In this paper, we identify a critical issue in the weighted repeat sales (WRS) method—the omission of market risk in the weight estimation model specified by Case and Shiller (1989). We demonstrate that the omission of market risk is conceptually unjustified. We propose a modified WRS model that is empirically supported, but also contributes to the broad discussion on the holding period dependence of real estate risk. We also show that the Case-Shiller weighting method is likely to be mis-specified in nine of the ten cities where the Case- Shiller metro indices are ‘‘tradable’’ with housing options and futures contracts listed on the Chicago Mercantile Exchange. Using a large sample of repeat sales from the Washington DC area, the original repeat sales method of Bailey, Muth, and Nourse (1963) and the Case-Shiller method are compared against the modified WRS. The results indicate that market risk plays an important role in the index estimation.
