fannie's collateral underwriter - big changes and what it ...today+1... · fannie's...

IN THIS ISSUE

Disability - your greatest risk . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .4

Fannie's Collateral Underwriter - big changes and what it means for you

January 2015–©Appraisal Today–PAGE 1

VOLUME 23 • ISSUE 1 • January 2015

Fannie wants to reduce appraisalrisk and has been collecting UAD

data since late 2011 to do this. WithCollateral Underwriter (CU) everyappraisal will receive a risk scorefrom 1.0 to 5.0. This does not meanan appraisal is "good" or "bad". Itreflects the risk for the lender. Forexample, a huge house in a tract ofsmall homes or declining values.

CU is an appraisal review tool,much more sophisticated than previ-ous rules-based tools. .

In my opinion this is finally a gooduse for the UAD data that appraisershave been providing for three years.Hopefully, the skill level and profes-sionalism of residential appraiserswill be increased.

It will make us get more supportfor our adjustments and think moreabout our adjustments.

More important, it will identifyappraisers who are not competent, tryto make values higher, put incorrectcomps and analysis in their reportsjust to get appraisals done faster,don't check for typos, etc. Someappraisers are already receivingwarnings for inconsistent Q&C rat-ings, and low GLA adjustments.Many more warnings will be sentout.

For appraisers, the most significantchange will be a focus on adjust-ments. Your adjustments will becompared with your "peers" andFannie's "model" which is regressionbased.

Note: I have used some directexcerpts from Fannie documents.They are included in quotationmarks.

Acronyms• CU - Collateral Underwriter - start-ing 1/26/15 - will be integrated intoDU • DU - Desktop Underwriter - inter-face for underwriters for credit andcollateral - used for many years • UAD - Uniform Appraisal Dataset -used since late 2011• UCDP - Uniform Data CollectionPortal - connects to CU starting1/26/15. Significant changes in mes-sages.

What about FHA, VA, and non-Fannie loans

No UAD. No CU. However, theymay start looking for adjustment sup-port as Fannie tends to be the mort-gage leader. Fannie is pro- appraisers

In contrast with many in the lend-ing industry, Fannie has always beenpro-appraiser.

FIRREA passed in 1989, which didnot require appraisals for transactionamounts under $250,000. But, Fanniestill required appraisals.

They consider the "boots on theground" appraisers to be the mostknowledgeable and objective sourceof values and description of a proper-ty. AVMs, real estate agents, Zillow,etc. don't even come close.

What are the pluses for appraisers?I really think that CU will make

better appraisers of most of us. I havenot been spending enough time onadjustment supports. Comparing myappraisals with those from otherappraisers will make me think more.

www.appraisaltoday.com

http://www.appraisaltoday.com

©Appraisal Today–January 2015

I will be more careful checking fortypos.

If you work in rural areas, CU willbe able to determine if the comp dis-tance from the subject is appropriate.

If you appraise unusual homes, andwho doesn't, CU will consider factorsothe than distance, such as time, ageand sq.ft. when determining whichcomps are appropriate.

Fannie is focusing on what isimportant, instead of checking to seeif the boxes are checked correctly orbeing perfect on filling out UAD,which have little or nothing to dowith values.

For appraisers who cut corners or"push" values and make us all lookbad, they will be identified and getwarning letters, or worse.

Working for low fees and/or fastturn times and cutting corners willbecome much more risky. Maybefewer appraisers will do this.

Maybe experienced and knowl-edgeable appraisers will be treated asprofessionals, including decent fees,and stop leaving the business.

Fannie vs. lender requirementsFannie requirements are changing.

No one knows for sure what theywill mean for appraisers, but at leastit is a widely accepted set of require-ments.

Unfortunately, lenders can askwhatever they want. If you accept anassignment be sure you know whatthey are. If the AMC does not sendyou the lender's requirements, orchange them after you have complet-ed the appraisal, you need to dumpthat AMC. Well run AMCs have webaccess to client requirements, tell youwhat they are, and don't changethem.

Some AMCs have very long listsof requirements. I suspect they arecompililed from all their lenders. Youneed to dump these AMCs.

Will appraisals take more time todo?

I suspect that when CU starts therewill be a lot of confusion. I don'tknow if appraisals will take moretime or less time because of CU dur-ing the adjustment period.

If you have not been using yourcomps database, you will need tostart using it.

If you don't have support for theadjustments that CU looks at, youwill need to get ready for adjustmentquestions.

If you try to make few, or noadjustments, you will have to changeand learn how to make adjustments.

There may be fewer stupid requestsfrom Fannie's UCDP. The CU warn-ing messages are much more specific.

If you don't carefully check yoruappraisals for typos, there will bemore time required to fix your errorslater.

If you try to get higher values, youwill probably get warning letters, orworse. That will be very good for theappraisal profession.

What is Collateral Underwriter?Per Fannie: "Collateral Underwriter

is a proprietary appraisal reviewapplication developed by Fannie Maethat performs an automated analysisof appraisals submitted to theUniform Collateral Data Portal(UCDP)."

"CU leverages an extensive data-base of property records (and salestransactions), market data, and pro-prietary analytical models to analyzekey appraisal components includingdata integrity, comp selection, adjust-ments, and reconciliation ".

Results of CU's automatedappraisal analysis include the follow-ing:"• A comprehensive Risk Score on ascale of 1.0 (lowest risk) to 5.0 (high-est risk) Identifying both high valuesand why• Risk Flags to identify factors con-tributing to high risk scores

• Detailed messaging to highlightspecific aspects of the appraisal thatmay warrant further attention• The CU web interface providesadditional content and functionalityto assist with deeper analysis of theappraisal.• Fannie Mae utilizes CU as part ofour ongoing appraisal quality andcollateral risk management efforts."

What is the purpose of CU?Per Fannie, "The purpose of

Collateral Underwriter is to identifyappraisals with heightened risk ofproperty eligibility or policy compli-ance violations, overvaluation, andappraisal quality issues."

"CU provides lenders with addi-tional transparency and certainty byproviding them access to the sameappraisal data and analytics used inFannie Mae's quality control frame-work."

"In summary, our objective is todistribute CU to support more proac-tive management of appraisal qualityby empowering lenders to addresspotential appraisal issues prior to loandelivery."

Comparing your report to your"peers". My experience fromrelocation appraisals

I have been doing relocationappraisals for almost 30 years. Myappraisals are compared with one ortwo other appraisers. We all appraisethe same property.

Here are a few examples. It is veryseldom that we have the same GLA.If we do, it may be suspicious.(Maybe we compared "notes" on theappraisal.) One appraiser made alarge view adjustment and one of theother appraisers did not. Are the two(or three or more) values close, ornot? Why?

For many years, there was a smallgroup of local appraisers who didrelocation appraisals. Our descrip-tions of the subject and market condi-tions were similar. We used similarsales and listings.


But, more recently, many experi-enced relocation appraisers are cut-ting way back or quitting appraising.Sometimes the other appraiser(s)have little relocation experience, orexperience in the market. I have tospend a lot of time explaining what Idid as compared with the otherappraiser. I did not like this andrecently have been turning downmore relocation assignments.

I have no idea what the "peers"Fannie will be comparing us to. Howmuch experience do they have? Whattype of education and training dothey have? Do they have localknowledge?

Maybe all appraisers will not beseen as the same in the future. MaybeAMCs will select appraisers that aremore knowledgeable, can explainwhat they are doing, etc., particularlyfor the higher risk loans.

What data on adjustments will beavailable to the underwriter online?

The comp selection informationbelow will be accessed throughFannie's UCDP Web interface.• Comps up to one year old from theappraiser and other data sources(CU-other appraisers and publicrecords). Fannie selects additionalcomps by using Census Tract Blocks(CTB). The comps are ranked.Warning messages are seen.Appraisal comps and 20 modelselected comps are rated on a spread-sheet. Users can select and deselectcomps and sort comps. Users can doa new comp search by using criteriasuch as GLA and age.• Aerial and street views for the sub-ject and comps by clicking a button• Local market trends graphs• "Heat maps" displaying CensusTract Block Group-level statisticssuch as median sales price,price/GLA, median days on market,etc.

I have included several pages fromFannie's Underwriter webinars at theend of this newsletter to see whatthey look like.

What are Census Tract Blocks?Unfortunately I was unable to find

a source of maps for CBG so you cansee where Fannie looks for comps.When I find a source, I will let youknow.

Before census tract numbers wereavailable online, I used census tractprinted books, and still have them inmy office. Often, in my area, theywere a good indicator of small neigh-borhoods boundaries. This is the onlywidely accepted way that Fannie candefine neighborhoods for CU to use.

Fannie says they prefer market areaboundaries that the appraiser says isappropriate, but for the underwriterto use this area, it has to be inputmanually into CU by the underwriteror reviewer.

Underwriters and appraisalsAbout 20 years ago, when Desktop

Underwriter was first starting, Iattended a Fannie training class forunderwriters.

They all said they hated to under-write appraisals because they werehard to understand and evaluate.

Compared with appraisals, creditunderwriting is much, much easier asit is based on facts - income, creditscore, debt, etc.

To help underwriters, later UCDPwas set up and specific warning mes-sages were added to help underwrit-ers evaluate appraisals.

With CU, the messages are muchclearer and are based on facts, suchas an appraisal which has GLA thatvaries significantly from otherappraisers and from the appraiser'sprevious appraisals. Rather than thewarning that the GLA in the subjectdescription was not the same as inthe sales comparison grid. Or, thebedroom count does not correlatewith the room count. And, of course,

from the appraiser's point of view,who cares?

Will appraisers still be needed?Yes. Appraising is an art and a sci-

ence. AVMs have limitations. Only"boots on the ground" appraisers willunderstand everything that influencesproperty value.

For example, I think that CU willput a premium on experienced,knowledgeable appraisers who canresolve issues with appraisals withhigh risk scores. Unfortunately, manyhave left appraising and/or will beleaving soon due to AMCs.Hopefully, CU will slow down the"brain drain".

I am predicting that the higher therisk rating, the more experienced,knowledgable appraisers will beneeded. I don't think AMCs will bebroadcasting orders for $200 to allthe appraisers on their lists for theseappraisals.

UAD has standardized propertydata for the first time

Standardized property data did notexist until the UAD. MLSs vary onwhere they get their data. Assessor'soffices vary widely and some are notcomputerized, although some AMCsseem to consider them more reliablethan appraisers' data. Some states arenon-disclosure, so sales prices are notavailable publicly. Another problemis how recent is the data? Almostevery data source has outdated infor-mation, except appraisals submittedfor loans.

Most consider appraisal data to bethe "Holy Grail" as it is consideredthe most reliable, but until UAD itwas not standardized and was notavailable.

Freddie and Fannie have been col-lecting data from appraisals for manyyears, but it was hard to use as it wasnot standardized. Appraisers havebeen supplying Fannie with UADdata since late 2011.

