The reliability of selected income producing real estate appraisal models : an empirical study

Abstract

Historical cost has been and continues to be criticized for lacking relevance for decision making. Because of its unique nature, income producing real estate (IPRE) has been in the forefront of the historical cost criticism. The consensus of opinion is that only fair market value is relevant for IPRE decision making. When the FASB addressed the problem of changing prices of IPRE in SFAS No. 41, fair value was not required as supplemental information. The FASB does not deny that information about the fair market value of IPRE would be relevant. The concern is that information about the fair market value of IPRE cannot be measured with sufficient reliability. It was hypothesized that there would be differences among average absolute and relative reliability measures for four measurement models. The four measurement models were: (I) Cost Approach-Marshall Valuation Service; (II) Cost Approach-Composite Construction Cost Index; (III) Income Approach; and (IV) Multiple Regression Technique. In general, it was found that Model III produced estimated fair values that were different from the other three models. It was also hypothesized that the average absolute and relative reliability measures among the four property types would differ. The four property types were office buildings, warehouses, apartments and shopping centers. This hypothesis was not supported by the data. The last hypothesis was that there were differences among the trend of the average absolute and relative reliability measures across the four appraisal models for all property types. The results of the testing procedures were mixed. Further investigation revealed that again Model III most often differed from the other models. Also, no significant difference was found between Model I and Model II in every case. Contrary to the beliefs of industry participants, Model IV-Multiple Regression Technique was found to be significantly different than Model I and/or Model II in only three out of twenty cases. The results of the study indicate that multiple regression may have some merits.