Academic journal article Real Estate Economics

The Accuracy of Owner-Provided House Values: The 1978-1991 American Housing Survey

Academic journal article Real Estate Economics

The Accuracy of Owner-Provided House Values: The 1978-1991 American Housing Survey

Article excerpt

The availability of information on houses in panel data studies like the American Housing Survey (AHS) and the Panel Study of Income Dynamics (PSID) provides valuable sources for estimation of hedonic house price models. However, the main source of information about the price of each house in the AHS and the PSID is the owner's valuation rather than the sales price. Because of the possibility that owners are inaccurate, among other reasons, researchers have focused on different sources, such as sales or assessors' data. However, sales data may not be representative of all houses in the market, and both sources usually lack information on the occupants and neighborhood characteristics,(1) which can limit their usefulness in economic research. The AHS does include the sales prices for some houses, though again the results may be affected by sample selection bias and the sample sizes will be small. On the other hand, if owner-stated values are accurate, the AHS and PSID can be confidently used by researchers.

In this paper, we use the AHS to test the accuracy of owner-reported values for three cities over a fourteen-year period. We compare owners' valuations with the sales prices of houses that sold in the twelve months prior to an interview and find that, on average, new owners state home values that are 8.4% higher than the sales prices. Further analysis indicates that these new owners, as compared to owners with longer tenure, overvalue their houses by 3.3%, on average. Thus, we find that the average owner will overstate the value of his house by 5.1%. We also find that owner errors are not related to characteristics of the owners (except length of tenure), the house or the neighborhood. This suggests that hedonic equations based on owners' valuations will provide unbiased estimates of house price indexes as well as the prices of house and neighborhood characteristics.

The accuracy of owners' valuations has been analyzed by others (see Table 1), and most find evidence that owners tend to overstate the value of their homes. We add to this literature in a number of ways. First, we rely on two models that are based on a theoretical rationale for why owners overvalue their homes (see the Appendix). Underlying these models is the notion that individuals will only buy a house if they perceive the value to be at least as great as the selling price. The difference between these two amounts can be explained by the self-serving bias (Lowenstein, et al. 1993) that arises because the value new owners place on their house reflects their perceived abilities in bargaining in the housing market. In both models, this difference erodes over time as individuals place more weight on market information relative to their own valuation at the time of sale.

Second, we provide a general model of owners' valuations that encompasses both the models we propose as well as those used by Goodman and Ittner (GI, 1992) and DiPasquale and Somerville (DS, 1995). We then test the restrictions on this general model that are consistent with these more specific cases. The statistical evidence does not allow us to accept the GI and DS hypotheses and generally favors our models.

Third, by analyzing the housing markets for three cities over fourteen years we can see if Ihlanfeldt and Martinez-Vazquez's (IM-V, 1986) and GI's results hold in local markets over a much longer time period. Fourth, we enhance our data set by adding census tract information on the neighborhood in which the house is located. This enables us to test for a relationship [TABULAR DATA FOR TABLE 1 OMITTED] between owner errors and neighborhood characteristics, something that previous analyses that used the AHS (IM-V, GI and DS) were unable to do. Fifth, our access to proprietary data also allows us to show that the results on owner accuracy are not affected by the top-coding and bracketing that occurs in the public-use version of the AHS.

The paper is organized as follows. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed


An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.