Some time recently, Fannie started

hoods. I always know what is impor-tant to buyers in a particular market,and that an adjustment is needed, butI don't know the exact adjustmentamount.

What makes me nervous is that mystate regulator wants to see supportfor my adjustments, not that Fanniewants to see it. If I lose my appraisallicense I will not be able to get muchappraisal work, even non-lenderwork. Fortunately, much of my workis commercial and 5+ unit apart-ments, where dollar adjustments aretypically not used.

What adjustments do you need tosupport or respond to whenquestioned by an underwriter orreviewer?

Below I have a list of what adjust-ments CU will be looking at.

No, it will not be looking at pools,outbuildings, etc.

CU looks at what Fannie considersimportant and works well withregression models. GLA, bedroomand bath counts, location, view, base-ment size, Q&C ratings, age, etc. Seethe list below.

Fannie is also using other analysismethods such as comparing Q&C rat-ings that are not consistent among anappraiser's appraisals.

What appraisal data will Fannie beanalyzing?

There is lots of UAD standardizeddata, but much of it is not used in CUnow. For example, Design and Carstorage. Text fields are also not usedas they are not standardized.

But, for now, Fannie will be usingthe UAD standardized data belowfrom the subject and comps:• Sales price• Lot size• Above-grade bathroom count• Above-grade bedroom count• Age• Below-grade area• Finished basement area

PAGE 4–©Appraisal Today–January 2015

using the UAD data. It was first usedinternally by Fannie. I suspect it wasused Appraiser Quality Management(AQM). A few appraisers were pro-hibited from doing appraisals forFannie loans. As of December, 2014,the new list did not include any newnames. Hopefully more of the "pushthe value" and "get them done as fastas possible" appraisers will be identi-fied.

What date did UAD start and howmuch data has been collected?

UAD started for conventional loansdelivered to the GSEs on or afterMarch 19, 2012 (and with applicationdates on or after December 1, 2011).

Over 14 million appraisals andover 20 million transaction recordshave been collected by UCDP so far.About 20,000 appraisals are comingin every day.

When will CU start?The effective date is January 26,

2015. Lender "beta testers" startedusing it about 6 months ago. Youmay have already received some ofthe warning message requests forcorrection.

Why is Fannie using CU?CU is being used for collateral risk

reduction, before the loan is made.Credit risk reduction is much, mucheasier as it is based on facts, such ascredit scores, income, etc.

Appraisal risk levels will be quan-tified, with a score from 1.0 to 5.0,allowing lenders to focus onappraisals identified as higher risk.

Typical for lending, there are sig-nificant changes after crashes, suchas the 1989 S&L crisis, whichinvolved primarily commercial prop-erty loans. This resulted in FIRREAand appraisal licensing.

Looking at the recent past, Fanniehad been only looking at loans afterthey were closed, which is too late to

identify problems, as we all learnedin the 2008 mortgage crash.

What you MUST do now - haveconsistent UAD data among yourappraisals

You can't control what your "peers"do, but you can make sure your datafor a comp is consistent when ever itis used.

Use your software vendor's data-base and populate it with all yoursubject and sales back to when youfirst started using UAD. Hopefully,you have already done this.

Even if you think that you seldom,if ever, use the same subject or compmore than once, you may be sur-prised. Also, UAD data is only 3years old now, but as the years passyou will not be able to remember ifyou are using comps again.

Otherwise, you will not be able toavoid having different data in differ-ent appraisal reports. For example,for a comp at 123 Smith Ct. in XXcity for an appraisal for 234 Jones St.you used 4 bedrooms. When youused the same comp in anotherappraisal you used 3 bedrooms. Thisprobably will result in a CU warningmessage.

There may be a good reason whyyou changed the data. Write it in yourreport. Even if it does not get read bythe underwriter or reviewer, it is thereand you don't have to spend timelooking for it.

Support for adjustments - a verybig change for appraisers

No one knows yet how much, andwhat type of, support underwritersand reviewers will be looking for.Fannie does not say anything abouthow to support adjustments, onlysends a message that yours are notconsistent with the "model" or yourpeers adjustments.

I have been appraising for 40 yearsand often don't have good "support"for adjustments as I don't typicallywork in conforming tract neighbor-


• Location• Location rating• View• View rating• Quality• Condition• GLA• Total bedroom count• Total bathroom count• Total living area

Be sure you check your compsdatabase to be sure you are using thesame data you used before.

How to provide support foradjustments

What Fannie does not like is low,or no, adjustments just to makeappraisals "look good" and get few"stips".

If you work in large conformingtracts, it is reasonably easy.

AMVs are regression based.Regression works well in conform-ing tracts with conforming homes.The farther your subject is from this,the less reliable is regression.Regression is typically best on GLA,which all properties have.

You will have to use other meth-ods, such as matched paired sales,broker interviews, cost analysis, etc.

Some appraisers have been doingregression analysis on MLS and pub-lic records, using their own methodsor software they have purchased.

There are some non-statisticalmethods, such as cost approach andmatched paired sales.

I will be looking at software that isavailable for adjustments in theFebruary issue of Appraisal Today.

So far, the best I have seen, andeasiest to use, is Redstone, devel-oped by Bradford forms software.Bradford developed several products,CVR and Compcruncher, that werenot widely adopted. Bradford hasbeen working on these products for5-6 years. They used these product tohelp develop Redstone. Pricing willbe per use, tentatively $5 to $15 per

use. I will be reviewing this in theFebruary issue.

On which properties will Fannie notbe able to obtain data?

Fannie loans not using fullappraisals will not provide UAD data.

Data from loans not sent to Fanniewill not be captured, for examplejumbo loans, VA, FHA, etc.

There have been many all cashsales with no loans on purchase.However, some investors refinance sothey can get money to buy more prop-erties.

Is anyone else using risk reductionsoftware?

VA is using LoanSafe fromCorelogic which analyzes appraisalsfor risk, but does not capture appraisaldata. However, VA is now requiringUAD so they are planning something.

Is CU required for lenders?No.

What about AMCs?Their lender clients decide what

they send to their AMC vendors.Some have access to the UCDP portaland some don't.

AMCs will not have access to CU.They may have access to UCDP. Theywill be able to view risk scores, flags,and messages in UCDP, but will nothave access to the Web-based CUapplications.

CU will generate warning messagesand hard stops, but it is the lender'sdecision which to send to their AMCvendors.

The only part of Fannie's two webi-nars that addresses AMCs is from theend of the second seminar(Understanding the CU Risk Score...)with recommendations to AMCs andLenders.

Here is a direct quote: "We'vetalked a lot about lenders' use of theCU messages, but realize manylenders have designated appraisalmanagement companies, or AMCs, as

your agents to submit to UCDP onyour behalf and consume GSEappraisal feedback. If that is the case,please make sure you work with yourAMCs to clearly establish policies andprocedures for how they may use theCU results. You should also confirmthat they've attended our overviewand training sessions."

What about reps and warranties andbuy backs?

The reason that lenders keepincreasing their requirements is thatthey are afraid of buy-backs, wherethe lender has to pay Fannie back forthe cost of loans that have defaulted.Representations and warranties reducethis problem for lenders. Fannie isloosening their requirements for creditto help reduce this risk for lenders.

Per Fannie: "Fannie Mae is not cur-rently offering representation and war-ranty relief with the use of CU.However, CU will be integrated withDesktop Underwriter (DU) in the firsthalf of 2015 to give lenders a moreholistic view of risk by displaying theCU risk scores, flags, and messages inDU."

"The integration will provide afoundation for future rep and warrantyrelief on property value. Fannie Maeis working with our regulator, FHFA,on details and timing, and we willprovide more information as itbecomes available."

Fannie will be using UAD appraisaldata and public records

Fannie will have data from all theUAD appraisals plus public records,but lacks full MLS data.

Appraisers have full MLS data butno UAD data

I don't think Fannie has MLS datadirectly from the MLSs. They mayhave some from a data partner. Lackof full MLS data could definitelymean that the Fannie model will nothave access to the primary data sourcefor almost all appraisers. This could


Individual Cooperative InterestAppraisal Report; and Form 2095,Exterior-Only Inspection IndividualCooperative Interest AppraisalReport."

What about rural areas?Per Fannie, "In rural areas, it may

be necessary for the appraiser toselect sales from a greater distanceand further back in time than in areaswith higher sales density and morehomogeneous housing stock. CUcompares the appraiser-providedcomps against a pool of observedsales transactions in the subject mar-ket and not against arbitrary, "rule ofthumb" guidelines."

"CU will not provide a high riskscore solely because the comparablesare dated, located several miles away,or require significant adjustmentfrom the appraiser. If CU's compara-ble selection model finds the apprais-er-provided comparables to be thebest available, the appraisal will like-ly receive a low risk score if no otherpotential issues are detected."

Building permitsFannie documents refers to build-

ing permits several times, but doesnot discuss how they will be used.

They could be used to see if the aproperty has been remodeled. Forexample, an appraiser says averagecondition, but there has been exten-sive remodeling or the opposite.

Fannie will only have access toonline sources, such as Buildfax.

What are warning messages?CU generates about 110 appraisal

warning messages. Lenders are notrequired to "fix" them. They may, ormay not, send the messages to theirAMC vendors.

See the attached "Appendix A" forthe list of warning messages.

What are "hard stops"These 21 messages stop the

appraisal from being submitted to

definitely cause data discrepanciesbetween appraiser comps and theFannie model's comps.

Particularly relevant for appraisalsis pendings and listings. Be ready toaddress this problem when askedabout the comps you used and yourdata source.

What geographic coverage does CUhave?

Per Fannie "CU is able to scoreabout 97 percent of appraisals sub-mitted from the 50 U.S. states andthe District of Columbia. Propertiesin the U.S. territories cannot be ana-lyzed."

What appraisals will not get CUrisk scores?

"The primary reason for unscoredappraisals is an inability to geocodethe subject property or an inadequatenumber of appraiser-provided com-parables. Rarely is a lack of alterna-tive sales observations the cause ofan unscored ("999") appraisal."

"Currently, CU is unable togeocode properties in the U.S. terri-tories (Puerto Rico, the VirginIslands, Guam, American Samoa,and the Northern Mariana Islands).In some cases, CU may be unable togeocode new construction becausethe address is new and not yet recog-nized by the geocoding service."

What appraisal forms will be used?Only 1004 (Uniform Residential

Appraisal Report) and 1073(Individual Condominium UnitAppraisal Report) will be used.

"Appraisals submitted on otherforms, including the following, arenot analyzed: Form 2055, Exterior-Only Inspection ResidentialAppraisal Report; Form 1075,Exterior-Only Inspection IndividualCondominium Unit AppraisalReport; Form 1004C, ManufacturedHome Appraisal Report; Form 1025,Small Residential Income PropertyAppraisal Report; Form 2090,

Fannie Mae. They must be fixed or"overrided" by the underwriter. Manyare probably "typos", but some are agood way to identify appraisers andsupervisory appraisers who are notallowed to submit appraisals toFannie or require 100% review.

