from CoreLogic
— this post authored by Yanling Mayer
One of the many challenges when underwriting a loan is to make sure the appraisal provides a reliable assessment of the collateral’s value. Fortunately, to meet the challenge and to stay compliant in a demanding regulatory environment, appraisal reviewers and underwriters have increasingly incorporated automated appraisal review technology for faster and more cost-effective collateral underwriting and process control – while ensuring the appraisal report reflects industry’s best practices, standards and guidelines, as well as requirements from Fannie Mae, Freddie Mac, FHA, and VA, and providing lenders with confidence in the value of the collateral supporting their credit decisions.[1]
From a lender’s perspective, an important challenge in appraisal review is to know whether the provided information is valid. For this reason, one important functionality is to help identify potential issues related to information accuracy of reported comparable transactions in the appraisal. In appraisal terminology, comparable transactions are those current or recent market transactions deemed most similar to the appraisal property in terms of physical and economic characteristics (e.g., location, physical attributes, condition of sale). They serve as the baseline information upon which the appraisal is developed.
Industry best practices include expectations that appraisers use care and due diligence to ensure the information provided by third-party data sources, such as a multiple listing service (MLS) or vendors of MLS data and/or public record data, is accurate. The MLS data are particularly important in providing the latest information on the local property market. When a large sample of 2016 purchase-loan appraisals were analyzed, information on nearly 9-in-10 of the comparable sales in the appraisal report contained a direct reference to the local MLS as the data source.[2]
A reliance on MLS data presupposes that the data are accurate. Figures 1 and 2 examine the accuracy of comparable prices taken from an MLS in the appraisal report, using public records as a means of validation.[3] To avoid potential confounding effects of regional differences in the quality and availability of public records data (for example, caused by state disclosure laws), the analysis shown here is based on a single state, California, and examines only purchase transactions in 2015 and 2016.
For an overwhelming majority of the comparable transactions (91 percent), the appraisal-report price was identical to what was captured at county recorders. A total of 6 percent show a price difference of “less than 0.25%,” which in dollar terms amounts to less than $1,000 in reported price difference on a property valued at $400,000. Thus, it is a rather rare incidence when the two sources of price data either do not match or are not very close. Only about 2.1 percent of the comps show a price difference of 1% or more.
When the MLS and public record data disagreed, which was about 9 percent of cases, it was the MLS that tended to have lower sales prices: Across the 156,000 sample, the MLS price was higher than the public records data in 2.7% of the cases, versus 6.3% of the cases where the MLS data indicated a lower price. Excluding the 91 percent of cases where the two sources matched, the MLS prices averaged about $605 (0.085%) lower than the public records data.[4]
Of course, in those 9 percent of cases when the two data sources do not exactly match, it is difficult to identify which source might have erred. For this reason, automated appraisal review tools will typically send a message to alert appraisal reviewers and underwriters when detecting a material discrepancy between the appraisal report and validating data source.[5]
Footnotes
1 One of the appraisal review technologies is CoreLogic’s Generally Accepted Appraisal RulesTM, known as GAAR®. The GAAR engine consists of thousands of rules that identify potential compliance and valuation risks based on the data and information provided in the appraisal report.
2 The analysis included about two million appraisals on single-family homes located across the U.S. in CoreLogic/FNC’s National Collateral Database (NCD).
3 The public recording office from counties across the U.S., known as county recorder, maintain databases of real estate sales and ownership transfers. The CoreLogic/FNC NCD contains regularly updated public record on residential property sales.
4 Researchers at Florida Gulf Coast University (Marcus T. Allen, Kenneth M. Lusht, and H. Shelton Weeks, “Reported Price Errors: A Caveat for Appraisers,” The Appraisal Journal, Fall 2015) compared MLS sales price data with the price recorded on the corresponding HUD-1 settlement statement. Using a sample of 400 transactions from one MLS during 2004-08, they reported that the MLS and HUD-1 sales prices matched in 91.2% of the sample. In contrast to our analysis, they reported that when the data sources disagreed, the MLS price was more likely to be higher. They reported that the MLS price averaged 0.3% higher across the 400 transactions (and 0.5% higher based on medians); across the 35 transactions where the sales price differed in the two sources, the MLS price averaged about 4% higher (the median price was about 1% higher). Compared with our analysis, the differences could reflect different location, time period and data source.
5 For example, CoreLogic’s GAAR engine automatically generates an alert message to collateral underwriters for further review when it identifies a material discrepancy.
© 2017 CoreLogic, Inc. All rights reserved.
Source
http://www.corelogic.com/blog/authors/yanling-mayer/2017/03/the-accuracy-of-comparable-property-data-in-an-appraisal-report.aspx