Fannie and Freddie data collection
Under direction of FHFA, Freddie Mac and Freddie Mac (the GSEs) have developed the Uniform Mortgage Data Program (UMDP) to enhance the accuracy and quality of loan data delivered to each GSE. The Uniform Appraisal Dataset (UAD) is a key component of the UMDP which defines all fields required for an appraisal submission for specific appraisal forms and standardizes definitions and responses for a key subset of fields.
For conventional loans delivered to the GSEs on or after March 19, 2012 GSEs required appraisals to be completed using the field-specific standardization requirements. Appraisal software forms providers incorporated the UAD field-specific standardization requirements into their appraisal report form software. The appraisal data form must conform to the UAD and be delivered through the Uniform Collateral Data Portal (UCDP), the joint portal through which lenders will submit electronic appraisal reports for delivery to Fannie Mae or Freddie Mac. The UAD and UCDP will help lenders, the GSEs, and other industry participants manage collateral risk through efficient collection and enhanced quality of appraisal data.
With the UAD, the GSEs will require that appraisals be completed with standardized responses in certain appraisal form fields. The standardization of certain data points will support consistent appraisal reporting regardless of geographic location of the property or any localized reporting conventions, by addressing vague or disparate data currently included on some appraisal reports. The UAD standardized response requirements. The UAD supports improved quality and accuracy of the appraisal data while preparing each GSE’s ability to determine how the data relates to loan performance and loan quality. None of the UAD requirements inhibit or limit appraisers responsibility to comply with the Uniform Standards of Professional Appraisal Practice (USPAP). As with all appraisal report forms, there is no limitation on appraiser’s ability to present additional information in the appraisal report.
Collateral Underwriter (CU)
For almost 30 years or more, we have enjoyed personal computer power to analyze data, Each year great advances are discovered and used to make life easier. Augmented Reality is one of these advances, which is making its way into the real estate industry. For example; ariel or satellite imagery can produce sketches which calculate the properties gross living area (GLA), or drones could be sent to an address to measure the property using lasers and photography. Data is also aggregated through building permits, contractors, and social media. Pictometry has access to home improvement data where condition inferences can be used to determine the interiors condition. Combining all data, data scientist can map spatial analysis and derive reports on foreclosures, abandoned homes, condition of homes, subdivision characteristics, value influence on playgrounds, community pools, and noise exterior obsolescence. Big Data, GIS, and analytics are all coming together, this is the new science of valuation.
Computer driven Automated Value Methodologies (AVM) have been used since 1981, Core Logic produced over 1 billion avms last year. AVMS modelers are built by economists, PHDs and data modelers. An AVM report can be produced in seconds with large batches of data. Whereas, Real Estate Appraisers typically take over 6 hours to produce a report on a single property, they have to ensure compliance with USPAP, USDP-UAD, CFPB, Dood-Frank, IAG, Federal and State laws. So why not use computer models?
So what is the Collateral Underwriter (CU)?
Jan 26, 2015, Collateral Underwriter was started by Fannie Mae and Freddie Mac. The purpose is to support a more proactive management of appraisal quality, by empowering lenders to address potential issues prior to loan delivery. The VA recently added it's own loan safe appraisal manager, risk score, and integrity score.
Each appraisal will have an overall underwriter risk score of 1-5, underneath that will be appraisal quality flag, an overvaluation flag, and a eligible - compliance flag. The flags may help the lenders use better resources for determining the value. Some of the appraisal quality flags measure data integrity, comparable selection, adjustment used for compatibles, and reconciliation practices.
Data integrity analyzes several factors in each appraisal
- Is the data consistent with ones self?
- Is it consistent with the appraisal peers?
- Is the data plausible?
The Comparable selection measures;
- CU ranks the appraisers provided compatibles against a pool of available compatibles, and does not use arbitrary guidelines.
- Sales from up to 1 year are considered with aggregated time adjustments
- Date of sale received more weight in rapidly increasing and decreasing markets.
- Proximity of recent sales to comparables is dependent on the market.
Adjustments need to be physically similar; GLA, Site, etc, (regression analysis is used to analyze). Time adjustments, negative or positive adjustments are applied based on price appreciation, and location is measured by census block group level where regression determines acceptable adjustments. Sale types (REO, Short, Estate, etc) are also being adjusted based on regression.
Old rules in appraising required the appraiser to limit the adjustments to 15% for line items, and 25% for gross adjustments, the CU does not adhere to these rules any longer. Rules of thumbs are not longer acceptable, they have to be statistically proven, such as price per square foot.
During the reconciliation of an appraisal report, additional flags may indicate higher risk with the value in such cases as; Are appraised values far outside the range of unadjusted comparable values? Appraised values out side the adjusted comp sales prices, appraised values with support from only one single comparable, and values at the lower or upper end of the range of comp values. A flag can be raised for an over valuation appraiser with heightened risk, but no estimate of value is given. No message is given for under valuation.
In Summary, CU is a risk management tool, it doesn't say if the appraisal report is good or bad, or if the appraiser is good or bad. The CU does not accept or reject appraisals. The CU does not provide the lender with a value for the property. Access to the CU database is not provided to the appraisers. Finally, its sole purpose is to access risk of the appraisal. The CU does analyze massive amounts of data using sophisticated models, its actually a sophisticated AVM.
Now we understand why UAD data is being collected and is now being analyzed by the GSEs to limit risk, why don't they use it as AVMs and produce their own valuations? This is where we look at artificial intelligence vs augmented intelligence. Much of real estate is still purchased based on human emotions. An appraiser still needs to be part of the process to determine what comparables to use and the adjustments required. With so much data being collected and analyzed, GIS spatial analysis utilized, and local appraisal knowledge, the analysis is getting more detailed, and accurate.
At Clear-valuation, we saw this trend happening over 8 years ago. We invested in big data, and we are one of only a few appraisal companies that has access to a national residential database of appraiser peer verified, geoscored homes, much like the GSEs. We use a proprietary tool which allows us to quickly choose comparables, collect data, and produce USPAP reports for Dodd-Frank Compliance. Call us to see how we can provide appraisal reports using augmented reality.