By Craig Morley, MNAA, GAA
Featured in “Appraiser Focus Magazine” Q3 2017
UNIFORM APPRAISAL DATASET
The purpose of the Uniform Appraisal Dataset (UAD) was to standardize the appraisal data in residential appraisals being sold to Fannie Mae to adhere to the newly instituted Absolute Ratings. Absolute Ratings is a program within the UAS introduced to rate various property characteristics, mainly the ratings of neighborhoods, views, quality and condition, with quality and condition ratings based on a numeric value based on a numeric value of one to five. This new reporting standard required absolute ratings that, once established by the appraiser for a property, should not be changed when said property was used as a comparable in another appraisal. This concept sounds good in theory, but in practice, there are limitations.
There are a number of restrictions with the rating system established by Fannie Mae. Properties rarely fit nicely into the categories of location, view, quality and condition. The other is that most seasoned appraisers were taught to rate properties relative to each other. The underlying principles of the Sales Comparison Approach is that of substitution, which is based on the concept of comparing similar homes in the market to that of the home being appraised and accounting for relative difference. Applying absolute ratings and description to a process of making relative comparisons has some practical problems. Appraiser began to find that most properties were falling between the lines in quality and condition ratings. At what point does a house move from “average” to “good”, and how do I maintain a consistent rating system with the other appraisers in my market area?
Marshall & Swifts definitions for quality rating is reported to have been used by Fannie Mae. However, looking at Marshall & Swift’s Cost Data shows that there are very large difference in cost from similar-sized properties from one quality class to another. These conflicting definitions caused many appraisers to question how to rate a property that is not clearly an “average” or a “good” quality house, but seems to fall in between those categories and how to consistently account for the differences from one property to the next.
As party of Fannie Mae’s data collection, they found more weaknesses in the analysis of the appraisals obtained. It seemed there was little correlation in the age and size adjustments made to comparable sales, regardless of property type or price. This was not a new revelation to any practicing review appraiser. Data revealed that many appraisers were adjusting comparable sales based on antidotal data, at best.
For instance, one case found that appraisals completed by the same appraiser on vastly difference properties ranging from manufactured homes to large, custom homes applied the same age adjustment of $1,000 per year and $35 per square foot. The properties in questions were selling at $50 per square foot in one report and well over $200 per square foot in another. Fannie Mae found that, across the country, similar adjustments were being applied, regardless of price per square foot.
Fannie Mae’s solution to this problem was to require the appraiser to better support the adjustments being made to the comparable sales by the appraiser. Thus, “bracketing” has become a watch word for loan underwriters country-wide. As typically applied by most lenders, a subject with a physical characteristic must present at least one comparable with the same physical characteristic. In other words, if the subject has a swimming pool, then at least one (preferably two) comparable sales used should have swimming pools as well.
Problems arise, however, when a comparable sale cannot be located with a similar feature, then the contribution value of that feature is called into question. This is particularly a problem with underwriters who unrealistically expect every physical feature of the property to be bracketed, including features with adjustments that are very small and reflect a preference for that feature in the market, but that have a minimal impact on the value conclusion.
For instance, not bracketing the Gross Living Area (GLA) of a house and relying on sales with adjustments of 5 to 10 percent of the sale price is much more problematic than using sales that do not bracket the site size of the subject where the adjustment rates may be 1 to 3 percent of the total sale price. To ask an appraiser to bracket the site size of the subject by including an otherwise dissimilar sale is not necessary, but asking the appraiser to bracket the GLA of the house where large across-the-board adjustments have been applied would most likely result in better support for the value opinion.
Alternatively, common sense would suggest that a feature or characteristic (such as a swimming pool) is desirable and most buyers in that market would pay something extra for that feature, but homes near the property with similar features cannot be located. This leaves the appraiser with the conundrum of either making the adjustment per market conditions, or stating that the feature makes no contribution to value because it cannot be bracketed.
The Absolute Ratings system and the expectation to have every physical feature bracketed has caused an increase in revision requests from underwriters. With multiple correction requests from underwriters per report, many appraisers have adjusted their reporting behavior to reduce the number of revision requests. Unfortunately, many of the underwriters requesting the revisions have little or no property valuation training and are simply going through a checklist. As a result, the following trends have been observed by reviewing appraisals:
√ If the appraiser rates everything the same with no or few adjustments, the appraiser has far less revision requests.
√ Many appraisers are reluctant to take on appraisal assignments in rural areas or accept appraisal assignments for non-traditional properties.
√ Where there are relative differences between the subject and the comparable sale, no adjustments are made.
In an attempt to aid appraisers with better tools in supporting adjustments, all kinds of new analytical tools are being developed by software venders. Many of these tools can be very helpful if used properly. However, use of these tools require appraisers to up their skill set to properly when the results from these tools are not reliable. I am aware of appraisers who have been disciplined by state licensing boards for using regression tools incorrectly.
