Journal of Property Tax Assessment & Administration


This paper compares the performance of different measures of vertical equity using simulated data. The goal of the paper is to detect the basic tendency of a test to correctly or falsely reject the null hypothesis. To accomplish this, simulated data was constructed so that the true relationships are controlled. In scenarios where there are measurement errors in the use of prices, distortions are of the basic form, i.e., variance distortions, to capture the tendency of the tests toward regressivity or progressivity. While more simulation studies that allow for other types of distortions such as outlier effects, nonlinearities, small samples, etc. could produce additional observations and possibly affect our conclusions, the results that would hold true regardless would be that (1) ASR and PRB have at least some endogeneity bias by construction with the ASR likely having more bias than the PRB, and (2) the methods belong to two classes: those with no measurement error assumption (non-EIV models) and those which assume measurement errors exist (the EIV models). Because these tests are constructed under different assumptions and operate under a different null (model conditions under the null are different), their behaviors are different. We do not recommend one particular test over another but rather that a suite of tests be reported to support the existence or absence of vertical equity. The tests of Paglin-Fogarty and Cheng, as well as Gini-based (distributional) tests, the Spearman Rank test, and the Clapp test are good candidates. It is important to note that Paglin-Fogarty and Cheng methods are subject to regressivity bias when assessed values are estimated from a regression.


Assessment administration; Equity; Assessment performance measures