UBC Theses and Dissertations
Measures of the quality of real property assessments : an examination of their validity Ghert, Bernard Irvin
The tax on real property continues to be a major factor in the fiscal structure of municipal governments. Therefore, justice dictates that the impact of this tax be distributed equitably, however "equitable" may be defined by the legislative authority. This is an administrative problem. In many jurisdictions the administration is directed to distribute the tax impact on an ad valorem basis of current market value. Statistical measures have been devised in order to measure the uniformity of the assessment of current market values. The random variable is defined as the ratio of assessed value to actual sale price, and the sample is the result of property sales which occur under conditions which are circumscribed by the definition of market value. The estimated parameters are then used to make direct statistical inferences regarding the level and uniformity of assessment. These parameters are then compared to pre-selected standard parameters in order to judge the relative uniformity of the assessment roll. First, market generated sales do not produce a random sample of the assessment roll. Therefore, direct statistical inferences from the sample do not apply to the assessment roll. Second, it is demonstrated that the standard parameters are implicitly based on an assumption that the universe of assessment-sale price ratios is normally distributed. The results of empirical investigation upon the Vancouver assessment roll indicate that the normality assumption cannot be justified. Therefore, the pre-selected standards are useless as measures by which to judge the relative uniformity of an assessment roll. Two alternative measures of assessment quality are proposed. First, a binomial model based on correct and incorrect valuation is examined. Though this has some practicable possibilities, it cannot, without some important value judgements, be used to measure degrees of quality or uniformity. A second alternative is proposed in order to overcome the above limitation. The alternative suggests that a good assessment roll have assessment-market value ratios which are normally distributed, and that the actual distribution of assessment-sale price ratios be tested for "goodness-of-fit" to the ideal distribution. The test parameters can be transformed to probability levels which would measure the quality or uniformity of the assessment roll. This measure may indicate that a highly uniform assessment roll on which most of the properties are under-assessed is of lower quality than an assessment roll on which more properties are correctly assessed, but less uniformly assessed. Though it is unlikely that uniformity of assessment can be achieved without correct valuation, it cannot be held that an assessment roll on which more properties are correctly assessed is of higher quality of assessments than a roll on which most properties are under-assessed, if the uniformity of assessment on the former roll is not at least as high as the uniformity of assessment on the latter roll. Since the measures of assessment quality and uniformity examined are not satisfactory, it is concluded that statistical analysis in assessment administration may be more usefully applied to the analysis of the causes of assessment errors than to the measurement of the number and size of the errors.
Item Citations and Data