A PCA Factor Repeat Sales Index for Apartment Prices in Paris

Abstract

This paper addresses the issue of building a repeat sales index based on explanatory factors. An earlier paper (Baroni, Barthélémy, and Mokrane, 2004), built a factorial index based on a selected linear function of economic and financial variables. Here, a more general and robust model based on Principal Components Analysis (PCA) is developed and applied to the Paris residential market over the 1973-2001 period. The PCA index for Paris is estimated and its characteristics and robustness are analyzed depending on estimation period, choice of observations, periodicity and reversibility. It is then compared to a standard repeat sales index (WRS), which was estimated using the same data, and the resulting indices are quite similar except in very particular market circumstances. Finally, contrary to the WRS index, the findings indicate that the PCA index can be efficiently used to forecast apartment prices.

Keywords: Real estate indices, Repeat sales, Factors, PCA, Index forecasting

JEL Classification: C20, G00