Housing Bubble, Revisited

After reading Kash’s first housing post, I read the paper he cited, “Are Home Prices the Next Bubble?” (FRBNY Economic Policy Review, Dec 2004). While there is a lot to chew on, I share Kash’s skepticism. In particular, the authors rely heavily on quality adjusted housing prices to reach their conclusion that there is no housing bubble. For example, consider their Chart 6:



The numerator is the total payment, including interest, for a “quality adjusted” house, meaning that they use hedonic regressions to deflate the numerator to adjust for factors such as larger and newer homes, more appliances, lot size, and so forth. The denominator is just median income. As the graph clearly shows, after adjusting for quality, housing is clearly consuming less — not more! — of families’ income. So there’s no bubble. Or, as the authors write at one point, “Perhaps we should be asking why home prices did not rise even more under the circumstances.”

But then I thought about my friends who own really nice, fancy, homes. For the most part, they all moved to the suburbs, often the outer suburbs, in order to afford their fancy homes. In the typical case, a new child created need for more space and a large home was simply not affordable in the close-in parts of the city. Hence the move to the suburbs. If the authors are going to rely upon quality adjustment, they should also at least account for the corresponding increases in commute times.

In fact, the value of an hour of commuting time is usually estimated to be about $15.00. If moving to the suburbs entails an extra 10 minutes of driving to work each way, for 250 days per year, that adds up to 5,000 minutes, or about 83 hours, which is valued at roughly $1200, per year. Depending on the discount rate, that dollar amount has a present discounted value of roughly $10,000 to $15,000, which should be subtracted from the quality-adjusted price. A change of that amount in the quality adjusted home prices used in the above figure would, I suspect, notably alter the above figure. In particular, it would offset much of the decrease.

Without their income numbers, it’s impossible to tell exactly how large the effect would be, but I can ballpark it. Median income in 2003 was about $43,300; using that in the denominator implies a quality-adjusted monthly payment of $505 at the end of their sample, where the ratio is around .14 ($505 is not the actual payment, it’s the quality-adjusted payment; the actual payment is some larger amount). Adding $10,000 to the principal to incorporate the higher commuting time implies an increased monthly payment of about $62, for a commute-adjusted monthly payment of $567. Using that number increases their end of sample ratio to 15.7%. Even higher commute times, or accounting for two commuters, would increase the ratio even further. (Of course, not all home buyers move to the suburbs.)

Also, read the note below the figure: “Authors’ calculations assume an 80 percent loan-to-value ratio.” I believe that putting 20% down on a house is now the exception, not the norm. With less than 20% down, the payments are higher due to both a larger loan size and mortgage insurance. In terms of Chart 6, to the extent that families are in fact putting less money down, the numerator used to construct the series will be larger, and the decline even less pronounced.

Overall, I’m not convinced that the authors are wrong, but I’m even less convinced that they are right. In fact, the stage 3 of a bubble theory, described in a follow up post by Kash yesterday, seems like the most compelling explanation for the evidence in the FRB study:

Third, highly intelligent economists, puzzled by the duration of the overvaluation, evolve theories of why things are different this time, and why this time the overvaluation is perhaps sustainable after all.

AB

P.S. Long time readers will, apparently, appreciate the absence of mentions of TV shows in this post.