Per Fannie: ..."lenders will receivea "Not Successful" status when oneor more of the 21 appraisal messagesare issued. Lenders will be requiredto review the Fannie Mae appraisalmessage(s) to verify if the informa-tion is correct as submitted or if anew or corrected appraisal isrequired. If the information is veri-fied as correct and it is determinedthat there is no impact to loan eligi-bility, the lender may request a manu-al override and provide a reason codeto change the submission status to a"Successful" status in UCDP."

Examples:• The appraisal indicates that the sub-ject property has legal nonconform-ing zoning and cannot be rebuilt tothe current density. This data indi-cates that the property is ineligiblefor delivery to Fannie Mae. • The sales contract was not analyzed• State certificate is not provided ontransaction amount over $1 million• Appraiser license does not matchsubject property state.

See the attached "Appendix B" forthe list of hard stop messages.

CU warning messages aresignificantly different from previousmessages

Fannie Mae started using warningmessages in UCDP a few years ago.

Per Fannie, "The Fannie MaeProprietary Messages in UCDP(were) rules-based alerts focused ondata reasonableness, property eligi-bility and policy compliance. CU ismodel-based and performs a morecomprehensive analysis of dataintegrity, comparable selection,adjustments, and reconciliation."

Example: Number of bedroomsOld warning (FNM0025): the sub-


ject property's number of abovegrade bedrooms relative to the totalnumber of rooms appears excessive.

New warnings • (#FNM0409): The appraiser hasreported materially different above-grade bedroom counts in one or moreappraisal reports• (#FNM0410): The reported above-grade bedroom count is materiallydifferent than what has been reportedby other appraisers

Per Fannie, "Prior to CU, FannieMae Proprietary Messages werelaunched in 2013. These messagesfocused primarily on data reasonable-ness, property eligibility, and policycompliance. Those messages were agreat start, but Fannie Mae has alsobeen working for several years on thedevelopment of more advancedappraisal analytics, which led to thecreation of Collateral Underwriter.Now, we are able to perform a morecomprehensive analysis of theappraisal, focusing on things likedata integrity, comparable selection,adjustments, and value reconciliation,by leveraging an expansive databaseof market data collected throughUCDP and proprietary analyticalmodels."

How much data does Fannie havefrom different appraisers for anindividual property?

Some of the warnings result fromyour appraisal being compared withappraisals from your "peers".

Fannie says they have, on average,5 data sets for each property - subjector comps, from appraisers. To me,this seems like a lot of records oneach property accumulated over onlythree years.

Science vs. art in appraisingI have been writing about AVMs

since 1994. They sometimes havesevere limitations. They only get cer-tain types of data. There are oftensignificant discrepancies among dif-ferent data sources.

For example, 5 appraisers havevery similar GLAs on a property, butthe 7th appraiser has another number.Who is correct? Sometimes it is the7th appraiser who saw somethingthat the other appraisers missed?

I now work in a very small geo-graphic area, my city. I regularlyattend brokers open houses and knowall the local agents. I follow localdevelopment trends. I am veryknowledgeable in my city. However,previously I worked a wider geo-graphic area. I was definitely not asinformed in other cities.

That's why Fannie realizes thatappraisers are important. When thereis a high risk score, they will need touse these experienced appraisers tofigure out what is happening.

Why don't appraisers have accessto CU?

Fannie wants appraisers to do inde-pendent appraisals. Only appraisersare "boots on the ground" and cansee what computers can't see.

If appraisers have access to CU,they will use computer softwareinstead of their own analysis. Whenthey find out what CU says, they willbe tempted to just use it.

Even the data is difficult to use asthere are multiple sources of thesame data. GLA is a good example -public records plus an average of 5appraisers.

Who has access to CU?"CU risk scores, flags, and mes-

sages will be provided in real timeafter an appraisal is submitted toFannie Mae through the UniformCollateral Data Portal® (UCDP®).The CU risk scores and messageswill be provided on the Fannie Maetab and the UCDP SubmissionSummary Reports. Any UCDP userwho submits an appraisal to FannieMae will have access to the CU riskscores, flags, and messages."

"For in-depth appraisal analysis,CU will be available to Fannie Mae

sellers via a web-based application(registration required).Correspondents will also have accessto the web-based application in 2015,but on a different timeline fromapproved seller availability."

"Other UCDP users that are notFannie Mae sellers or correspondents– including appraisal managementcompanies (AMCs) that have UCDPaccess as lender agents – will be ableto view the CU risk scores, flags, andmessages in UCDP, but will not haveaccess to the web-based application."

Does CU provide a value or specificadjustments?

Per Fannie, "CU does not providean estimate of value. CU is intendedto be an appraisal review utility and isnot an automated valuation model, orAVM. Lenders will be alerted toappraisals with potential overvalua-tion, but will not receive a value or arange of values."

"CU does not provide specificadjustment numbers. It comparesadjustments in the appraisal with theModel, adjustments in otherappraisals done by the same appraiser,and with other appraisers' adjust-ments."

CU Risk ScoreLenders may use the CU risk score

(from 1.0 to 5.0, with 5.0 the highestrisk) to segment appraisals by riskprofile, resulting in more efficientresource allocation, workflow man-agement, and collateral risk manage-ment processes.

Per Fannie "the objective of CU isto assist lenders with assessing prop-erty eligibility and appraisal quality. Itdoes not provide approvals or denials,nor should it be used as a basis for acredit decision."

"It's important that you think of theCU risk score in terms of "low riskand high risk," and not in terms of"good or bad". As with any automatedanalysis, there is opportunity for falsenegatives and false positives. While

CU is effectively predictive ofappraisal defects and overvaluation, ahigh score does not necessarily meanthe appraisal is bad. Lenders shouldperform additional due diligence onappraisals with high risk scores todetermine if corrections are neces-sary. If they are not, it is beneficial tohave adequate commentary from theappraiser and documentation in yourloan file."

Property eligibility/policycompliance risk flags

"You have probably already seenthe fatal Uniform Appraisal Dataset,or UAD, edits. These are the jointFannie Mae/Freddie Mac edits thatcause a Not Successful submissionstatus in UCDP and require resubmis-sion of an appraisal. CriticalProprietary Messages or "hard stops"are the 21 messages that will requiremanual override or resubmission of acorrected appraisal. They are notused very often and are discussedbelow."

"Overvaluation is critical toFannie. Sophisticated statistical mod-eling is used to identify appraisalswith higher probability of materialovervaluation. There are no moregranular messages associated withthis risk flag, but other messagingand information contained in the userinterface will assist lenders in assess-ing this risk."

Appraisal quality risk flags - DataIntegrity - GIGO

Fannie is "looking to determine ifthe physical attributes and transac-tion terms of the subject property andcomparables are accurately reported.This is the bottom of the pyramidhere because it is the foundation forany appraisal report. We all know theold saying "garbage in, garbage out".If the subject or comps are not accu-rately represented, it can influenceour judgment of the entire appraisalreport. It affects whether or notcomps look similar to the subject, thedirection and magnitude of the

adjustments, and which compsshould receive most weight in recon-ciliation."

Is the data plausible? For example,is the lot size 40,000 acres or 40,000sq.ft.?

Consistency within a singleappraiser's body of work. Is thecomp data the same? For example,for 123 Main St., the appraiser useda sales price of $200,000 and inanother appraisal used $225,000.

"Consistency between the apprais-er and his/her peers. We expect thatin nearly all cases, an appraiser'sdescription of any comp should beconsistent with the description pro-vided by their peers."

"It is important to know thatFannie Mae is not looking to splithairs or start he-said/she-said argu-ments between appraisers. For exam-ple, we would not flag small differ-ences in gross living area or cases inwhich appraisers are split down themiddle between C3 and C4 condi-tion. We are looking for material dif-



ferences and outliers." "In summary, appraisals with mul-

tiple data discrepancy messagesand/or egregious errors will get high-er CU risk scores. Users should con-firm the accuracy of the appraiser-provided data when they see thesemessages, particularly in cases wherethere are a number of errors or mate-rial discrepancies."

My comment: Of course, much ofthis is due to appraisal "typos",which we all make, or failure tocheck the previous times a comp wasused in your appraisals. Consistencywith peers is more tricky. If you dis-agree with data from MLS or publicrecords, which is used by manyappraisers, be sure to explain why inyour appraisal.

Data Integrity - Comp SelectionPer Fannie, "The appraiser is

responsible for providing comps thatare most similar to the subject prop-erty. If they are not, results of theappraisal may be compromised.

Sample warnings:"• The reported sale price is material-ly different than what has beenreported by other appraisers. This iswhat we call a "peer discrepancy".This message fires because CU seesfive other appraisers reporting a saleprice of $300,000, while our apprais-er is the only one reporting a saleprice of $325,000. Obviously, bothof these prices can't be right.• Our appraiser states a sale price of$279,900. Three other appraisersreport a sale price of $240,000. Ourappraiser has also reported $240,000in a previous report."

If an appraiser does this on all, ormost of, the comps used, this is verysuspicious.

This is another reason for usingyour comps database. If the appraiserjust likes to "push" value, this is asignificant red flag, much more seri-ous than using different Q ratingsbecause he/she did not check a compdatabase.

"The appraiser provided the bestcomparables available"

"Nearly every report Fannie Maesees has some comment to the effectof "the appraiser provided the bestcomparables available".

"CU will help lenders determine inwhich cases this comment is true andin which cases it is not."

Data Integrity - Comp selection -time, distance and physicalsimilarity

Per Fannie, "When comparableshave been selected, the appraisermust make adjustments for differ-ences between the subject and thecomparables. How they adjust is acritical factor in appraisal quality.Adjustments should reflect the typicalmarket reaction to these differences,and not generic rules of thumb."

"Collateral Underwriter's compselection model is one feature thatmakes it unique. CU takes intoaccount physical similarity, time, anddistance when analyzing the overall


relevance of comp transactions. Thesignificance of each of these factorsmay vary from market to market andmay change over time. CU does notassign a fixed weight for each ofthese factors, but instead usesadvanced statistical analysis to treateach appraisal and each market differ-ently."

"It is important to know that ameni-ties like swimming pools, outdoorkitchens, accessory units, and out-buildings are not reported in standard-ized format and therefore cannot beconsidered by the CU model. Sowhen you encounter subject proper-ties with these types of amenities,please take this into considerationwhen reviewing the CU results."

"The significance of each physicalfeature is also model-derived andmarket-specific. For example, inWashington, DC, a double lot mayonly mean a few thousand extrasquare feet of lot size, but could sig-nificantly impact value. In ruralWashington state where many proper-ties have several acres or more,another acre may contribute only afew thousand dollars to marketvalue."

