By

Otis W. Gilley

Department of Economics and Finance

College of Administration and Business

Louisiana Tech University

Ruston, Louisiana 71272

(318)-257-3468

and

R. Kelley Pace

LREC Endowed Chair of Real Estate

E.J. Ourso College of Business Administration

Louisiana State University

Baton Rouge, LA 70803

(225)-388-6238

FAX: (225)-388-6366

Kelley@Pace.am

www.spatial-statistics.com

We would like to thank John Boyce for his valuable comments. Both authors gratefully acknowledge the research support they have received from their respective institutions.

This manuscript is a version of the article which appeared in:

Gilley, O.W., and R. Kelley Pace, "On the Harrison and Rubinfeld Data," Journal of Environmental Economics and Management, 31 (1996), 403-405.

Academic Press holds the copyright and has granted us permission to distribute this copy.

In a well-known paper, Harrison and Rubinfeld (1978) investigated various methodological issues related to the use of housing data to estimate the demand for clean air. They illustrated their procedures using data from the Boston SMSA with 506 observations (one observation per census tract) on 14 non-constant independent variables. These variables include levels of nitrogen oxides (NOX), particulate concentrations (PART), average number of rooms (RM), proportion of structures built before 1940 (AGE), black population proportion (B), lower status population proportion (LSTAT), crime rate (CRIM), proportion of area zoned with large lots (ZN), proportion of nonretail business area (INDUS), property tax rate (TAX), pupil-teacher ratio (PTRATIO), location contiguous to the Charles River (CHAS), weighted distances to the employment centers (DIS), and an index of accessibility (RAD).

Belsley, Kuh, and Welch (1980) used the data to examine the effects of robust estimation and published the observations in an appendix. It also is one of the few moderate sized hedonic data sets available on the Internet (via STATLIB). Many authors have used the data to illustrate various points. For example, Krasker, Kuh, and Welch (1983), Subramanian and Carson (1988), Brieman and Friedman (1985), Lange and Ryan (1989), Breiman et al. (1993), and Pace (1993) have used the data to examine robust estimation, normality of residuals, nonparametric, and semiparametric estimation. Essentially, a cottage industry has sprung up around using these data to examine alternative statistical techniques.

Unfortunately, these data have some incorrectly coded observations and an unsuspected censoring problem. In the process of conducting another study, we rechecked the data against the original census data. We discovered eight miscoded dependent variable observations which appear in Table 1. Moreover, we discovered the Census Bureau censored tracts whose median value was over $50,000. Hence, all tracts with a median value equal to or greater than $50,000 appeared as $50,000. Table 2 identifies the 16 censored observations.

To examine the sensitivity of the Harrison and Rubinfeld results to these changed data, we ran (1) the original uncorrected OLS regression; (2) the OLS regression on the corrected dependent variable observations; and (3) a TOBIT using the corrected dependent variable observations. The results of these three regressions appear in Table 3. The goodness-of-fit as measured by R² rises somewhat when employing the corrected observations. However, the magnitudes of the coefficients de not change much and the qualitative results from the original regression still hold.

We attempted to examine the independent variables for similar problems, but we could not replicate any of these.