Never reason from a population change

I realize that people get tired of me continually harping on the “never reason from a . . . ” maxim, but it’s a continual problem.

A recent NBER study of housing by Schuyler Louie, John A. Mondragon, and Johannes Wieland had the following abstract:

The standard view of housing markets holds that the flexibility of local housing supply–shaped by factors like geography and regulation–strongly affects the response of house prices, house quantities and population to rising housing demand. However, from 2000 to 2020, we find that higher income growth predicts the same growth in house prices, housing quantity, and population regardless of a city’s estimated housing supply elasticity. We find the same pattern when we expand the sample to 1980 to 2020, use different elasticity measures, and when we instrument for local housing demand. Using a general demand-and-supply framework, we show that our findings imply that constrained housing supply is relatively unimportant in explaining differences in rising house prices among U.S. cities. These results challenge the prevailing view of local housing and labor markets and suggest that easing housing supply constraints may not yield the anticipated improvements in housing affordability.

When I first read that abstract, I was skeptical. How is it possible that building constraints are not the primary cause of high prices in places like California and the Northeast? It turns out that my skepticism was justified. A new paper by Michael Weibe has the following abstract:

Louie et al. (2025) (henceforth LMW) claim to provide evidence that housing supply constraints are quantitatively unimportant in understanding changes in housing prices and quantities in the United States. In this short comment, I show that LMW’s empirical model is unidentified. LMW models housing demand as a function of population and average income. However, supply constraints are a key determinant of population growth: newcomers are less able to move into a city that restricts the construction of new housing. Hence, LMW does not have exogenous variation in demand. Accordingly, the empirical results are uninformative about the role the housing supply constraints.

If you are having trouble understanding the problem in the first paper, consider the following analogy. Suppose someone said, “High gas taxes are not the real reason why Europeans consume less gasoline (per capita) than America, the actual reason is that they have smaller cars and use public transport.” The problem with that claim is pretty obvious, right? The tendency of Europeans to use smaller cars and public transport partly reflects the fact that gas taxes are extremely high in Europe. Economists call this the “identification problem.” It is important to identify whether market changes are caused by demand shifts or supply shifts. Neither price nor quantity can answer this question.

Not very many people have moved to California in recent years, but that’s not because there’s little demand to live here. Rather it’s because supply constraints have pushed housing prices in California to levels far higher than in most other states.