A0923
Title: Understanding real estate matches through a semiparametric panel model of the matching function
Authors: Taining Wang - Capital University of Economics and Business (China) [presenting]
Feng Yao - West Virginia University (United States)
Abstract: A semiparametric panel model of the matching function is proposed, extending the conventional (log) Cobb-Douglas function to allow coefficients to vary with environmental variables. The model captures various unobserved searching frictions during the matching process by accounting for latent heterogeneity in both cross-sectional units and unobserved common shocks over time. Furthermore, the model allows for significant flexibility, enabling varying coefficients to appear in both constant and time-varying regressors (supply) and accommodating different environmental variables (mortgage rates and unemployment rates) in distinct coefficient functions. A two-step estimator is proposed without requiring a normalization of the fixed effects. The first step estimates the varying coefficients with series-based estimators, eliminating fixed effects through multiple differencing. The second step performs a one-step kernel backfitting to improve the estimation efficiency. It is demonstrated through Monte Carlo simulations that the estimators are computationally efficient and perform well relative to a profile-based kernel estimator. Matches in the U.S. real estate market over 100 metro cities during 2012-2018 are studied. It is found that nonlinear effects of the mortgage and unemployment rates decrease the matching elasticity of housing sellers and buyers, and the magnitude of the effect varies with matching efficiency.
