A Diagnostic Approach to the Multicollinearity Problem for Better Model Selection in the Hedonic Pricing Method
摘要
The hedonic pricing method is an effective technique for estimating land and building prices. Nonetheless, numerous analysts have expressed concerns regarding the potential for multicollinearity to inflate the variance of the estimated coefficients. Hedonic pricing often involves spatially situated objects, and so strong correlations between two or more variables frequently arise. There are several diagnostics for evaluating multicollinearity, although most—sourced from explanatory variables—only measure the severity of this condition. To assess the risk of multicollinearity in a comprehensive manner, it is imperative to develop diagnostics that consider both the probability of occurrence and the severity of multicollinearity. This paper presents a numerical method for evaluating the influence of multicollinearity from these two perspectives.