Underappraisal Disparities and Time Adjustments to Comparable Sales Prices in Mortgage Appraisals

[Joe Flacco]

I reviewed around thirty appraisals this year. They weren't purchase or refi valuations, they were for government program documentation. Narratives but with form sections in many. Bottom line these were prepared by folks claiming extensive knowledge and experience. They were embarrassingly replete with errors and poor methodology. But to the topic at hand, in particular, the market condition analyses were just simply absent. I could be preaching to the choir here. These reports I was looking at were prepared in 2022 and 2023. The property types were infrequently transferred. This resulted in, among other things, a long research time period. Many had sales from 2018 to 2023. This a rather unusual handful of years as we all recall, in terms of the value of the dollar, real property transfer frequencies, etc.

So, there would be like five comps usually. Spread all over time. Sizes were also pretty wide differentials. So the per acre prices were relatively tight, lets say. But The older sales were much smaller than the more recent ones. The entire market conditions analysis would be a sentence that said the collected data showed no indication for market condiution adjustment. Or something along those lines. I mean the dollar had fallen like 17% or something between some of these sales, and you could eyeball that sizes differences were muting the time differential, at least there should have been a suspicion! Saw this in report after report. I kept asking myself if maybe they didn't mention this in preparation for the license tests any longer.

I can understand if you've real comparable data, real close geographically, and real close sale prices, all real recent, probably ok with saying no market con adjust. But if you know there's inflation galore, your sales are from three four years apart, and prices are close but there are some differentials, do you stop there or do you do some market condition analysis. What sort of analysis is considered adequate support to apply an adjustment, how much is enough to reject it. What data sources do you study? I don't know if this article has hit on something, but i definitely noticed this trend this year.

Prices of all properties didn't go up. Look at this one.

 

[CGinMN]

Under appraisal is defined as appraised value below contract price, without any adjustment for seller concessions.

Here’s another study that doesn’t account for the fact that seller concessions inflate contract prices, and that seller concessions are disparately used by FTHBs and minorities.

We focus on an important source of misvaluation and underappraisal: inadequate adjustments for rapid house price movements and the role they play in driving disparities.

The study doesn’t attempt to differentiate between underappraisal and misvaluation, rather they conflate the terms in a shell game. I have noticed the language these studies use is carefully selected.

As a benchmark, we construct a predicted time adjustment to ascertain if a comparable sales price needs updating. To do so, both a high time frequency and geographic precision are needed. We choose a source that meets those criteria and is freely available: the monthly ZIP Code-level, non-seasonally-adjusted, price indexes from Zillow.

So in the study, to cure underappraisals, they apply their own time adjustment based on Zillow zip code level price appreciation to comps that lack a time adjustment and voila, they cure a lot of under appraisals. It really shouldn’t be surprising that a positive adjustment to the comps - in any category - would cure a lot of underappraisals. Whether it is actually warranted, or accounted for in another way in the analysis, isn’t considered.

We know many appraisers account for rising prices via qualitative analysis (reconciling up) rather than by making an adjustment. So in many cases applying Zillow-based time adjustments is double dipping to overstate the result.

An AVM will apply a time adjustment to 100% of the comps. An appraiser, who understand seasonality and is doing it correctly might only apply time adjustments 3-4 months out if the year.

IMO is another misapplication of big data analysis.

 

[shrubberyvaluation]

Under appraisal is defined as appraised value below contract price, without any adjustment for seller concessions.

Here’s another study that doesn’t account for the fact that seller concessions inflate contract prices, and that seller concessions are disparately used by FTHBs and minorities.

We focus on an important source of misvaluation and underappraisal: inadequate adjustments for rapid house price movements and the role they play in driving disparities.

The study doesn’t attempt to differentiate between underappraisal and misvaluation, rather they conflate the terms in a shell game. I have noticed the language these studies use is carefully selected.