"With distance, comps in closeproximity are obviously preferred ifall else is equal. However, model-derived location factors are consid-ered in addition to straight line dis-tance. We also consider the availabili-ty of other comparables and their dis-tance from the subject."

"Going back to our prior example,travelling one mile for a comparablein Washington, DC could take youacross several neighborhoods and dra-matically different price tiers. In arural area, you might go 5, 10, 20miles away or further with no signifi-cant impact on value. CU would treateach of these situations differently."

"The same goes for time. Themodel will go back one year from theeffective date of the appraisal forcomparables, but date of sale receivesmore weight in comp selection in

rapidly increasing or declining mar-kets. For example, if you were in amarket last year with double digithome price appreciation, CU wouldprobably prefer more recent sales.Conversely, if you are in a marketwith very stable values, the modelmay go back 9 to 12 months to findcomps that are more proximate orsimilar."

"CU ranks the appraiser-providedcomps against a pool of availablesales, not against arbitrary time, dis-tance, or similarity parameters."

"Appraisals with material disagree-ment between the appraiser-providedand model-selected comps willreceive higher risk scores. In thesecases, lenders should research if othermore relevant comparables wereavailable but not utilized by theappraiser. If so, you may wish to askthe appraiser to consider these salesor provide explanation of why theywere not used."

I have included the Comp SelectionExample in the attachment to thisnewsletter. It is on Page 9 of theFannie Webinar "Understanding theCU risk score, flags, and messages.

Adjustments"Collateral Underwriter produces

statistically-derived market-specificadjustments for all UAD-standardizedphysical characteristics, date of sale,location, and sales type.For physical similarity, CU usesregression analysis to produce adjust-ments for property features. Differentmodels are used for single family andcondo property types. The magnitudeof the adjustment for each physicalfeature may vary from market to mar-ket."

"CU also makes time adjustmentsto comparables. These adjustmentsmay be negative or positive. The fre-quency of time adjustments fromappraisers is startlingly low, particu-larly over the past few years when wehave seen rapid home price apprecia-tion in many markets. In fact, looking

at some of the hottest markets, wesaw time adjustments on only fivepercent of comps over 6 months old." "There is a common misconceptionthat Fannie Mae will not accept posi-tive time adjustments – in fact, FannieMae is looking for an accurate marketvalue, and accepts both positive andnegative time adjustments as appro-priate."

Differences in location are alsoadjusted for. In absence of clearneighborhood definitions, CU usesCensus Block Groups, or "CBGs" asdefined by the US Census Bureau,and make adjustments based on saleprices within these areas."

"CU also makes market-basedadjustments for Sales Type. REOsales, Short Sales, Relocation salesand non-arms length transactions areadjusted against arms-length transac-tions based on observed discounts andpremiums between these sales types.This too, varies from market to mar-ket."

"The key point here is that CU doesnot adhere to the widely recognized –and often misinterpreted – 15% netand 25% gross adjustment guidelines.Many believe that Fannie Mae willnot accept comps with adjustments inexcess of these guidelines, which isalso not true. Long-standing relianceon generic guidelines has led to whatFannie Mae believes is widespreadunder-adjustment by appraisers. Asstated in our Selling Guide, adjust-ments may be an indicator of compa-rability, but lower adjustments are notnecessarily more accurate adjust-ments."

"In summary, rather than take a"lower is better" approach to theadjustments, CU will flag adjustmentsthat are in the wrong direction or aresignificantly different than theappraiser's peers and the adjustmentsderived by our statistical models. Inthese cases, lenders should determineif the appraiser's adjustments are ade-quately supported and reflective ofmarket reaction."


the data. It may have been first usedin Appraiser Quality Management(AQM)to get rid of the very "bad"appraisers. A few appraisers wereprohibited from doing appraisals forFannie loans. UAD data providedFannie with objective, standardizedinformation, such collecting thenames of all appraisers for all Fannieappraisals. For example, an appraiserwho was signing off as the onlyappraiser doing 100 or moreappraisals per week. (As ofDecember, 2014, the new AQM listdid not include any new names.)

Recently, Fannie has been sendingwarning letters to appraisers aboutinconsistent use of Quality andCondition ratings by comparing thedata from all their appraisals andfinding inconsistencies. In the pastmonth or so, appraisers have beengetting warning letters about usinglow GLAs. Using UAD data Fanniefound that $25 per sq.ft. was widelyused all over the country, even thoughthere were substantial difference insales prices and construction costs.

Now, after three years of data col-lection, with Collateral Underwriter,Fannie will be using more of theUAD data.

Fannie compares your appraisaldata and adjustments with "otherappraisers"

There is no way you can knowwhat every "other appraiser" will do.But, I have some suggestions below.The phrase used is "materially differ-ent".

On the plus side, this will "catch"appraisers who change the data tomake fewer adjustments, higher Qand C ratings, etc.

Fannie is looking primarily forappraisers who overvalue propertiesand for incompetent appraisers.

For example, Quality and Conditionratings are not exact and it is hard toknow what other appraisers will use.But, if the subject is old and notupdated, you would not expect anoth-er appraiser to use a high rating.

there can be problems if the reconcili-ation is not sound. To provide anaccurate opinion of value, appraisersshould give the most weight to themost relevant comparables in theirreconciliation."

"In the example, The first thing wenotice is a very wide range of saleprices, from $510,000 to $720,000.Even after adjustments, the range stillextends from under $550,000 to$720,000. Comps #1 and #2 appearquite similar to the subject and arelocated on the same street. They soldfor only $510,000 and $550,000respectively. Comp #3, however, soldfor $720,000."

"So one of these things is not likethe other. Why may that be? Closerexamination of this comp shows ithas the largest GLA, nearly threetimes the lot size, and is farthest awayfrom the subject. Research suggeststhat property values in this neighbor-hood are dramatically higher than inthe subject neighborhood."

"So not only is it physically theleast similar to the subject, but it islocated in a higher-priced neighbor-hood. Nonetheless, the appraisergives nearly all weight to this compin reconciliation. So while this is"bracketed", the appraiser is givingmost weight to the least relevant com-parable, and as a result is inflatingvalue."

My comment: I have included thefull Reconciliation Example in theattachment to this newsletter. It is onPage 13 of the Fannie Webinar"Understanding the CU risk score,flags, and messages. The appraiserdefinitely appears to be "pushing"value.

How Fannie uses UAD dataFreddie and Fannie have been col-

lecting data from appraisals for manyyears, but the data was not standard-ized and was difficult to use.

Appraisers have been supplyingFannie with UAD data since late2011. Recently, Fannie started using

"In the Adjustment Example, hereare some of the messages:

"Comps #1, #2, and #3 would allreceive Message 607 that states theappraiser's GLA adjustment is smallerthan peer and model-derived adjust-ments. As this message states, foradjustment messages like this to fire,the appraiser's adjustment must besignificantly different from the major-ity of their peers and be significantlydifferent from the CU model."

"In this case, comps are selling for$200 to $300 per square foot of grossliving area, but the appraiser is adjust-ing only $15 per square foot. Boththeir peers in this market and CU'smodel-derived adjustments are muchhigher than $15 per square foot." - "Last, the appraiser makes a nomi-nal $3,000 adjustment for a secondgarage stall across all comparables.Garages are something you may notpay much attention to, but it still mat-ters to buyers and sellers. This exam-ple property happens to be in a coldweather climate with long harsh win-ters and buyers in an $800,000 pricetier. Many buyers there would noteven consider a home with a singlecar garage. $3,000 equates to lessthan a dollar per day amortized overthe life of a loan. Buyers in this pricetier would certainly expect a muchlarger discount to consider a one-stallgarage."

"Although Net and Gross adjust-ments are very low in this example,they are not reflective of market reac-tion, and the comps may not be assimilar as the appraiser's net andgross adjustments suggest."

My comment: I have included thefull Adjustment Example in theattachment to this newsletter. It is onPage 10 of the Fannie Webinar"Understanding the CU risk score,flags, and messages.

Value Reconciliation "Even when Data Integrity, Comp

Selection, and Adjustments are great,

Appraisal TodayISSN 1066–3900

Appraisal Today is published 12 times per year by Real Estate Communication Resources.

Subscription rate: $99 per year, $169 - 2 yearsPublisher

Ann O'Rourke, MAI, SRA, [email protected]

Subscriber ServicesTheresa Lua

M,T,W 7AM to noonFriday 7AM to 9 AM (Pacific time)[email protected] (24 x 7)

CirculationHancock Mailing Service

Editorial and Subscription Offices2033 Clement Ave., Suite 105

Alameda, CA 94501 Phone: 1-800-839-0227Fax: 1–800–839–0014

Email: [email protected]

Appraisal Today is sold with the understanding that the publisher,editors, and others associated with the publication are not engagedin rendering accounting, legal, or other professional services. Itdoes not attempt to offer specific solutions to individual problems.Questions about specific issues should be referred to the appropri-ate professional for analysis. ©2015 by Real Estate Communication Resources. All rightsreserved. The contents of this publication may not be reproducedeither whole or in part without consent.


If you think your data sources arenot correct, but other appraisers mayuse them, make a comment in yourappraisal. Then you will remember.An underwriter or reviewer may readit.

For example, public records says1,560 sq.ft. GLA. You measured itand that is not correct - it was notmeasured correctly by the assessor, adata entry error, or a permitted addi-tion is not included. Or, you includeda "walk out" basement because this isa standard practice for real estateagents, sellers and buyers in yourarea.

Appraised value conflicts withinyour appraisal

This is easy to avoid and can showappraisers who are trying to get ahigh value. • Above or below the range of theadjusted comps • Above or below the range of the

MBA Loan Volume Application Index – 1/13 to 12/14

1200.0

1400.0

1600.0Market Index

unadjusted comps• Near the maximum or minimumadjusted comparable sale value withsupport from only that comparablesale

How is Fannie estimatingadjustments?

The documents refer to a "model".This is a statistical model, probablyregression-based.

Adjustments - the biggest changeIf questioned, it is good if you can

show how they got their adjustments.If you can't, I assume that somehowyou will be able to change your adjust-ments, hoping they are acceptable.

Adjustments - compare with peersand model• GLA - smaller or larger than peerand model adjustments. This isFannie's current "hot button". Justdon't use the same small adjustment inall your appraisals whatever the medi-an home price or size. • Condition and Quality - smaller orlarger than peer and model adjust-ments

For some reason, Fannie says "mate-rially different" from peer and modeladjustments for the adjustments for:• Lot size • View

• Location

Other adjustment warningmessages• Wide range of adjusted valuesindicates potentially inadequateadjustment. • The adjustment is in the wrongdirection• Negative or postive net adjust-ment indicates the comp is materi-ally superior to the subject, but theappraised value of the subject ismaterially higher

A brief history of appraisalproperty data - why Fannie wantsappraisal data

Before licensing, appraisers couldonly get access to MLS by quarter-ly "sold books" in many MLSs. So,there were various appraiser co-opsaround the country where apprais-ers contributed data from theirappraisals. One of them wasCMDC (California Market DataCo-op) started by S&L appraisersin the 1970s. I was an active mem-ber until licensed appraisers gotMLS access in the early 1990s.There was another large co-opbased in Atlanta, GA.

