Estimating the housing market matching function through Internet traffic analysis

In the absence of any empirical estimation, standard theoretical models of the housing market follow the framework developed for the job market that assume a Cobb-Douglas form with constant returns to scale for the matching function. The present study aims to fill the gap and provides a first estimate of a matching function for the housing market. To that end, we assemble a unique dataset, through Internet traffic analysis, with measurements of the numbers of home buyers and home sellers active in the forty-four largest urban areas in France, from October 2013 to June 2017. We confirm that a Cobb-Douglas form fits the data well. However, our results indicate decreasing rather than constant returns to scale. Two extensions of the baseline model random matching are explored. The first accounts for the intensity of the buyers’ search, the second, for some characteristics of the market participants. Both bring evidence that these elements matter in the way buyers and sellers meet and transact and that the matching on the housing market is not purely random.