Recent empirical work in public finance uses the housing price response to public investment to assess the efficiency of local durable public good provision. This article explores the theoretical justification for this technique. It points out that the logic justifying the technique for evaluating nondurable public good provision does not translate to the durable case. A model in which investment is determined by the interaction between a budget-maximizing bureaucrat and a community's residents is used to demonstrate that the technique can falsely predict underprovision, falsely predict overprovision, or perform without error. * Manuscript 4 COATE AND MA idea, Brueckner develops a model in which if housing prices rise following a small, permanent increase in local nondurable public good provision, then it can be inferred that the good is underprovided. Conversely, if housing prices fall, the good is overprovided (Brueckner, 1979(Brueckner, , 1982. This model has been used as the basis for a number of empirical studies of the optimality of local public good provision (see, e.g., Brueckner, 1982;Barrow and Rouse, 2004;Lang and Jian, 2004).Can housing prices be used to assess the social optimality of local durable public good provision? In an ambitious and creative paper, Cellini et al. (2010) employ the approach to detect whether local school districts are over-or underproviding public school facilities. Using a static version of Brueckner's model, they argue that if housing prices in a district rise following an investment in public school facilities, then such facilities are underprovided. Conversely, if housing prices fall, facilities are overprovided. To estimate what house prices would be in the counterfactual situation in which an observed investment is not undertaken, Cellini et al. exploit the fact that investments must be approved by residents in a referendum. Drawing on the regression discontinuity literature, they then compare housing prices in school districts in which referenda have just passed with those in which they have just failed. If prices are higher in the just passing districts, they argue that school facilities are underprovided. This is indeed what they find for California school districts.The intuitive appeal of this approach notwithstanding, there are important conceptual differences between an investment in a durable public good and a permanent increase in a nondurable public good. First, because of depreciation, the benefits from investment in the durable public good will not be permanent. Rather, they will diminish over time. Second, again because of depreciation, whether or not the investment in question is undertaken, future investments will be made by the community. Moreover, the nature of these investments will depend on the stock of the public good and hence on the fate of the investment in question. This creates a linkage between the current investment and the future investment path in the community. These differences raise the question of whether the logic that underlies the no...