In one interview with a seller, the property being appraised had an updated kitchen with granite counters and tile floors. The appraiser had used comparable sales with laminate counters and vinyl floors with no adjustments made. When asked why no adjustment was made, the appraiser explained that those things don’t make any difference in the value. The appraisal was below the purchase price and there were some very happy people involved. Unfortunately for the appraiser, the seller also happened to be an appraiser.
Several appraisers in Utah got a hold of a regression application that seemed like a very useful tool, providing attractive outputs in graphs and tables. One of the appraisers shared this new find with some associates without getting any training on the use of these tools, which resulted in disciplinary for those involved by the Appraisal Board for USPAP violations.
Collateral Underwriter (CU) introduces a new layer for appraisers, most of which are completely unaware of how it works. In fact, Fannie Mae is likely the only entity that actually knows how the rating system works and they don’t seem willing to share.
CU is an underwriting tool used by Fannie Mae that provides a 1 to 5 risk rating for a property based on the appraisal, with 1 being low and 5 being high.
A CU score is not supposed to rate the appraiser, only the risk associated with a property. Appraisals that come back with a risk rating for the property of a 4 or 5 put the originating lender at risk of a potential buy-back request by Fannie Mae. Most lenders do not want to repurchase a loan once it has been sold to Fannie Mae. The result is that appraisers doing work in areas with limited data or diverse property types often end up producing appraisals that have a higher risk rating (no fault of the appraiser, just the nature of the property being appraised). Apparently CU has data from all the appraisals being completed and sent to Fannie Mae together with public record data. CU takes the information from the appraisal, develops a model and evaluate sales that it predicts are most similar to the subject. When sales selected by the appraiser are not consistent with the sales predicted by the model, an appraiser may be required to explain why certain sales were not used.
I recently completed an appraisal for a newer home in a tract development where they had only built a few two-story homes over a walkout basement. In an effort to locate homes that were most physically similar to the subject, the sales selected were located several miles from the subject in competing areas where predominant values were similar and the neighborhoods were considered to be interchangeable for most market participants. The CU score came back at a 5. The only feedback was that one of the comparable sales I used had three and a half baths, but apparently my peers showed something different. There was one comparable sale behind a small neighborhood park with a small beneficial view, and apparently my peers did not rate the view the same. In the overall scheme of things, these differences may have had an impact of less than 2 percent. We completed an usual house in a rural area that we were sure was going to come back with a high-risk rating. The lender said the CU came back at 1-go figure.
Fannie Mae has not disclosed how the CU scores are developed. We have had some lenders suggest that CU scores are increased when the appraiser uses sales that are not predicted by the CU model. We have been told that where the appraisal has lots of differences between the subject and the comparable sales that result in zero, that it increases the CU score. For example the more times that appraiser inputs a zero where there is a difference in the description, the higher the CU score is. For example: An appraiser rates a view for the subject as a N;Res; (Neutral Residential) and a view on a comparable sale as a N;Mtn; (Neutral Mountain). If there is no difference in value, the appraiser should put a zero, reflecting that the difference has no value difference. The same applies for differences in design, where one home is rated as a Rambler and the other is a Ranch. A zero would be provided to reflect that there is no value difference. Where there is any difference in porches, patios, landscaping, etc., the more zeros that show up on the report, the higher the CU score. In many cases, there are not hard and fast definitions for design or view, but the way the appraiser presents and adjusts the data may potentially increase the CU score.
Lenders want to use appraisers with a history of lower CU scores. This presents a problem for appraisers doing work in rural areas where the risk associated with the properties is understandably higher due to the nature of the market area. The properties in these areas have poor market data and a limited pool of comparable sales with significant property differences. With lenders looking for lower CU score appraisers and the inevitability of rural properties to produce high CU scores, not only does it limit the number of appraisers willing to do the work, but has also created concern that lending for properties in rural areas will be drastically limited.
Fannie Mae recently introduced a new refinance option for lenders: Day-One Certainty. This loan program allows a lender to refinance a property without an appraisal and the low fee ofS75 to use as a proprietary valuation model to guarantee that the lender will not have to repurchase the loan as a result of the appraisal or valuation. It would seem that any lender would prefer to use this loan over any loan program that requires an appraisal, as it is faster, less expensive and has virtually no risk of loan repurchase. The irony is that the Automated Valuation Model used is typically a less reliable valuation product, but with lower risk to the lender … remember 2006? In Fannie Mae’s defense, the program is reported only to apply to a very small number of properties that fit certain criteria. However, with increasing appraisal fees and difficulty in finding appraisers in some areas, the program is very attractive for lenders. Will the program guidelines be expanded? One thing is certain: There is pressure in the market to find lower-cost valuation solutions and as appraisers, we need to be aware of those needs and be part of the solution!
While there are concerns about the changing nature of the industry, I am confident that the appraisal profession can be both a profitable and rewarding career for those who are forward-thinking and can embrace new tools so they can succeed in the quest for the right value conclusion.