A random walk down main street?

In this paper we provide a statistical test for the presence of a random wa lk component in house prices. This assumption is widely made when estimatin g repeat sales price index models. The surprising aspect of the test is tha t it is based on repeat sales data in which each house is observed to sell only twice. We do not have a time series of data on each house, nor a panel , and thus conventional tests for unit roots do not apply. Despite the fact that we have only two time series observations per house we show that our test, which is easy to apply with available software, has desirable charact eristics. When applied to the well known Case and Shiller (1989) [Case, Kar l E. and Robert J. Shiller, 1989. The efficiency of the market for single f amily homes. American Economic Review, 79, 125-137.] data the test reveals no evidence of a random wall; component in four large U.S. cities. (C) 1999 Elsevier Science B.V. All rights reserved.