Appraisal Bias

[alebrewer]

USPAP has no problem with a range, nor point value, nor a benchmark you use to label something as higher or lower than something else.

To me our analysis should 'tighten up' the adjusted values to a reasonable range... And that is nothing more than what the grid does. If your grid is actually wider after adjusting than before, then it is certainly wrong. But once you get to that supported range, then the final value you pick is based upon one of two things. Some final adjusted value based upon one sale or data point that is the most similar and most supported to the subject, or, your randomly selected "value" is just that. And in many cases, there are no dead to nuts machine precision arrows pointing to the exact number...even if all three sales are adjusted to an identical value, surely all three are not identical properties to the subject, even in cookie cutter Horton cheap built homes. At that point, the number YOU pick might be different from the number SOMEONE ELSE picks and it neither makes either one of you 'right' nor 'wrong'. At some point that final number is result of your heuristic analysis. And appraising is not science but it is not voodoo either.

heuristic (adj)
proceeding to a solution by trial and error or by rules that are only loosely defined.

I've always been intrigued by the number of folks who haven't ventured into the theoretical aspects of our profession. In my mentorship sessions, I'll often ask which should be tighter - unadjusted sales range or adjusted sales range. The number of blank stares I've had over the years is enlightening... we're basically performing a manual multi-variate regression (or should be) when performing the SCA.

 

[glenn walker]

I only call balls and strikes and I see no winners

 

[Zoe]

FHA, VA and GSE won't allow it George.

George have you ever argued with a stop sign?

CFPB is on verge of doing serious damage of current system. I can feel it in my bones.

 

[alebrewer]

But it would put you out of business. Do you understand that ?

I think we're finally getting to the crux of your angst, J. All your high brow integrity aside, methinks you're really just worried AI can do as well as you at producing a credible range of value...

 

[George Hatch]

I don't; deny the fact, that a property can sell for slightly different prices, but what does it have to do with anything ??

these appraisers are arguing against an appraiser providing a point value. Why do you refuse to address it ?

You keep picking on me instead and imagine I believe things that are not true of me, such as there can only be one value or one price for a property.

Citation needed. This is the post that set you off and it doesn't say or suggest anything about appraisers not complying with the terms of their assignments. It speaks to the users changing their expectations and it provides a reason for that which has zero to do with value creep in the market, that being a consistent theme with you. And then from there you commenced to criticize everyone who disagrees with you as having low standards.

 

[George Hatch]

FHA, VA and GSE won't allow it George.

Everyone knows what their current requirements are. The discussion has been about what they should or shouldn't allow, not about what they currently do/don't allow. .

They won't allow a range instead of the final number. OTOH, they also don't prohibit a range in addition to the final number. I don't think anyone actually foresees any changes on their end.

 

[alebrewer]

I don't agree.
I'll give a very simple example. The same subject house, a refinance so no distracting SC price - same subject 2000 sf house in a suburb. Adjusted Value range of the comps is 290k-308k

if the market was declining, wouldnt' an informed buyer pay less, thus it makes sense that the subject might be a point value of 290k or 295k??

And if the market was inclreasing, wouldnt' an ifnormed seller get a higher price, thus subject might be a point value of 305k or 308k?

Above is a very simple and clear reason why one point value is more supportable than the other, - yet you keep insisting that no point value is more or less supportable in a range.

Seems like you don't even trust your own range - how the heck could you trust a point within that range?

Let's say you're reviewing an appraisal I completed (and if there is a God - I sincerely hope that never happens - for your sake). Regardless - I conclude a range of between $300,000 and $306,000, and select a point value of $302,000 to conclude to, as that is the contract sales price. Now - that point value I selected could have been any digit within my range - and it would have been JUST AS SUPPORTABLE. Nevertheless, as an agreed condition of the assignment, I conclude to a point value - and use the contract price as basis for that conclusion. Back to you reviewing - you do your analysis, and you believe $303,000 is a MORE supportable point value than $302,000. Are you really that good? (hint: no).

 

[Terrel L. Shields]

FHA, VA and GSE won't allow it George.

I've always been intrigued by the number of folks who haven't ventured into the theoretical aspects of our profession.

Zoe obviously hasn't. No one is arguing that bankers want a range of values, nor courts, nor whatever. But the appraisal theory lacking in this profession is astounding. I bet in 9 of 10 appraisers offices, you'd be hard pressed to find as many as 3 hardcover textbooks on appraising. If they didn't get it during a class, they certainly never tried to buy an appraisal textbook. USPAP is their textbook and maybe a digital copy of FNMA selling guide and the 4000.1.

The range of the adjusted sales is one indicator of the "quality" of the data you are using and the greater the range, the less reliable the point value is going to be.

 

[Zoe]

The most powerful regulators in the nation are on that appraisal bias hearing. The writing is on the wall.

 

[alebrewer]

This is different than an AVM providing a range. Why? Because AVMs pick point values too! And yet, there are still major differences in the usefulness of an AVM and appraisal.

AVM's absolutely pick point values - and most often associate those with a confidence interval. Which is very closely related to an appraiser's adjusted sales range...