Regression Modeling for Vertical and Horizontal Property Tax Inequity

Abstract

Early property tax regression models were used to estimate vertical inequity across assessment districts. By their nature, these models contain measurement error bias. Clapp (1990) used an instrumental variable to overcome this problem. However, there remains the possibility of omitted variable (OV) bias. Addition of variables for neighborhood inequity provides critical information at the local level. It also overcomes possible OV bias in estimates for district vertical inequity. An extended model is developed and illustrated. Conditions are given for significant district OV bias. The methodology is broadly applicable to linear models when there is measurement error in the independent variable.