Arthur Nelson is quoted in yesterday’s USA Today as saying the ownership rate will drop one-half percentage point per year until it reaches 63.5 percent. His foundation for this forecast is a paper he was good enough to send me–I would link to it but I can’t find it on the Web.
It is certainly possible that the ownership rate would fall to its lowest level since 1985, but I am afraid I don’t find Professor Nelson’s argument to be convincing, because his paper contains no model. Rather, he puts together some survey results and short-term trends, expresses his long-held disdain for suburbs, and argues that 57 percent of housing units added to the stock between now and 2020 will be rental units. This would be a sea change, as there has not been a single year since 1973 where less than half of units built were single-family, and in many years, the single-family share exceeds 70 percent.
An appropriate model of tenure choice would statistically model the relationship between tenure choice and a variety of characteristics, including age, marital status, presence of children, race, residency status, education, income and the relative cost of owning versus renting. Here is what we know based on the econometric literature on tenure choice (none of which Nelson cites): old people are more likely to be homeowners; married couples are more likely to be homeowners, permanent residents are more likely to be homeowners, white people are more likely to be homeowners (even after controlling for other characteristics), and homeownership rises in income and education. We also know that when owning is a good deal relative to renting, people are more likely to buy.
So a paper that purports to forecast owning needs: (1) a well specified model; and (2) forecasts of variables that predict owning, including income and relative costs. I don’t know what either income or relative costs are going to look like in 15 years, and neither does Professor Nelson.
BTW, FWIW, I will take the over on 63.5 percent.