Identification of Multidimensional Hedonic Models

Abstract

Nonparametric identification results for scalar nonseparable hedonic models were worked out in Heckman et al. (2010). This paper extends this work to multidimensional hedonic models. A fully nonseparable multidimensional hedonic model is point identified. Identification requires policy invariant normalizations, requires data from multiple markets with ``rich" price variation, and requires observable multidimensional aggregate supply or demand shifters. Estimation is based on nonparametric estimation of the joint distribution of the data. When data from multiple markets, is not available, point identification can be obtained with additional restrictions. The paper shows that an additive specification results in a multidimensional transformation model and shows how to extend results for a scalar transformation model to the multidimensional case. These models are identified up to location and scale.