I was involved in an effort in thelate 1980s to develop data standardsfor appraisers. There were many

mailto:[email protected]





lender, we encourage you to use CUto inform conversations with yourappraisers. Because information inCU is sourced primarily through datasubmitted to UCDP, local appraisersshould have access to much of thissame data – particularly comparablesales and local market trends."

"So you should be able to haveconstructive dialogue with appraisersto resolve appraisal questions or con-cerns. However, you should not makedemands or provide instructions tothe appraiser based solely on auto-mated feedback. In other words, don'tassume they're wrong just becauseyou see a CU message."

"For example, there may be a com-pletely valid reason the appraiserselected the comparables that wereused, or their description of a compa-rable may be correct even though itconflicts with descriptions from otherappraisers."

"Fannie Mae recently reviewed anappraisal in which one appraiser dis-agreed with 23 other appraisers onthe square footage, number of bed-rooms, and number of bathrooms ona comparable. After further review ofthe appraisal and other data sources,we realized that one appraiser got itright and 23 got it wrong. The oddsof this happening are very slim, butwe suspect risk managers and qualitycontrol associates would still be gladthat they took a closer look at thisone."

Where to get more informationIn this article I have read all the

relevant Fannie documents. Theywere all written for lenders, particu-larly underwriters. None were writtenfor appraiser. Some have lots ofinformation for appraisers and somehave little. I have included a fewpages from the Fannie Webinarsattached at the end of this newsletter.

The December 2014 FannieWebinars (Introduction to CollateralUnderwriter and Understanding theCU Risk Score, Flags, and Messages)are worth listening to if you like tohear rather than read. Both are inPDF and can be printed. The intro-duction webinar does not includenotes. The other webinar has somevery useful notes. The print link is inthe Resources tab in the upper rightof the first screen on each webinar.Registration is required for the webi-nars, but you can stop and go backlater.

Link to Fannie Webinars: fan-niemae.com/singlefamily/collateral-underwriter

Bradford forms software had a onehour webinar which covered an intro-duction of big data and CU plusabout ½ hour on their Redstone prod-uct that provides adjustments usingregression. Webinar recording athttps://goto.webcasts.com/viewer/event.jsp?ei=1050667

Fannie will be releasing moreinformation, including underwritertraining classes, in January 2015. Iwill have them listed in my freeAppraisal Today email newsletter assoon as they are available. If youdon't want to subscribe, check theemail archive link onwww.appraisaltoday.com. They aresent out on Thursday, except whenthere is a holiday, such as Christmas.

discussions of "pull down menus" foruse in classifying such characteristicsas type of roof, floor coverings, etc.The choices were controversial andnothing was ever adopted. Plus, thereare wide differences among geo-graphic area in terminology. But, Ilearned a lot about the difficulties ofstandardizing appraisal data.

Recommendations for AMCs andlenders on how to use CU results

"CU is effectively predictive ofovervaluation and appraisal defects,but produces some false positivesand false negatives, just like anyother automated analytical tool. Ahigh CU risk score does not neces-sarily mean an appraisal is bad, nordoes a low score necessarily mean anappraisal is good. The same is truewith individual messages. For thisreason, lenders should not makeappraisal review decisions basedsolely on this automated feedback."

"Lenders may choose to use theCU risk score to triage workflow andallocate resources more effectively.However, Fannie Mae will not pro-vide specific guidance or recommen-dations on where to set any thresh-olds. We recommend that lenderstake time to familiarize themselveswith CU and establish their own risktolerance."

"As we discussed in the score andmessage overview, risk flags identifyfactors contributing to high CU riskscores, and messages highlight spe-cific aspects of the appraisal thatmay require further attention.However, they do not affect theUCDP submission status and areonly warnings. We recommend thatyou consider these messages whileyou perform your review, but youdon't need to "clear" the message."

"Additional data, analytics, andfunctionality in the CU user interfacewill provide users with further con-text and clarity behind the CU scoresand messages. Separate training willbe available on the user interface."

"Last, whether you're an AMC or a

https://goto.webcasts.com/viewer/eve


Many appraisers worry about therisk of getting sued for an

appraisal, but one of your greatestrisks is becoming disabled and unableto work. To appraise at your fullcapacity, you have to be able to walk,hear, and see. If disabled, you may beable to continue working, but atreduced capacity. Or, you may not beable to do field work, but you can dodesktop appraisals and reviews. But,you will probably not be able to workat all for a period of time.

Since appraisers spend a lot of timedriving, getting in an auto accident isa much higher risk than for peopleworking in an office. Other risksinclude getting injured during aninspection, plus the risks we all haveof a serious medical problem.

Jack Jones had a successfulappraisal business, netting him over$75,000 per year. Combined with hiswife's income of $45,000 per year asa dental hygienist, they have a verycomfortable life, with two kids in col-lege and one in high school.

One day, Jack was driving downthe freeway and was rear-ended by alarge truck. He had emergencysurgery for massive injuries. His fam-ily was informed that he will survive,but it will be many months before hecan return to work, even on a part-time basis.

After the initial jubilation that hewill live, his wife starts thinkingabout how they will make it finan-cially. His medical insurance willcover the medical and rehabilitationcosts. All of Jack's net income wasgenerated by his own appraising andreviewing. His overhead costs willcontinue even when he is not working(MLS dues, CE, E&O, etc.). Jack andhis wife have enough in savings tocover 4 months of lost income, butthey don't have any disability insur-ance.

When Jack used to have associateappraisers in his office to doappraisals, he still would have pro-duced about 30% of the income him-self. Plus, he reviewed and co-signedall their work and was the primarycontact for clients.

We all have insurance for autos andfire. Most of us have life insuranceand E&O insurance. But, your great-est risks aren't not being sued over anappraisal or having your house burnup.

The Social Security Administrationestimates that one in four 20-year-olds will become disabled and unableto work before they reach the age of67.

Do It NowDon't wait. Do it now. I had back

surgery in 1988, two years after start-ing my business. I was very fortunate- it was successful and I had a fullrecovery with only 6 weeks off ofwork. I had an office manager andtwo associate appraisers which reallyhelped. Did I have disability insur-ance before the surgery? No. Did Iget it? Yes, but it excluded any backproblems.

An appraiser I have known formany years was diagnosed withadvanced breast cancer 9 years ago atthe age of 62. Her prognosis was notgood and she did not expect to sur-vive very long. She had a 5 year dis-ability policy, up to age 65, for a rela-tively low premium with "own occu-pation" coverage for appraising andreceived $2,000 per month until shewas 65. It was partial income replace-ment but really helped. She survivedand is healthy now. She regularlyencourages appraisers to get disabilityinsurance.

Life vs. disability insuranceShould you have life, disability, or

both?

The reason for life insurance is toprotect your family financially if youdie. As you get older and your chil-dren leave home, this is less impor-tant. I had a disabled spouse and hadlife insurance for many years as hewas unable to work full time. Whenhe passed away, I dropped the insur-ance.

Disability is a much greater riskthan death, but the coverage is formuch lower amounts - monthly pay-ments up to 60% of your income.Self employed persons can't getworkers comp coverage (unless theyare incorporated). Most people arenot disabled for the rest of their lives.This insurance can help you if unableto work for a short or long period.

Disability insurance keeps yougoing while you are recovering. I nolonger have disability insurance as Ireceive $3,000 in Social Security permonth. (I waited until age 70 to startcollecting.)

Risk of becoming disabled forappraisers

There are lots of varying statisticson becoming disabled at some timeduring your working life, varyingfrom 80% to 20%. Much of the dataand analyses come from the disabilityinsurance industry.

But, these numbers are averages.Appraisers work in the field, which ismuch riskier than an office job.

I used www.whatsmypdq.org todetermine Personal DisabilityQuotient(PDQ) for appraisers. This isfrom a non-profit organization,Council for Disability Awareness.

The example I used: Female, Age45, Height 5 ft. 3 inches, weight 125.There were four choices for type ofjob - Mostly Office, Little Office,Little Physical, Mostly Physical. Iselected Little Office. No tobacco,average healthy lifestyle, no treat-


Disability - your greatest risk

http://www.whatsmypdq.org


ments for high blood pressure, etc. Here are the results:

• Your chances of being injured orbecoming ill and unable to workbefore you retire for 3 months orlonger - 22%• If you do become disabled for 3months, your chances of the disabili-ty lasting 5 years or longer - 43%• The average length of a long-termdisability for someone like you is 86months

To calculate the financial risk orEarnable Income Quotient (EIQ), Iused an income of $60,000 per year,annual increase of 2%, and retire-ment at age 65. The answer was$1,457,842.

Results for selecting Little Physicalfor type of job the probability ofbeing disabled for 3 months or longerincreased to 28%. The other numberswere the same.

I strongly recommend doing thisanalysis for yourself. It only takes afew minutes and is very easy to do.

Social SecurityYou have Social Security coverage,

if you have enough quarters in. But,it typically takes over a year to getstarted, and the benefits are relativelylow compared with current earnings.

Also, you must be unable to per-form any gainful activity and the dis-ability must be expected to last atleast 12 months or result in death.

Where to get disability insuranceDisability insurance policies are

available from many insurance bro-kers and companies. Check with yourpersonal insurance broker. LiabilityInsurance Administrators, who sellsappraisers E&O insurance offers it atwww.liability.com. The AppraisalInstitute offers a group disability andprofessional overhead coverage to itsmembers through REAGIT. NAIFAalso has a disability group plan.Check to see if any of your groupsoffer it.

Please note that some people say

that it can be difficult to collect froman insurance company and it maytake awhile. Be sure to google com-panies you are interested in to see ifthere are any problems getting paid.

Private disability insuranceDisability insurance can only be

purchased for partial, not full,replacement of income, typically60%. The insurance company wantsto be sure you are financially moti-vated to return to work. However,disability income from a private poli-cy is not taxable, so it's probablyclose to your after-tax income.

What to look for in a policy:1. Coverage for "own occupation".2. Both non-cancellable and guaran-teed renewable.3. Optional inclusion of partial dis-ability.4. Optional cost of living adjust-ments.

Primary cost factors:1. The dollar amount of income to bereceived. (The higher the income, thehigher the premiums.)2. Length of waiting time before ben-efits start. (The shorter the wait, thehigher the premiums.)3. Duration of the benefits. (Thelonger the duration, the higher thepremium.)

