An analysis of appraised values and actual transaction prices in the US CMBS market

This thesis addresses the characteristics of transaction-based indices and appraisal-based indices and compares the difference between appraisal and transaction price in the United States Commercial Mortgage-Backed Securities (CMBS) market. The examination is based on the transaction database of Real Capital Analytics, Inc (RCA). A hedonic regression model is applied to data for the period 2000-2006 to produce national indexes at the all-property, office and retail levels. The hedonic model examines the relationship between appraised value or transaction price and NOI, property characteristics, and time. The results are used to create price and appraisal indices. Moreover, the results also prove that multivariate regression analysis is a cost-effective statistical procedure for estimating property values in a time-varying approach. Despite the characteristics influence on price, the relationship between transaction and appraisal behavior is demonstrated in this article. The transaction-based index reflects the timing and changes of market price more accurately and effectively than appraisal-based index does during the examination period. Comparing two appraisal indices, the one without transactions (refinancing) is less volatile than the one with transactions (sale). The underlying reason is appraisers have more pressure when there is a transaction occurred comparing with only for refinancing deal. Therefore, they will appraise those properties with transactions higher than refinancing ones. In addition, after comparing appraisal index without transaction (refinancing) and transaction index (sale), I learn that transaction index for sure leads appraisal index at least one period due to its lagging issue. Therefore, we can predict appraisal index return based on transaction index.