Appraisal Issues Found On 37 Percent Of Reverse Mortgage (poll)

[Doug Wegener]

Do you think that FHA findings that 37 Percent of Reverse Mortgages are overvalued by 3% is credible?

 

[Doug Wegener]

Do you think that FHA findings that 37 Percent of Reverse Mortgages are overvalued by 3% is credible?

Sorry, flawed poll. I read an article in the WSJ that did not mention that they used an AVM which was their methodology. Ignore choice #3

 

[Randolph Kinney]

What was the standard used to know the actual value? What was the standard error?

 

[Doug Wegener]

What was the standard used to know the actual value? What was the standard error?

The FHA told reporters that out of the 134,000 appraisals inputted into its automated valuation model, approximately 50,000 (37%) were inaccurate by at least 3%

 

[Randolph Kinney]

The FHA told reporters that out of the 134,000 appraisals inputted into its automated valuation model, approximately 50,000 (37%) were inaccurate by at least 3%

Nonsense. Their model to determine value needs to be validated by the market, not some global warming model that can't predict reality before it occurs.

 

[KHS445]

From a mostly non-bias point of view I would say without a great deal of additional information I would have to be somewhat suspect of the methodology and frankly even how the conclusion is framed. It sounds like they are saying an AMV is 100% accurate in all cases all of the time, so that is the standard an appraisal should be judged by. Rather than saying you know these AMV things are pretty accurate because in 63% of the cases there was less than a 3% variance from a regular appraisal. I really doubt there wouldn't be a similar valuation spread if you had 10 different appraisers prepare a report on the same property.

It would be interesting to know where the properties were located, as I would suspect that the valuation of a tract home with a number of nearby sales should be fairly accurate. On the other hand a more rural property with limited sales in the immediate area might have a greater variance. I would also like to be able to drill down into the actual appraisal and the AMV for a particular property to see where the actual difference is. Might it be condition, view, basement finish, outbuildings, etc. Were the appraisal and the AMV prepared using the same effective date? In a stagnant market the difference of up to six months may not make a huge difference, but in an increasing or decreasing market it could. Any idea if the AMV was prepared using the same guidelines as an appraisal as far as the date of the comparable sale, the distance from the subject, bracketing the subject, etc.?

All in all probably a good start but hardly a damning of appraisals or AMVs. The devil is in the details. In my opinion the use of AMVs for routine valuations may actually be a good thing for many in the appraisal community. Sure it may cause some problems for the "turn and burn' outfits, but it will leave the tougher assignments for the licensed appraiser to complete. More difficult and time consuming reports result in higher fees. I could see where fees for a residential property that can not be satisfactorily valued using an AMV could run $500 to $1000+. As someone's tagline said a few years ago "4 @ $200 is different that 2 @ $400".

 

[Elliott]

Here's the stats for Zillow for Portland, OR:

Portland, OR Four Stars Zestimates, 762,000 homes 65.3% were within +/- 5%, 84.8% were within +/- 10%, 93.3% were within 20% and viola, Zillow gives itself a median error of 3.4%! and 4-gold stars (What??--BS, Classic "regression to the mean", or a meaningless massaged number).

Use enough statistics and you can confuse anybody. I was at an AI Seminar and speakers made claims of R^2s of 0.20 and even 0.04 as being a significant correlation. The misuse of statistics in appraising is appalling.

FHA bought a black box that gave them the number they wanted to hear, nothing more.

 

[Joe Flacco]

One of my clients has a habit of putting in a estimate of value when they send a order. I wonder where this estimate comes from so I look on Zillow and that seems to be where they get it. They are probably very often disappointed when they get the appraisal report.

 

[Joe Flacco]

What these AVM's don't understand or account for is that one or two factors can have a big impact on value. It is not all the little differences each have a little bit of impact on value which is what AMV's seem to do. Like a parking space. It does not understand that not having a parking space in a condo project in the city is going to be a big impact on value.

 

[Deleted member 128537]

3% is negligible. Again when doing RELO work the standard was 5%. 2 appraisers had to be within 5% of each other on the same house; or we had to discuss our value differences. My question is 3% from what? 3% from the 5% margin on other independents valuations? Or 3% from the actual sale price. I don’t see how ANYONE can determine the latter.

The problem with the WSJ article is that it’s read by people that have no idea about the complexity of appraisal estimates. People go 3% and freak. I see 3% and say, “NOT BAD.”