As a benchmark, we construct a predicted time adjustment to ascertain if a comparable sales price needs updating. To do so, both a high time frequency and geographic precision are needed. We choose a source that meets those criteria and is freely available: the monthly ZIP Code-level, non-seasonally-adjusted, price indexes from Zillow.

So in the study, to cure underappraisals, they apply their own time adjustment based on Zillow zip code level price appreciation to comps that lack a time adjustment and voila, they cure a lot of under appraisals. It really shouldn’t be surprising that a positive adjustment to the comps - in any category - would cure a lot of underappraisals. Whether it is actually warranted, or accounted for in another way in the analysis, isn’t considered.

We know many appraisers account for rising prices via qualitative analysis (reconciling up) rather than by making an adjustment. So in many cases applying Zillow-based time adjustments is double dipping to overstate the result.

An AVM will apply a time adjustment to 100% of the comps. An appraiser, who understand seasonality and is doing it correctly might only apply time adjustments 3-4 months out if the year.

IMO is another misapplication of big data analysis.

Applying adjustments based on zip code. Sometimes different areas of the same zip code change differently including sub markets and different price points.

 

[bsilver]

Prices of all properties didn't go up. Look at this one.

View attachment 94253

I didn't say all prices went up. Any particular market sector in any particular geography is going to have a different experience over any given time period. My point is only that making a conclusion that there was no change, without presenting ANY data, not even an explanation, not even sales in the in house database and talks with realtors. Just a statement, no market conditions were required. Do you do that? I saw dozens of reports like that, which made me think maybe there techniques to test the need are not known to a good percentage of appraisers.

 

[Sadie]

No data for over-appraisals? They are more harmful than under appraisals.

 

[bsilver]

No data for over-appraisals? They are more harmful than under appraisals.

Mis-valuation due to a lack of analysis data support is equally harmful regardless of if results in over or under valuation. Understand that appraisals have many users and many intended uses. An over valuation in a divorce case is going to hurt one of the ex-es and help the other. In a condemnation case with a Before and After, an over on the total difference hurts the taxpayer and helps the condemned. If there's an under on the before and an over in the after, the condemned is harmed TWICE, with errors in both directions. Comparative weighting of the degree of impropriety does not abrogate the impropriety.

 

[Sadie]

Mis-valuation due to a lack of analysis data support is equally harmful regardless of if results in over or under valuation. Understand that appraisals have many users and many intended uses. An over valuation in a divorce case is going to hurt one of the ex-es and help the other. In a condemnation case with a Before and After, an over on the total difference hurts the taxpayer and helps the condemned. If there's an under on the before and an over in the after, the condemned is harmed TWICE, with errors in both directions. Comparative weighting of the degree of impropriety does not abrogate the impropriety.

IMO. in residential markets, over valuation has many more consequences than an under appraisal. Its just that they are farther down the road than an under appraisal. The 'study' I believe was referring to residential appraisals used for the purchases.

 

[Joe Flacco]

I didn't say all prices went up. Any particular market sector in any particular geography is going to have a different experience over any given time period. My point is only that making a conclusion that there was no change, without presenting ANY data, not even an explanation, not even sales in the in house database and talks with realtors. Just a statement, no market conditions were required. Do you do that? I saw dozens of reports like that, which made me think maybe there techniques to test the need are not known to a good percentage of appraisers.

No, I don't do that.

 

[Fernando]

Prices of all properties didn't go up. Look at this one.

View attachment 94253

Prices aren't going up.
I'd been keeping an eye on this one property which has been listed for 3 months and still on market.
I appraised it 3 years ago and hate to think that I may have appraised it too high back then.

 

[Joe Flacco]

Prices aren't going up.
I'd been keeping an eye on this one property which has been listed for 3 months and still on market.
I appraised it 3 years ago and hate to think that I may have appraised it too high back then.

Some properties still going up. Not the rate of pandemic rate but more normal 5% per year or so. Some properties are flat since around 2022 and some are lower. Some properties have not changed much since the beginning of the pandemic.