A Monte Carlo Exploration of the Vertical Property Tax Inequity Models : Searching for a 'Best' Model

This paper examines the empirical complexities of using linear, log-linear, and nonlinear models to test for vertical property tax inequity in residential real estate data. We apply the existing vertical inequity models to real estate data from Lubbock, Texas. Subsequently, we use Monte Carlo simulations to explore the performance of each inequity model on a variety of different predetermined inequity patterns. Using the Lubbock data as a starting point, we generate eight different contrived inequity patterns from three different data generating processes, which create 24 stylized data sets. The first data generating process treats assessed values as a noisy function of sale prices, which reflect home values. The second process considers the sale price as a noisy function of assessed values. The final process considers both sale prices and assessed values to be a noisy measure of home values. Although our findings do not yield a “best” model, the results do allow us to draw some conclusions and provide practitioners and academics with a road map to test for vertical tax inequity in future real estate data