Freddie Mac: Quality & Condition Ratings

[NC Appraising]

Supposedly there is software out there which can interpret pics to derive quality and condition ratings. A consumer version of that kind of app might help appraisers calibrate their own judgement and result in more consistency.

"Hey Siri, can you check the pics in this folder and develop overall quality and condition ratings? While you're at it, print a report with your analysis for my workfile." ​

+-14 minutes in and other parts as well.

I would suggest Every appraiser to watch this. THIS is the future of appraising for the GSEs.

 

[TerryRohrer]

Certainly there is factual information in our appraisals but valuations by definition are opinions so they are, and always will be, subjective. Another issue I have is that I think our estimated values should be listed as a range of values and tied to a confidence rating because in most cases we aren't good enough to say the property is only worth X amount, no more no less.

I have never seen a comment or an appraisal where an appraiser said, certified, or implied that their opinion of value was "X, no more no less." Nor have I ever understood some appraiser's desire to claim publicly and often that they are not "good enough" to conclude and convey and defend "their opinion" of the value of a property.

But those issues aside, how is the "confidence rating" of a range in values being proposed developed and presented and supported and defended? It seems to me if one can't develop an opinion of value they can believe in themselves, they won't have much more luck in convincing someone an even more deficient measure of "credibility" will add much. I expect it would serve to mislead more than to actually improve anything in the appraisal process.

 

[alebrewer]

Nor have I ever understood some appraiser's desire to claim publicly and often that they are not "good enough" to conclude and convey and defend "their opinion" of the value or a property.

Fully agreed that appraisers should be ready and able to defend their opinions of value. That said, its my opinion that point values don't exist. We report them all the time, but in reality - there's no such thing. Again, IMO.

 

[TerryRohrer]

Fully agreed that appraisers should be ready and able to defend their opinions of value. That said, its my opinion that point values don't exist. We report them all the time, but in reality - there's no such thing. Again, IMO.

I don't disagree. Market values don't exist, either.

 

[FLApr04]

I have never seen a comment or an appraisal where an appraiser said, certified, or implied that their opinion of value was "X, no more no less." Nor have I ever understood some appraiser's desire to claim publicly and often that they are not "good enough" to conclude and convey and defend "their opinion" of the value or a property.

But those issues aside, how is the "confidence rating" of a range in values being proposed developed and presented and supported and defended? It seems to me if one can't develop an opinion of value they can believe in themselves, they won't have much more luck in convincing someone an even more deficient measure of "credibility" will add much. I expect it would serve to mislead more than to actually improve anything in the appraisal process.

The value of the appraisal is implied to be the exact value is it not? Banks don't use some value close to our value. No, they use our exact value to base their lending decisions. If appraising was so exact then if you gave 10 different appraisers the same assignment they should all come back the same exact valuation right? We all know that doesn't happen. In fact, you would mostly likely get 10 different values and the more difficult the assignment, the wider range of values you're likely to get.

This is why I think a range of values with a certain confidence level is a better, and more honest way to value a property. That doesn't mean appraisers should have some crazy wide range of values on every assignment but it should be within a small range. Some values could even be exact like when you're in a PUD and you have model match homes with no variations but it would be more accurate to have a range on more difficult properties. When I said we "good enough" I meant it's just our opinion and yes that can be supported but so can another appraiser who comes up with a different value who's value is also supported. You see statistics and data can be viewed and used in many different ways supporting different conclusions so if they are both supported who is to say one is right and one is wrong? Are you saying you're opinion is absolutely right every time and anyone else who comes to a different conclusion is wrong?

 

[J Grant]

I completely agree and have been saying the same thing to my colleagues for years. I like how they try to make the condition fit into some neat little box as if that's how the real world works so there has to be room for variations. Certainly there is factual information in our appraisals but valuations by definition are opinions so they are, and always will be, subjective. Another issue I have is that I think our estimated values should be listed as a range of values and tied to a confidence rating because in most cases we aren't good enough to say the property is only worth X amount, no more no less.

I know you mean well but the above, but I warned appraises about for years ( but some kept saying it anyway) The above is how lending appraisals became replaced with WAIVERS, which is what you say you want - an AVM range of value because appraisers are not "good enough" to stand behind a point value. Good god, if we say we are not good enough to do our jobs, what do they need us for? And that is how they saw it, since too many here kept saying it.

Our value opinion (not estimate ) does not mean the property is worth and no more or less as some universal fact - our value opinion is credible and reliable for its purpose, end of story..

