Optimal Comparable Weighting and Selection: A Comment

Vandell (1991) and Gau, et al. (1992) have recently developed rigorous methods for selecting and weighting comparables for the market comparison approach to valuation. This paper compares the statistical properties of the Vandell and Gau approaches; specifically, it examines the bias and variance produced by both approaches. Under classical OLS assumptions, Vandell's method is preferable; when these assumptions are violated, the size and direction of the violation will determine which method is preferable. Finally, the paper concludes that when the regression producing the adjustment factors has an heteroskedastic error structure, corrections should be made to the regression itself, rather than as part of the comparable weighting and selection process. Copyright American Real Estate and Urban Economics Association.