For an appraiser, you should getcoverage for "own occupation", orcoverage if you are unable toappraise. Many companies try todeny insurance by saying you mustbe able to do "any gainful occupa-tion".

Your policy should be both non-cancellable and guaranteed renew-able.

You may want to include coveragefor partial disability: unable to per-form one or more parts of a job.

A cost-of-living adjustment ensuresthat your benefit goes up over time.This is particularly important for anyvery long-term disability.

Disability coverage for you, yourspouse, or both?

How much income do you want toreplace if disabled? If you or yourspouse provide most of your familyincome, you will probably want toinsure only the higher-earning person.

If you want as much protection aspossible, insure yourself and yourspouse.

Long term individual coverage costs You can expect to pay between 1%

- 3% of your annual gross income fora quality policy.

Example: If you are earning$50,000 per year you can expect topay between $500 - $1500 per yeardepending on your occupation, ageand the level of benefits and optionalriders that you include on your poli-cy.

Quotes vary widely, so shoparound. All the options make it con-fusing sometimes.

Here's a sample quote for "ownoccupation": The cost varies withage, gender and occupation. Ahealthy, 35-year-old man who earns$75,000 per year in a white-collarprofession would pay about $100 amonth to The Guardian for a policywith a $4,000 monthly income bene-fit that continues to age 65 if he's per-manently injured. If he buys it at 40,the same insurance costs about $120a month.

Sample quotes are below for: pay-ment to age 65, $3,000 per monthincome, 90 day waiting period, totaland partial disability.

As you get older, your risk increas-es. Below are sample comparativerates.Partial disability:Age 35: $1,200 per year ($40 per$100 of coverage)Age 45: $1,800 per year ($60 per$100 of coverage)Age 55: $2,550 per year ($85 per$100 of coverage)

http://www.liability.com

Only full disabilityAge 35: $1,020 per year ($34 per$100 of coverage)Age 45: $1,560 per year ($52 per$100 of coverage)Age 55: $2,220 per year ($74 per$100 of coverage)• Cost of Living adjustment increasespayments by about 20%• Increasing the waiting period from90 to 180 days would decrease thepayment by about 10%. Reducing thewaiting period further, from 60 to 30days would increase the payment anadditional 20%. • A lifetime benefit would increase thepayment by about 20%. However,there are many variables, such aswhen the disability started andwhether it was due to accident orsickness.

Short term disability coverage -another option

Most disability lasts less than oneyear. Short term insurance is muchless costly than long term.

New businesses and part-timersIt is very difficult for new business

owners to obtain disability insuranceas they have no income stream.

Part timers who work less than 26to 30 hours per week typically can'tget coverage as insurers figure there'sno incentive to return to work.

Professional overhead insuranceAnother cost-effective option is

professional overhead insurance,which is short term, typically one totwo years, and covers actual businessoverhead expenses such as officerent, support staff salaries, and insur-ance. You must continue to be inbusiness after becoming disabled.

For a 45-year-old, with a 30-daywaiting time, and a 12-month period,a typical premium is $27 per year per$100 of coverage. For example, a$3,000 per month coverage wouldcost $810 per year. A 24-month bene-fit would cost $34 per $100, or$1,020 per year.

Group insuranceCheck with organizations you are a

member of to see if they offer agroup disability insurance plan,which may be cheaper than a person-al plan. The Appraisal Institute hasdisability and professional overheadgroup insurance. NAIFA also has agroup plan.

Key person disabilityWhat if your partner or a key

employee became disabled? Thisinsurance reimburses the business forthe loss of a key employee andallows funding of temporary replace-ment or training of a successor.

Some sample stats- One in five Americans aged 35 andolder will experience a long-term dis-ability — one that lasts three monthsor longer — before their 65th birth-day - you are three times more likely tobecome disabled beforeage 65 than you are to die. - 23,000 people are killed in acci-dents at home each year....350,000will be disabled. - 13,000 people are killed in work-related accidents eachyear....2,200,000 will be disabled. - 21,000 people are killed in publicaccidents each year....2,700,000 willbe disabled.

- 51,000 people are killed in vehicu-lar accidents each year....2,000,000will be disabled.

Where to get more informationThere is a good article from the

New York Times athttp://www.nytimes.com/2010/02/06/your-money/life-and-disability-insur-ance/06money.html explaining all theconflicting information, dealing com-panies that sell disability insurance,and links to other web sites.

Most of the companies offeringdisability insurance are large lifeinsurance companies such asNorthwestern, Equitable, andPrudential. Using a broker experi-enced in disability policies is a goodoption. Be sure to google the insur-ance company to see if there are anyproblems getting paid.

Check with your appraisal associa-tion or other business group to see ifthey have a group plan for personaldisability or professional overheadinsurance.


http://www.nytimes.com/2010/02/06/


On the next 12 pages are excerpts from The December 2014 Fannie Webinars -Introduction to Collateral Underwriter and Understanding the CU Risk Score, Flags, andMessages

The first 6 pages are from the Understanding the CU Risk Score, Flags, and Messages - good examples and com-ments and include pages 9 to 14 of the webinar:Page 9 - Comp Selection Example. - maps of which comps CU has selected plus comments below.Page 10 - Adjustments - Comments on 4 adjustments - Physical simlarity, Time, Location, Sale type. Page 11 - Adjustment Example - Grid with comps plus comments belowPage 12 - Reconciliation - What CU looks forPage 13 - Reconciliation Example - Grid with comps plus comments belowPage 14 - CU Data Quality and Model Error Messages - a few examples, many are probably typos

The next 6 pages are very different. They are from the Introduction to Collateral Underwriterwebinar and are the pages that underwriters see. AMCs and appraisers do not see these pages.Page 1 - Appraisal Page - Comp map, Messages, and comp gridPage 2- Dataappraisal - List of appraiser and Model selected comps. Sortable columnsPage 3 - Configurable Comp Searches - the underwriter uses this page to change comp selectionPage 4 - Messages - ExamplesPage 5 - Aerial and Street-view Photogaphy - What the underwriter has access toPage 6 - Local Market Trends - Graph and “Heat Map” - where price changes are located and comp locations

The last 6 pages are from UCDP Fannie Mae Appraisal Messaging Change Notification - November 18, 2014 andincludes all the Warning Messages and the Hard Stops. Please note that Warning Messages do not require fixingthem. The Underwriter decides which ones to send to the AMC or lender. Hard Stops require a correction on theappraisal to get the appraisal accepted by UCDP.

Here are location maps for two appraisals with one mile radius circles drawn around the subject properties. These are from real appraisals of two different properties submitted through UCDP. The maps were generated in CU.

You will see the subject property in green and the appraiser-provided comparables in blue. The dotted black circle represents a one-mile radius around the subject. By the way, Fannie Mae does not have a 1-mile guideline, but that is a common rule of thumb and we’ll use it here for the sake of illustration.

Note the distances to the comps. For the appraisal on the left, the comps are three to six miles from the subject. For the appraisal on the right, all the comps are less than a mile from that subject.

Both of these appraisals have commentary from the appraiser stating that the comps they chose were the best available.

Now, which of these is most troubling to you? Many of you may say the appraisal on the left, due to the distance of the comparables.

Now let’s look at alternative comparables found by CU.

On the appraisal on the left, you’ll see a few other sales noted by the orange icons. We see that all the alternative comps for the property on the left are at similar distances, several miles from the subject. It turns out that this subject is an older, small home on a large lot that is surrounded by newer, larger homes on much smaller lots.

On the appraisal on the right, you’ll see more than a dozen sales within a few blocks that were ignored by the appraiser. It just so happens that these homes are, by all accounts, very similar to the subject and selling for far below the appraised value.

In summary, you could easily get turned around if you’re using generic guidelines and a checklist approach to comp selection. You may let the appraisal shown on the right slide by with no further action taken, while you engage the appraiser on the left with requests for comparables in closer proximity when none exist.

Using Collateral Underwriter, you may act differently. The appraisal on the left received a low CU risk score, while the appraisal on the right received a comparable selection message and a high CU risk score.

So when we say CU is “model based” and doesn’t rely on arbitrary guidelines, this is a good illustration.

9

Administrator

Sticky Note

The next step in the process is adjustments. Collateral Underwriter produces statistically-derived market-specific adjustments for all UAD-standardized physical characteristics, date of sale, location, and sales type.

For physical similarity, CU uses regression analysis to produce adjustments for property features. Different models are used for single family and condo property types. The magnitude of the adjustment for each physical feature may vary from market to market.

CU also makes time adjustments to comparables. These adjustments may be negative or positive. The frequency of time adjustments from appraisers is startlingly low, particularly over the past few years when we have seen rapid home price appreciation in many markets. In fact, looking at some of the hottest markets, we saw time adjustments on only five percent of comps over 6 months old. There is a common misconception that Fannie Mae will not accept positive time adjustments – in fact, Fannie Mae is looking for an accurate market value, and accepts both positive and negative time adjustments as appropriate.

Differences in location are also adjusted for. In absence of clear neighborhood definitions, CU uses Census Block Groups, or “CBGs” as defined by the US Census Bureau, and make adjustments based on sale prices within these areas.

CU also makes market-based adjustments for Sales Type. REO sales, Short Sales, Relocation sales and non-arms length transactions are adjusted against arms-length transactions based on observed discounts and premiums between these sales types. This too, varies from market to market.

The key point here is that CU does not adhere to the widely recognized – and often misinterpreted – 15% net and 25% gross adjustment guidelines. Many believe that Fannie Mae will not accept comps with adjustments in excess of these guidelines, which is also not true. Long-standing reliance on generic guidelines has led to what Fannie Mae believes is widespread under-adjustment by appraisers. As stated in our Selling Guide, adjustments may be an indicator of comparability, but lower adjustments are not necessarily more accurate adjustments.

In summary, rather than take a “lower is better” approach to the adjustments, CU will flag adjustments that are in the wrong direction or are significantly different than the appraiser’s peers and the adjustments derived by our statistical models. In these cases, lenders should determine if the appraiser’s adjustments are adequately supported and reflective of market reaction. Let’s look at an example.

10

In this example, the appraiser has provided four of the best comparables available. However, most are materially superior to the subject, and CU has issued a series of messages for suspected under-adjustment.

Here are a few of the potential issues and messages that would fire on this appraisal …

Comps #1, #2, and #3 would all receive Message 607 that states the appraiser’s GLA adjustment is smaller than peer and model-derived adjustments. As this message states, for adjustment messages like this to fire, the appraiser’s adjustment must be significantly different from the majority of their peers and be significantly different from the CU model.

In this case, comps are selling for $200 to $300 per square foot of gross living area, but the appraiser is adjusting only $15 per square foot. Both their peers in this market and CU’s model-derived adjustments are much higher than $15 per square foot.

On Comp #2, we see Message 611 that states the adjustments for lot size are materially different from the peer and model adjustments.

