13th A R Bergstrom Prize in Econometrics to James Graham

Congratulations to James Graham, who was awarded the 2019 A. R. Bergstrom Prize in Econometrics for his paper “House Prices and Consumption: A New Instrumental Variables Approach”. The Bergstrom Prize can be awarded every two years and aims to reward the achievement of excellence in econometrics, as evidenced by a research paper in any area of econometrics.

Fluctuations in house prices are thought to have a significant impact on the macroeconomy, especially through their effect on consumption expenditures. The household balance sheet channel is a commonly cited explanation for this effect, whereby movements in house prices induce wealth effects and changes in the value of collateral. The primary difficulty in trying to identify these effects empirically is that house prices are endogenous equilibrium objects. Instrumental variables strategies are one way to isolate potentially exogenous variation in such prices.

James’ paper assesses the response of household consumption to house prices using a new instrumental variable strategy, which follows the literature on Bartik instruments. These are often referred to as “shift-share”’ instruments, since they typically consist of some aggregate shock that differentially affects locations according to the share of economic activity exposed to that shock.

In the paper, the instrument for local house price growth consists of the local share of houses possessing particular physical characteristics, which is then interacted with regional growth in the marginal prices of those characteristics. The characteristics include features that capture the overall quality of a house, such as its age, number of bedrooms, or number of bathrooms. The intuition for the effectiveness of the instrument is that the more the housing stock in each location is concentrated in particular characteristics, the more exposed that location will be to shocks to the relative prices of those characteristics. For example, if San Francisco was composed mostly of two-bedroom houses built prior to the 1960s, while Las Vegas has mostly four-bedroom houses built in the early 2000s, then a general increase in the price of larger and newer houses would result in relatively faster house price appreciation in Las Vegas.

To conduct the empirical analysis, James uses large micro-data sets on house prices and household consumption. The house price instrument is constructed using data on millions of individual housing transactions from across the US. Because this data contains information on geography, house characteristics, and house prices, it can be used to construct the composition of house characteristics in each location and to run the hedonic house price regressions that are used to estimate the marginal prices of the house characteristics. Household consumption expenditures are taken from a large panel data set which, among other things, contains information on the location of each household. Each household can then be linked to local house price changes in order to estimate the effect of these prices on their consumption behavior.

James reports IV-estimated average consumption elasticities with respect to house prices in the range of 0.1 to 0.15. These estimates correspond to marginal propensities to consume out of housing wealth of approximately 1.2 to 1.8 cents in the dollar, in line with previous estimates in the literature. Another important contribution of the paper is to show that, when estimating consumption elasticities in a panel data setting, the Bartik instrument performs much better than popular alternative instruments in the literature (e.g. local housing supply elasticities). For instance, the instrument provides significant time-series variation in house prices, while other instruments are mostly only useful in the cross-section.

In their assessment, the adjudicators Professors David Fielding, Alfred Haug, and Dorian Owen noted that James’ paper “is an excellent piece of empirical research.  It is well crafted with great attention to detail, using new data on millions of individual housing transactions across the US. The author developed a novel instrument for house prices in order to overcome endogeneity problems when estimating the relationship between household consumption and house prices. The robustness of the empirical findings is convincingly demonstrated.”