boston), the filtering rate is glacially slow. In those metropolitan areas, the opportunity cost of building 40 market-rate units on scarce developable land could well outweigh the small benefit to affordability. The opportunity cost of building 40 market-rate units is much higher if the units are in or near the urban core, because it uses scarce land on which new affordable units can be built (low-income families generally need to live closer to the urban core because of transportation costs). The choice is never "40 market-rate units or nothing." Housing is a durable good, so the time horizon for alternatives needs to be much longer. It's extremely short-sighted to say "40 market-rate units is better than nothing, so build away!" Matt's second mistake is more troubling. He argues that giving developers an incentive to build both market-rate and affordable units are "the best kind of incentives," using a version of a "density bonus" as an example. Density bonuses are better than inclusionary zoning—which force developers to keep a certain percentage of the units they're building "affordable"—but both are part of the same family of policies, all of which are deeply misguided. At best, these policies are woefully inadequate; sadly, most of the time they're downright counterproductive. These policies are often touted as "market-friendly," but that's true only on an extremely superficial level. Density restrictions (commonly referred to as "large lot zoning") are a very large part of the reason it's unprofitable to build affordable housing units in the first place. The solution isn't to trade exemptions from large lot zoning regulations for a handful of affordable housing units; the solution is to change the large lot zoning regulations. Inclusionary zoning and density bonuses are an ineffective sideshow, and to the extent that these kinds of policies distract housing advocates from policies that would actually increase the supply of affordable housing, they're harmful to the cause of affordable housing.