In the appraisal report, the appraiser estimates the site value of the subject to be nearly half a million dollars. However, they make absolutely no adjustment for the fact that Comp #2 has a double lot. The appraiser’s peers in this market and the CU model indicate that an adjustment would be warranted for a lot size difference of this magnitude.

Last, Message 640 fires on all four comparables. This message states that the appraiser’s net adjustments are materially different from the model’s net adjustments. In addition to those issues already noted, there are other factors contributing to these messages.

Nominal adjustments for the extra bathroom on Comp #1 and Comp #3

Adjustments of only $2 to $5 per square foot for basement finish on Comps #1, #3, and #4. To put this in perspective, the appraiser is adjusting less than 1% for a fully finished basement on Comp #4.

Last, the appraiser makes a nominal $3,000 adjustment for a second garage stall across all comparables. Garages are something you may not pay much attention to, but it still matters to buyers and sellers. This example property happens to be in a cold weather climate with long harsh winters and buyers in an $800,000 price tier. Many buyers there would not even consider a home with a single car garage. $3,000 equates to less than a dollar per day amortized over the life of a loan. Buyers in this price tier would certainly expect a much larger discount to consider a one-stall garage.

Although Net and Gross adjustments are very low in this example, they are not reflective of market reaction, and the comps may not be as similar as the appraiser’s net and gross adjustments suggest.

In summary, utilization of superior comps with inadequate adjustments can lead to material overvaluation.

11

The last component CU analyzes is value reconciliation. This is a fairly straight-forward examination of the relationship between the appraised value and the adjusted and unadjusted range of comparable sales prices.

CU looks for the following:

Appraised values far outside the range of unadjusted comp prices. Bracketing is not a requirement, but it is often worth a closer look at the appraisal when the subject is worth far more or far less than any of the comparables.

CU also looks for appraised values outside the range of adjusted comp values. So for example, an appraised value of $250,000 when the adjusted range of comps spans $200,000 to $225,000.

If you think this sounds silly, know that Fannie Mae is on pace to receive more than 20,000 appraisals from lenders this year that have a value outside the range of adjusted comp values. Somewhere in your next 100 to 150 appraisals, you will probably see this.

Last, CU looks for cases with wide ranges of value and support from a single comparable. In layman’s terms, this is “putting all of your eggs in one basket”. Again, this may be justified, but is usually worth a look in order to validate.

The key point here is that CU looks at values at both the upper and lower ends of the range of comp values. Overvaluation and undervaluation may lead to higher CU risk scores and appraisal quality flags.

Appraisals with any of these characteristics will receive reconciliation messages and higher CU risk scores. In these cases, lenders should confirm that the appraiser has given the most weight to the most relevant comparables in their reconciliation of value.

Here is an example…

12

Here is another appraisal on a loan delivered to Fannie Mae.

Even though the comparables are among the best available, Collateral Underwriter generates a reconciliation message on this appraisal. Message 505 states that the appraised value is near the maximum adjusted comparable sale value with support from only that comparable sale. Simply put, the appraiser is putting all of their eggs in one basket.

Let’s look at some of the factors contributing to this message.

The first thing we notice is a very wide range of sale prices, from $510,000 to $720,000. Even after adjustments, the range still extends from under $550,000 to $720,000.

Comps #1 and #2 appear quite similar to the subject and are located on the same street. They sold for only $510,000 and $550,000 respectively.

Comp #3, however, sold for $720,000. So one of these things is not like the other. Why may that be? Closer examination of this comp shows it has the largest GLA, nearly three times the lot size, and is farthest away from the subject. Research suggests that property values in this neighborhood are dramatically higher than in the subject neighborhood. So not only is it physically the least similar to the subject, but it is located in a higher-priced neighborhood.

Nonetheless, the appraiser gives nearly all weight to this comp in reconciliation. So while this is “bracketed”, the appraiser is giving most weight to the least relevant comparable, and as a result is inflating value.

13

So we just covered the four key appraisal components and associated messaging. Those account for the majority of the new Fannie Mae messages and are the heart of Collateral Underwriter.

There are two other series of messages: one to highlight potential data quality issues, and the 999s that provide reasons for unscored appraisals

The first group is the CU Data Quality messages. These highlight suspected data entry errors.

Data Quality Messages include the following:

Invalid property addresses, dates in the future or distant past, or failure to report data in UAD format.

Implausible sales prices, appraised values, or property features. Keep in mind we’re talking about extreme cases here that are most likely the result of typos or truncated XML data.

Other messages highlight appraisals with less than three closed sales, multiple uses of the same comp within a single report, or net/gross adjustments that don’t match the sum of individual line item adjustments.

These Data Quality Messages are very infrequent and can usually be resolved by the appraiser.

The second set of messages – called Model Error messages, or 999s – explain why an appraisal was not scored.

Examples of Model Error Messages include:

Multiple data errors

Unsuccessful geocoding of the subject property or a number of comparable sales

Subject attributes that cannot be modeled, or areas with extremely limited sales activity.

Geocoding errors are the most common cause of unscored appraisals. It is extremely rare that subject attributes or location cannot be modeled.

Lenders should review these messages carefully to determine if the errors can be corrected by the appraiser or if they are unable to be resolved.

For example, a home that is actually 20,000 square feet might simply be out of scope for our model. However, if that was a typo and the appraiser meant 2,000 square feet, they can correct and resubmit.

It may be possible for CU to eventually geocode the subject property if the appraiser corrects an invalid address. However, if CU can’t geocode because the property is brand new construction with an unrecognized address, there may be no possibility of resolution.

14

9 Collateral Underwriter™ by Fannie Mae Collateral Underwriter I October 2014

Appraisal Page

The main appraisal page includes a comparable sales map, messaging center, details for the appraiser-provided comparables, and links to additional information and functionality.

Comparable Sales Map

Message Center

Appraiser-provided comparables

Subject and comp characteristics

Comparable Rank

CU Score and Risk Flags

Datappraisal


Datappraisal

The “Datappraisal” displays the results of CU’s comparable selection model. Appraiser-provided comparables are ranked along up to 20 model-selected sales in the subject market.

Appraiser-provided and model-selected comps

Users can select/deselect model comparables

Sortable Column Headers


Configurable Comp Searches

Users can perform comparable searches by defining specific parameters for geographic boundaries, time frames, and physical characteristics.

Users may also set specific parameters for comp physical characteristics and time frame.

Users can draw specific geographic areas from which to choose comps.


Messages

The Message center includes the Collateral Underwriter score, flags and messages in addition to UAD/UCDP edits and remaining Fannie Mae Proprietary Messages.

UAD/UCDP® Edits and FNM Proprietary Messages

CU Risk Score and Flags

Detailed data integrity, comp selection, adjustment, and

reconciliation messages


Aerial and Street-view Photography

Users can easily access aerial and street-view photography from the appraisal page with a simple point-and-click of the subject or any comparable on Collateral Underwriter’s property map.

CU provides a birds-eye view of the subject and comparables. Users can easily move between

properties, rotate camera angle, zoom-in/zoom-out, etc.

Street view imagery for the subject or any comparable can be accessed through the CU interface.


Local Market Trends

Collateral Underwriter’s Market Trend and Heat Map functionality provides users detailed insight into local market trends.

The Market Trend function shows market appreciation or decline relative to prior sales of the subject property.

Heat Maps display Census Block Group-level statistics such as median sales price, price/GLA, median days on market, etc.

© 2014 Fannie Mae. Trademarks of Fannie Mae. 11.18.2014 4

Appendix A – Collateral Underwriter Appraisal Messages

Message ID Message Text Severity Applicable

Forms

FNM0401 The appraiser has reported materially different sale price(s) in one or more appraisal reports.

Warning 1004/1073

FNM0402 The reported comparable sale price is materially different than what has been reported by other appraisers.

Warning 1004/1073

FNM0403 The appraiser has reported materially different GLA(s) in one or more appraisal reports.

Warning 1004/1073

FNM0404 The reported GLA is materially different than what has been reported by other appraisers.

Warning 1004/1073

FNM0405 The appraiser has reported materially different lot size(s) in one or more appraisal reports.

Warning 1004/1073

FNM0406 The reported lot size is materially different than what has been reported by other appraisers.

Warning 1004/1073

FNM0407 The appraiser has reported materially different above-grade bathroom count(s) in one or more appraisal reports.

Warning 1004/1073

FNM0408 The reported above-grade bathroom count is materially different than what has been reported by other appraisers.

Warning 1004/1073

FNM0409 The appraiser has reported materially different above-grade bedroom count(s) in one or more appraisal reports.

Warning 1004/1073

FNM0410 The reported above-grade bedroom count is materially different than what has been reported by other appraisers.

Warning 1004/1073

FNM0411 The appraiser has reported materially different age(s) in one or more appraisal reports.

Warning 1004/1073

FNM0412 The reported property age is materially different than what has been reported by other appraisers.

Warning 1004/1073

FNM0413 The appraiser has reported materially different below-grade area(s) in one or more appraisal reports.

Warning 1004/1073

FNM0414 The reported total below-grade area is materially different than what has been reported by other appraisers.

Warning 1004/1073

FNM0415 The appraiser has reported materially different finished basement area(s) in one or more appraisal reports.

Warning 1004/1073

FNM0416 The reported finished basement area is materially different than what has been reported by other appraisers.

Warning 1004/1073

FNM0417 The appraiser has reported a materially different location in one or more appraisal reports.

Warning 1004/1073

FNM0418 The location rating is materially different than what has been reported by other appraisers.

Warning 1004/1073

FNM0419 The appraiser has reported materially different view(s) in one or more appraisal reports.

Warning 1004/1073

FNM0420 The view rating is materially different than what has been reported by other appraisers.

Warning 1004/1073



Forms

FNM0421 The appraiser has reported a materially different quality in one or more appraisal reports.

Warning 1004/1073

FNM0422 The quality rating is materially different than what has been reported by other appraisers.

Warning 1004/1073

FNM0423 The appraiser has reported a materially different condition in one or more appraisal reports.

Warning 1004/1073

FNM0424 The condition rating is materially different than what has been reported by other appraisers.

Warning 1004/1073

FNM0428 The reported finished basement area is materially different than what has been reported by other appraisers, but the total living area is consistent.

Warning 1004/1073

FNM0429 The appraiser has reported a materially different finished basement area in one or more appraisal reports, but the total living area is consistent.

Warning 1004/1073

FNM0430 The sale price is significantly different than what has been reported by any other appraiser.

Warning 1004/1073

FNM0431 The GLA is significantly different than what has been reported by any other appraiser.

Warning 1004/1073

FNM0432 The lot size is significantly different than what has been reported by any other appraiser.

Warning 1004/1073

FNM0433 The condition rating is significantly different than what has been reported by any other appraiser.

Warning 1004/1073

FNM0434 The quality rating is significantly different than what has been reported by any other appraiser.

