Many recent papers have claimed that when housing services are treated separately from other forms of consumption in utility, a wide range of economic puzzles such as the equity premium puzzle can be explained. Our paper challenges these claims. The key assumption embedded in this literature is that households are not very willing to substitute housing services for consumption. We show that housing services and consumption must be much more substitutable than has been assumed for a neoclassical consumption model to be consistent with U.S. house price data. Further, when forced to match both historical house prices and stock returns, the lowest risk-free rate the model can generate is 11 percent.