An accurate gauge of property value is an important element of residential mortgage credit analysis. For this reason, Standard & Poor's Ratings Services encourages issuers of residential mortgage-backed securities (RMBS) to make an Automated Value Method (AVM) value on all securitized loans available both to us and to RMBS investors. The AVM value would complement the value generated by a traditional appraisal. We believe the inclusion of AVM values, or other valuations, provides a cross-check for determining an accurate property value and enhancing our analytics. This supplemental data will be considered as part of our operational risk analysis, and would facilitate our ongoing efforts to enhance our analytics and processes and help us to provide additional insightful research to the mortgage and securitization markets.
Standard & Poor's LEVELS® 6.1 Model File Format will provide additional fields to capture the AVM system, the AVM system value, and the date of the AVM.
We would like to receive feedback from the marketplace as to the usefulness of supplemental valuation methods other than AVMs. For example, how worthwhile are the use of "drive-by" appraisals, broker price opinions, or tax assessments as value-add methodologies in determining the most accurate valuation of a property. Comments should be directed to Leslie Albergo at 212-438-2381 (firstname.lastname@example.org).
For mortgage lenders, accurate information about both the borrower and the property being used as collateral for a mortgage loan is a fundamental requirement. The appraisal report, which is the basis for determining just what the real estate asset is worth, contains vital information on the property. So when appraisals are inaccurate or fraudulent, the consequences can be severe—not just for the lending institution, but also for investors in securitized products that include loans originated based on this faulty data.
Reports of abuses of appraisal practices and outright fraud have been mounting, despite many attempts—regulatory and procedural—aimed to prevent such practices. Standard & Poor's believes that the appropriate use of AVMs can help diminish the risk associated with the most commonly reported fraudulent appraisal practices by offering a second opinion on the collateral value, outside of the origination process. Indeed, the use of AVMs may help counteract fraudulent appraisal schemes and increase confidence in both reported property values and the loan-to-value ratios.
Standard & Poor's has been reviewing AVM models for several years, and we are beginning our next round of AVM testing. The results of these vendor-specific tests have long been a part of our loan-level credit enhancement models, using sales price thresholds and incorporating geographic location.
Mortgage fraud has grown into a problem serious enough to merit attention from the Federal Bureau of Investigation (FBI), which divides mortgage fraud into two general categories: "fraud for housing" and "fraud for profit." (For more information, see the related report, "Federal Bureau of Investigation, Mortgage Fraud: New Partnership to Combat Problem," dated March 9, 2007, and available at www.fbi.gov/page2/march07/mortgage 030907.htm.)
Fraud for housing occurs when borrowers misrepresent their income or employment when applying for a loan to buy a home. Given the underperformance of certain types of residential mortgages, there's reason to believe that the incidence of borrower fraud for housing has increased.
More serious appraisal fraud generally falls under fraud for profit, a phenomenon that involves collusion among industry insiders. The FBI categorizes roughly 80% of all fraud schemes as fraud for profit and focuses on this type of fraud in its general investigations.
According to the FBI report, typical schemes include:
Appraiser identity theft is a growing concern. A person skilled at computer technology can easily replicate an appraisal's digital signature and pass off a valuation—complete with the appraiser's license number below the signature—as legitimate. This ability, of course, greatly increases the chance that the fraudster will be successful and remain unidentified.
Third parties to the transaction can also contribute to appraisal fraud, particularly in situations where the mortgage broker needs the value of the home to be a certain amount for the borrower to qualify for financing. Appraisers continue to report pressure from mortgage brokers, originators, real estate professionals, and others—including appraisal management companies who hire them—to inflate the property values.
In fact, a new national survey found that 90% of appraisers reported that mortgage brokers, realty agents, lenders, and even consumers have pressured them to raise property valuations to enable deals to go through. This percentage is up sharply from figures in a parallel survey conducted in 2003, when 55% of appraisers reported attempts to influence their findings and 45% reported that such pressure "never" happened. Now the latter category is down to just 10%.
Both surveys were conducted by October Research Corp., a firm in Richfield, Ohio, that publishes Valuation Review, a popular industry newsletter. The latest survey involved 1,200 appraisers representing a statistical cross-section of the industry in 50 U.S. states, the District of Columbia, and Puerto Rico. The results have a margin of error of plus or minus 2.8 percentage points.
The mortgage industry has been taking various measures to prevent fraud. For example, the August 2007 Mortgage Bankers Association Appraisal Industry Outreach Task Force Report called for a national appraisal registry or national property database to address the competitive pressures on appraisers to inflate home values. Under this proposal, appraisers would submit reports to a centralized database that would be available to mortgage lenders, rather than funneling their reports back to brokers. The appraiser's valuation would thus be free from any other party's intervention.
In another move to head off fraudulent appraisals, the Mortgage Asset Research Institute (MARI®) recently announced enhancements to its Mortgage Industry Data Exchange (MIDEX®) anti-fraud database. MARI has now added "license verification" of appraisers and real estate agents to the platform. MIDEX-Licensing provides instant online access to licensing information on industry professionals (companies and individuals) from more than 60 federal and state mortgage regulators.
Other suggested practices include appraiser lists and report cards to monitor appraisers' reliability and ethics and more easily identify questionable parties.
While we support the use of AVMs, regular testing and validation of these models is vital to their acceptability.
Standard & Poor's will soon begin its next round of AVM testing. In order to enhance our testing capabilities, we've added the expertise of AVMetrics LLC to our AVM validation process. We believe this partnership will allow us to leverage the core capabilities of our respective firms. We will use AVMetrics LLC to help coordinate our testing process. We will continue to apply our specific loan-level criteria to the aggregated data we receive from each vendor participating in our validation process.
Standard & Poor's expects to test the accuracy of AVM models semiannually. The testing process begins with a large, geographically diverse data file of property addresses that have sale dates within the prior three to six months. The results are stratified and measured, allowing us to compare the variance of the actual sale price with the estimated value given by the AVM system according to the sales price bucket, state, county, and zip code.
The variance analysis allows us to develop system-specific assumptions at the loan level, which we incorporate into our ratings models. Our ratings assumptions with the use of AVMs include an analysis of the credit quality of the borrower, the location of the property, and the value of the home.
The quality of data supporting an AVM model determines the quality of its performance. The data at the core of any model's results should be as accurate, current, and complete as reasonably possible, a challenge that AVM vendors continually address. Accordingly, Standard & Poor's review process has become helpful to the overall credit risk analysis of residential mortgage portfolios.
AVM vendors interested in participating in this testing should contact Standard & Poor's immediately. For additional information, call Dan Hall at 212-438-1576 or Leslie Albergo.
For more information, please see the related article, "S&P Enhances Its Review of Data Quality Practices for U.S. Residential Mortgage Originations," published Oct. 10, 2007, on RatingsDirect, the real-time Web-based source for Standard & Poor's credit ratings, research, and risk analysis, at www.ratingsdirect.com.