CAPTURING GEOGRAPHICALLY LOCALIZED MISSPECIFICATION ERROR IN RETAIL STORE CHOICE MODELS

No retail store choice model, no matter how many relevant variables it might include, can realistically expect to model all the variation in store choice. There are always some variables that are left out, beca use they are difficult to measure, they have not yet been conceptualiz ed in theory, or their estimated parameter stability suffers when an e xcessive number of predictors are included. Because these omitted vari ables can be correlated with geographic location, model misspecificati on error may itself be correlated with location. Estimating the geogra phically localized misspecification errors therefore suggests itself a s a method for estimating (and predicting) the effects of these omitte d variables. The authors show that spatial nonstationarity of the mode l parameters may also be expressed as an instance of omitted variables and therefore be addressed using their method. They show, using both a simulation study and an empirical natural experiment, that estimatin g the geographically localized misspecification error can appreciably reduce prediction error, even when the predictor model is reasonably w ell specified.