I posted this numerous times in the past, that if appraisers kept saying "we are not good enough" , at some point Fannie and Freddie would believe them. I also said that if the appraiser did not pick th point value, then somebody else would - and that somebody now is the loan officer or CS price in a WAIVER ( as long as it falls within the FF AVM range )

The fact that a condition rating is somewhat subjective is not the problem, with enough competence the appraiser gets it right 99% of the time, enough to perform our job and the same with a point value. But we already lost the point value share of the business for a segment of lender work - good luck ever getting it back.

 

[George Hatch]

In terms of value conclusions, accurate is what calculators do. Reasonable is what appraisers do.

The first step in getting to "credible" is to establish expectations we can actually expect to meet. The assertion of unrealistic and unattainable expectations can only lead to failure. It only takes a few such failures to undermine our credibility, and with it our utility in the market.

One of the arguments out there is that overfitting the model is a problem because it implies a level of precision that won't hold up under further scrutiny. We can see examples of that in how some appraisers use the Fannie SC grids, but other valuation models are also vulnerable.

The reason appraisers don't take Zillow seriously is because we know their value conclusions are only sometimes within reason, let alone being "accurate".

Meanwhile, most of the valuation comes in comp selection. Not in the adjustments. In most cases if your comps are right then any valuation model can get to the reasonable conclusion. The wierdos are the exception primarily because of the lack of really comparable sales. If you had 20 recent sales of geodesic domes in your community that would be a pretty straightforward valuation problem, but when you don't have any it becomes a real headache. Regardless of which valuation model you use.

 

[J Grant]

The value of the appraisal is implied to be the exact value is it not? Banks don't use some value close to our value. No, they use our exact value to base their lending decisions? If appraising was so exact then if you gave 10 different appraisers the same assignment they should all come back the same exact valuation right? We all know that doesn't happen. In fact, you would mostly likely get 10 different values and the more difficult the assignment, the wider range of values you're likely to get.

This is why I think a range of values with a certain confidence level is a better, and more honest way to value a property. That doesn't mean appraisers should have some crazy wide range of values on every assignment but it should be within a small range. Some values could even be exact like when you're in a PUD and you have model match homes with no variations but it would be more accurate to have a range on more difficult properties. When I said we "good enough" I meant it's just our opinion and yes that can be supported but so can another appraiser who comes up with a different value who's value is also supported. You see statistics and data can be viewed and used in many different ways supporting different conclusions so if they are both supported who is to say one is right and one is wrong? Are you saying you're opinion is absolutely right every time and anyone else who comes to a different conclusion is wrong?

See my post 36 , though it is probably too late to get our lost business back from WAIVERS. I estimate we lost 40% of volume. Do you want to destroy the rest and put yourself out of a job?

Many appraisers who do not understand what our value is, deserve to be out of a job - we are not saying it is an "exact value"; it is a credible opinion of value supported by the appraisal data and development, and yes, our professional judgment is part of that. Appraising does not pretend to be exact. No point value on earth is exact, including ones from a computer or AVM.

A point value in lending sets the value for the LTV % ratio of the loan. That's it. It does not mean every other person on earth believes our value, or that other prices for a property are not possible.

 

[J Grant]

Whether a point value is generated by an AVM, regression, a wonder app, or an appraiser, it is a finite number and can not by its nature be "exact" as far as other possible prices a property might get in the market.
An AVM has a confidence score around their value to indicate that very thing. The appraisers is supposed to understand what they are doing wrt the limitations of their value opinion - the reason why it is for one specified client for a specified use and purpose with a disclosed definition of value as of one specified effective date etc.

The appraisal advantage is it is narrative and in English (or another language), so the reader can understand the process and can agree or disagree, use it or not., and accept it, or challenge it.

 

[George Hatch]

As for the range in values vs the point value there's no reason for appraisers to consider that an either-or choice. You can't even get to the one point value without first stopping off for a moment at a range. Even if only as the first step in your final reconciliation process.

Saying so in a report can at least provide the reader with some context for the number on the bottom line. If a reviewer concludes to a slightly different number but that number is still within the appraiser's original range then the reviewer can fairly consider BOTH point values to be within reason insofar as the concept of market value goes. They don't have to say the other appraiser was clearly unreasonable.

If Fannie/Freddie really thought precision was a doable expectation the UAD ratings would include more than the current number of ratings. Appraisers would be doing Q3.2 or C4.5 and applying market conditions adjustments by the day.