Multilateral index number theory and generalized repeat sales house price indexes

One of the most common approaches to constructing home price indexes is the repeat sales method. In this article, it is shown that this approach is part of a family of widely used multilateral index number methods known as weighted GEKS. Here, the price index is constructed in two stages. First, elementary indexes are constructed between each period. Second, these bilateral indexes are aggregated into a transitive set of multilateral indexes using weighted least squares regression. We show that the repeat sales index has precisely this form and identify the specific elementary index formula and weights used. This observation provides a basis upon which to consider generalized repeat sales indexes which use different elementary indexes and weights. We apply some of these methods to repeat sales data and show that these choices matter empirically.