Modeling Airbnb demand to New York City: A panel data approach

The present study is the first to quantify price and income elasticities of Airbnb demand to New York City at the listing level. In doing so, an empirical tourism demand model is estimated employing an unbalanced panel data set on the city’s Airbnb listings. The comprehensive panel data set employed in this research has been obtained from AirDNA and comprises N x T = 478,386 observations for the time period September 2014 to June 2016. After seasonal adjustment and statistical pre-testing (e.g., poolability, Hausman, heteroscedasticity, etc.), an econometric model with the occupancy rate as the dependent variable as well as the relative listing price and source-market-weighted real gross domestic product (GDP) as explanatory variables is estimated. To this end, a one-way fixed effects panel estimator with cluster-robust standard errors is applied. The standard errors are treated as clustered at the zip code area level in order to capture the spatial dimension of the data, i.e., the homogeneity of Airbnb listings within a given neighborhood and their heterogeneity across neighborhoods. Since all variables have been transformed to natural logarithms before estimation, the regression coefficients can conveniently be interpreted as price and income elasticities of demand, respectively. In addition to an overall satisfying model fit, both price and income elasticities are highly statistically significant and feature the algebraic signs expected from microeconomic theory. While Airbnb demand to Manhattan is generally price-inelastic, Airbnb demand outside this borough is price-elastic and comparatively more income-elastic. Differences in price and income elasticities compared to the total of the listings in Manhattan can also be observed for those Airbnb listings that are not entire homes or apartments, for those that are not offered by commercial Airbnb providers, as well as for those featuring combinations of these characteristics. The differences in price elasticities, for instance, have an important implication on the optimal pricing strategy of the single Airbnb providers who are assumed to have a certain degree of market power, enabling them to charge individual prices higher than their marginal production costs, which, in turn, reflects the heterogeneity of the Airbnb offerings.