Across The Board Time Adjustments?

[DARREN GODFREY]

When you grow up, your parents made you eat greens. Probably didn't like the taste at first, but they made you stronger. Same holds true to eating crow.

Sorry I had to get some sleep...so as I was saying . This appraiser adjusted every closed sale in the report based on his MC, I did more research on the MC and found that if you searched the subdivision there were 2 closed sales in the first period that reported much lower valuable one was about 50k lower a rancher in a sea of colonials and a foreclosure at about 70k below the other closed sales. In order to make sure I was not throwing this guy under the bus I removed those 2 sales from the MC report...the deviation was lower than his 7% adjusted for period adjustment...further more I found 2 almost Identical sales and did a paired sales analysis one from the first period and 1 from the last. The difference over almost a year was less than 1 percent.
The marketing times in this subdivision throughout the year are from 5 days to about 212 day...this is inconsistent with a red hot market. I feel the market is stable. In addition there are sales in the subdivision that are pretty much flips meaning there conditioning and upgrades have been bought up to a higher quality also leading to higher prices paid.
All interesting stuff.
Although I appreciate what you guys are saying , personally if I have to make across the board adjustment and when I do it is usually for a specific Item like a pool, porch etc. I go back to paired sales analysis to see if it is truly justified.

 

[NachoPerito]

Sorry I had to get some sleep...so as I was saying . This appraiser adjusted every closed sale in the report based on his MC, I did more research on the MC and found that if you searched the subdivision there were 2 closed sales in the first period that reported much lower valuable one was about 50k lower a rancher in a sea of colonials and a foreclosure at about 70k below the other closed sales. In order to make sure I was not throwing this guy under the bus I removed those 2 sales from the MC report...the deviation was lower than his 7% adjusted for period adjustment...further more I found 2 almost Identical sales and did a paired sales analysis one from the first period and 1 from the last. The difference over almost a year was less than 1 percent.
The marketing times in this subdivision throughout the year are from 5 days to about 212 day...this is inconsistent with a red hot market. I feel the market is stable. In addition there are sales in the subdivision that are pretty much flips meaning there conditioning and upgrades have been bought up to a higher quality also leading to higher prices paid.
All interesting stuff.
Although I appreciate what you guys are saying , personally if I have to make across the board adjustment and when I do it is usually for a specific Item like a pool, porch etc. I go back to paired sales analysis to see if it is truly justified.

If I were you I would go more macro on this. Hopefully your MLS shows market statistics with the median price on a macro scale.

 

[DARREN GODFREY]

When you grow up, your parents made you eat greens. Probably didn't like the taste at first, but they made you stronger. Same holds true to eating crow.

Sorry I had to get some sleep...so as I was saying . This appraiser adjusted every closed sale in the report based on his MC, I did more research on the MC and found that if you searched the subdivision there were 2 closed sales in the first period that reported much lower valuable one was about 50k lower a rancher in a sea of colonials and a foreclosure at about 70k below the other closed sales. In order to make sure I was not throwing this guy under the bus I removed those 2 sales from the MC report...the deviation was lower than his 7% adjusted for period adjustment...further more I found 2 almost Identical sales and did a paired sales analysis one from the first period and 1 from the last. The difference over almost a year was less than 1 percent.
The marketing times in this subdivision throughout the year are from 5 days to about 212 day...this is inconsistent with a red hot market. I feel the market is stable. In addition there are sales in the subdivision that are pretty much flips meaning there conditioning and upgrades have been bought up to a higher quality also leading to higher prices paid.
All interesting stuff.
Although I appreciate what you guys are saying , personally if I have to make across the board adjustment and when I do it is usually for a specific Item like a pool, porch etc. I go back to paired sales analysis to see if it is truly justified.

I am giving the OP some benefit of the doubt because I still am not sure what "across the board" means. Denis went there way back. I still don't know. Across the board makes me think of a set dollar amount or percentage. Like a concrete set amount. Like 1% or $5,000 period.

And, my advice to the OP would be to ask for more support since they believe the REO sale was relied upon so much.

Thanks for the benefit of the doubt lol... I would think more would give it to me since ...guess what I am doing the research, this is no knee jerk reaction. This is only the second time I am disagreeing with another appraisers opinion. I generally don't nit pick on these things because we all have the freedom of our own methods. The first disagreement I had was obviously some kind of fraud or outright number hitting as the appraiser simply left the neighbor to obviously superior neighborhood to get higher prices when there were plenty of comps available in the subject's neighborhood.

 

[DARREN GODFREY]

I


This can be true. I am a CG. But the time adjustment is not complex at all. Time adjustments are the easiest adjustments to make and to support.