Warning 1004/1073

FNM0435 The location rating is significantly different than what has been reported by any other appraiser.

Warning 1004/1073

FNM0436 The view rating is significantly different than what has been reported by any other appraiser.

Warning 1004/1073

FNM0437 The reported total living area for the subject is materially different than what has been reported in another appraisal of the same subject.

Warning 1004/1073

FNM0438 The reported total bedroom count for the subject is materially different than what has been reported in another appraisal of the same subject.

Warning 1004/1073

FNM0439 The reported total bathroom count for the subject is materially different than what has been reported in another appraisal of the same subject.

Warning 1004/1073

FNM0440 The reported lot size for the subject is materially different than what has been reported in another appraisal of the same subject.

Warning 1004/1073

FNM0441 The reported year built for the subject is materially different than what has been reported in another appraisal of the same subject.

Warning 1004/1073

FNM0442 The location rating for the subject is materially different than what has been reported in another appraisal of the same subject.

Warning 1004/1073

FNM0443 The view rating for the subject is materially different than what has been reported in another appraisal of the same subject.

Warning 1004/1073



Forms

FNM0444 The quality rating for the subject is materially different than what has been reported in another appraisal of the same subject.

Warning 1004/1073

FNM0445 The condition rating for the subject is materially different than what has been reported in another appraisal of the same subject.

Warning 1004/1073

FNM0446 The subject was excluded from data discrepancy computations due to a missing unit number.

Warning 1004/1073

FNM0483 The condition rating of "C1" conflicts with the reported age. Warning 1004/1073





FNM0488 The condition rating conflicts with the reported age (condition rating is likely "C1").

Warning 1004/1073

FNM0489 The condition rating conflicts with the reported age (condition rating is likely either "C1" or "C2").

Warning 1004/1073

FNM0490 The condition rating conflicts with the reported age (condition rating is likely "C1").

Warning 1004/1073

FNM0491 The condition rating conflicts with the reported age (condition rating is likely either "C1" or "C2").

Warning 1004/1073

FNM0501 The appraised value is above the range of adjusted comparable sale prices provided by the appraiser.

Warning 1004/1073

FNM0502 The appraised value is below the range of adjusted comparable sale prices provided by the appraiser.

Warning 1004/1073

FNM0503 The appraised value is above the range of unadjusted comparable sale prices provided by the appraiser.

Warning 1004/1073

FNM0504 The appraised value is below the range of unadjusted comparable sale prices provided by the appraiser.

Warning 1004/1073

FNM0505 The appraised value is near the maximum adjusted comparable sale value with support from only that comparable sale.

Warning 1004/1073

FNM0506 The appraised value is near the minimum adjusted comparable sale value with support from only that comparable sale.

Warning 1004/1073

FNM0603 <Feature(s)> adjustment(s) is (are) in the wrong direction. Warning 1004/1073

FNM0606 The GLA adjustment is larger than peer and model adjustments.

Warning 1004/1073

FNM0607 The GLA adjustment is smaller than peer and model adjustments.

Warning 1004/1073



Forms

FNM0610 The appraiser's wide range of adjusted values indicates potentially inadequate adjustment.

Warning 1004/1073

FNM0611 The lot size adjustment is materially different from peer and model adjustments.

Warning 1004/1073

FNM0612 The lot size adjustment is materially different from peer and model adjustments.

Warning 1004/1073

FNM0613 The condition adjustment is larger than peer and model adjustments.

Warning 1004/1073

FNM0614 The condition adjustment is smaller than peer and model adjustments.

Warning 1004/1073

FNM0615 The quality adjustment is larger than peer and model adjustments.

Warning 1004/1073

FNM0616 The quality adjustment is smaller than peer and model adjustments.

Warning 1004/1073

FNM0617 The view adjustment is materially different from peer and model adjustments.

Warning 1004/1073

FNM0618 The view adjustment is materially different from peer and model adjustments.

Warning 1004/1073

FNM0619 The location adjustment is materially different from peer and model adjustments.

Warning 1004/1073

FNM0620 The location adjustment is materially different from peer and model adjustments.

Warning 1004/1073

FNM0630 The appraiser's negative net adjustment indicates the comparable is materially superior to the subject, but the appraised value of the subject is materially higher.

Warning 1004/1073

FNM0631 The appraiser's positive net adjustment indicates the comparable is materially inferior to the subject, but the appraised value is materially lower.

Warning 1004/1073

FNM0640 The appraiser's net adjustments for the comparable sales are materially different from the model net adjustments.

Warning 1004/1073

FNM0797 The appraiser-provided comparables are materially different than the model-selected comparables.

Warning 1004/1073

FNM0801 The address is the same as the subject or one of the other comparables.

Warning 1004/1073

FNM0802 The raw USPS ZIP code data is invalid. Warning 1004/1073

FNM0803 The property address cannot be geocoded. Warning 1004/1073

FNM0804 The reported property value is outside the typical range ($10,000 - $10,000,000) and may be erroneous.

Warning 1004/1073

FNM0805 The reported sale date is invalid. Warning 1004/1073

FNM0806 The reported property age may be erroneous. Warning 1004/1073



Forms

FNM0807 The reported GLA may be erroneous. Warning 1004/1073

FNM0808 The reported above-grade full bathroom count may be erroneous.

Warning 1004/1073

FNM0809 The reported above-grade half bathroom count may be erroneous.

Warning 1004/1073

FNM0810 The reported above-grade bedroom count may be erroneous.

Warning 1004/1073

FNM0811 The reported lot size may be erroneous. Warning 1004/1073

FNM0812 The reported total below-grade area may be erroneous. Warning 1004/1073

FNM0813 The reported finished basement area exceeds the reported total below-grade area.

Warning 1004/1073

FNM0814 The appraisal has fewer than 3 closed sale comps. Warning 1004/1073

FNM0815 The appraisal's effective date is invalid. Warning 1004/1073

FNM0816 The reported condition rating is outside of the model range (1-5).

Warning 1004/1073

FNM0817 The reported quality rating is outside of the model range (1-5).

Warning 1004/1073

FNM0818 The reported location rating is invalid. Warning 1004/1073

FNM0819 The reported view rating is invalid. Warning 1004/1073

FNM0820 The appraised value is suspected to have a data entry error. Warning 1004/1073

FNM0821 The unadjusted sale price is significantly different than the appraised value.

Warning 1004/1073

FNM0822 The net adjustment is erroneous. Warning 1004/1073

FNM0901 The appraisal report contains multiple data errors. No CU Risk Score available.

Warning 1004/1073

FNM0902 A subject attribute is not covered by the model. No CU Risk Score available.

Warning 1004/1073

FNM0903 The subject address cannot be geocoded. No CU Risk Score available.

Warning 1004/1073

FNM0904 The subject's location is not covered by the model. No CU Risk Score available.

Warning 1004/1073

FNM0906 Fewer than 3 appraiser-provided comparables can be geocoded. No CU Risk Score available.

Warning 1004/1073



Forms

FNM0908 Due to multiple data problems, an insufficient number of appraisal comparable sales exists for risk assessment. No Collateral Underwriter Risk Score available.

Warning 1004/1073

FNM0999 An unknown error has occurred. No Collateral Underwriter Risk Score available.

Warning 1004/1073

FNM1000

The Collateral Underwriter Risk Score is <#> on a scale of 1 to 5 where 5 indicates highest potential collateral risk. A score of 999 indicates no Collateral Underwriter Risk Score available.

Warning 1004/1073

FNM1002 There is a heightened risk of appraisal quality issues. Warning 1004/1073

FNM1004 There is a heightened risk of overvaluation. Warning 1004/1073

FNM1006 There is a heightened risk of Property Eligibility and/or Policy Compliance violations on this appraisal.

Warning 1004/1073

FNM1093 The CU Analysis could not be performed. Warning 1004/1073

FNM1095 The CU analysis could not be performed. Only appraisals submitted on form 1004 or 1073 in UAD format are analyzed by CU.

Warning 1004/1073

FNM1096 The CU Analysis could not be performed. Warning 1004/1073


Appendix B – Fannie Mae Overridable Appraisal Messages


Forms

FNM0083 The sales contract was not analyzed. Overridable 1004/2055, 1073/1075

FNM0084 There was no comment on market conditions, even though one or more negative housing trends were indicated (declining, over supply, over 6 months).

Overridable 1004/2055, 1073/1075

FNM0085 Less than three settled sales were used as comparables. Overridable 1004/2055, 1073/1075

FNM0086 Research of prior sale was not performed. Overridable 1004/2055 FNM0087 Research of prior sale was not performed. Overridable 1073/1075 FNM0092 State certificate is not provided on transaction amount over

$1 million. Overridable 1004/2055, 1073/1075

FNM0093 Appraiser license state does not match subject property state. Overridable

1004/2055, 1073/1075

FNM0094 Supervisor license state does not match subject property state. Overridable

1004/2055, 1073/1075

FNM0096 Illegal zoning compliance has been indicated in appraisal. Review description to verify if the property may be eligible per the Selling Guide.

Overridable 1004/2055

FNM0097 Illegal zoning compliance has been indicated in appraisal. Review description to verify if the property may be eligible per the Selling Guide.


FNM0098 Present use is indicated as not highest and best use. Overridable 1004/2055 FNM0099 Present use is indicated as not highest and best use. Overridable 1073/1075

FNM0101 The subject property may be a hotel/motel or condo hotel. Overridable 1004/2055, 1073/1075

FNM0102 The subject property is in a condominium project that may be ineligible for delivery to Fannie Mae.

Overridable 1004/2055, 1073/1075

FNM0174 The project name suggests that the property may be a condo hotel. Verify the subject is located in a project that meets Fannie Mae’s Selling Guide requirements.


FNM0176

The appraisal indicates that the subject property has legal nonconforming zoning and cannot be rebuilt to the current density. This data indicates that the property is ineligible for delivery to Fannie Mae.


FNM0179 The appraisal indicates the subject property has a C6 condition rating. If the loan is not a DU Refi Plus or Refi Plus loan, the property is not eligible for delivery to Fannie Mae.

Overridable 1004/2055, 1073/1075

FNM0193 Based on the overall quality of this appraiser's work, Fannie Mae has decided to review 100% of the appraisals prepared by this appraiser for any loans delivered to Fannie Mae.

Overridable 1004/2055, 1073/1075

FNM0194

Based on the overall quality of this supervisory appraiser's work, Fannie Mae has decided to review 100% of the appraisals prepared by this appraiser for any loans delivered to Fannie Mae.

Overridable 1004/2055, 1073/1075

FNM0195 Fannie Mae will not accept appraisals from this appraiser. Overridable 1004/2055, 1073/1075

FNM0196 Fannie Mae will not accept appraisals from this supervisory appraiser.

Overridable 1004/2055, 1073/1075

fannie's collateral underwriter - big changes and what it ...today+1... · fannie's...

Documents