Improved IV Estimation of Vertical Property Tax Inequity

In this paper we propose a new IV estimator to be used in detecting vertical property tax inequity. We conduct Monte Carlo experiments to evaluate the bias of this estimator in comparison to traditional linear and log-linear regression based estimators. We find that the new estimator is more robust to bias across alternative average assessment ratios, even in the presences of errors-in-variables, than the IV estimator suggested by Clapp (1990) and frequently used for such purposes. Furthermore, the new instrument allows for an investigation into how strong the measurement error in sales prices relative to that in assessed values must be for the results to change from those of the traditional methods