Comment

In this paper, Guren, McKay, Nakamura, and Steinsson (Guren et al.) advocate using regional cross-sectional empirical estimates to measure microeconomic parameters such as the marginal propensity to consume out of housing wealth. The “divide-by-multiplier”method recovers the microeconomic parameter by dividing the regional response to a shock by a “demandmultiplier” estimated separately as the regional response to a government spending shock. This method has many potential applications in regional data. (Full disclosure: in Chodorow-Reich, Nenov, and Simsek [2019] I use it to measure the stock market wealth effect on consumption.) Guren et al. provide a comprehensive yet accessible theoretical exposition and a numerical evaluation in the context of a model of housing wealth. I divide my comments into four parts. First, I provide background on literatures using regional data in macroeconomics and on demand multipliers. Second, Guren et al. apply the method to a shock to preferences for housing. This choice introduces complications because housing is both part of wealth and an expenditure item, and both construction and consumer spending may respond to changes in house prices. I illustrate the relationship using a simpler yet still fully microfounded model with pure wealth shocks. Third, I use thismodel tomakemymain critique of the paper or, more accurately, of the paper’s title. A casual reader might infer that the divide-by-multipliermethod provides the only validway to interpret regional empirical estimates. My simple model illustrates an additional property of regional effects of demand shocks, namely, that they provide a lower bound for the appropriately defined aggregate effect of the shock. I conclude by reviewing other recent approaches to interpreting