With other adjustments I don't see how it is possible for a residential appraiser to justify many adjustments and still have time to get an appraisal done every day. You just don't get paid enough to do a lot of research.

Ok, the markets I am generally working in unless you carefully approach every comparable in the MC form you use , it can lead to some very unreliable #s due to flipping . The pre=flip and after flips values can make it look like a huge increase ...but then if you take 2 (for example ) C2 properties from the first and last period you will get a totally different level of increase. It is another reason why I am not a huge fan of this form.

 

[DARREN GODFREY]

If I were you I would go more macro on this. Hopefully your MLS shows market statistics with the median price on a macro scale.

I used paired sales analysis from the subdivision...it painted a much different increase than blindly throwing comps based on 20% +/- GLA . It is much more apples to apples. If this were a new subdivision , with all comps having the condition and basically the same levels of upgrades a broad sweeping form like the MC addendum would hold more weight.

 

[NachoPerito]

In the end, I think it is proper to make time adjustments for a changing market across the board. In my opinion, as appraisers we should make time adjustments in a changing market that reflect the market. In the search for market value it is the only reasonable thing to do. It would be wrong for you to ask the appraiser in this case to remove their time adjustments if they have it properly supported. I hope I am not missing anything here.

 

[DARREN GODFREY]

In the end, I think it is proper to make time adjustments for a changing market across the board. In my opinion, as appraisers we should make time adjustments in a changing market that reflect the market. In the search for market value it is the only reasonable thing to do. It would be wrong for you to ask the appraiser in this case to remove their time adjustments if they have it properly supported. I hope I am not missing anything here.

Well again , this is not a knee jerk reaction. I am questioning the basis of this adjustment...That is key. I can't say he is wrong unless the market data indicates he is. Hence recreating the MC addendum and going one step further with same subdivision paired sales analysis.

 

[hastalavista]

Well again , this is not a knee jerk reaction. I am questioning the basis of this adjustment...That is key. I can't say he is wrong unless the market data indicates he is. Hence recreating the MC addendum and going one step further with same subdivision paired sales analysis.

Which somewhat goes back to my original post (seems like years ago! ).

If the adjustment is supported, then it is supported. If the adjustment is supported, then it might support adjustments across the board.
If the adjustment is not supported, then it is not supported. If independent analysis by the reviewer concludes something else, then that independent analysis becomes the reviewer's basis for (a) disagreement with the analysis and (b) possibly a different value conclusion if an appraisal-review value is part of the review SOW.

What is not a basis is "I wouldn't do it this way, so another shouldn't" when all the recognized texts regarding market condition adjustments state the same thing: If they are supported, apply them; and if they are supported for all the sales in the grid, apply them to all the sales in the grid.

Now, separate from this specific appraisal under review, but applicable in general:
I find it hard to believe (but not impossible that such would exist) that a single data point at the front-end of the 1004mc would be such an outlier that it would impact the market conditions analysis. It would seem to me if that is the case, there simply isn't enough data in the entire analysis to make credible market-trend conclusions (and, I think that is one of the biggest complaints most raise concerning the 1004mc; by itself, it isn't always reliable). Further, it may not be a true comparable in the first place. I do think excluding it would be the appropriate thing to do if it was a flip/renovation, etc.

Second, if one is going to rely solely on the 1004mc for market conditions, one better understand its weaknesses.

I've posted this one before- I hope someone can point out where I have a flaw in this example because I cannot see it:

There are three buckets that contain data from different-sized periods and refine the trend to a data-point that reflects a median price. It may not be (and likely rarely is) the annualized price-trend many use it for.

Let's make this easy: Let's assume the distribution of sales is equal along the timeline within each bucket's period (and how often does that occur with a small data set?):
The first bucket is six-months (prior 7-12 months). Assume it has 9 sales and the median price is $450k. That means the median price is sale #5 as ranked by price. But it reflects all sales within that period, so the point-in-time may not be 12 months ago but on average would be approximately at the end of month 9.
The second buck is 3 months (prior 4-6 months). Assume it has 5 sales and the median price is $460k. That means the median price is sale #3 as ranked by price. But it reflects all sales within that period, so the point-in-time may not be 6 months ago but approximately 5.5 months ago.
The last bucket is the last 3-months (current to 3-months. Assume it has 5 sales and the median price is $465k. That means the median price is sale #3 as ranked by price. But it reflects all sales within that period, so the point-in-time may not be current but 1.5 months ago.
Here is how that data lays out:
9 months ago $450k.
5.5 months ago $460k.
1.5 months ago $465k.

What is the annualized appreciation? Most would say it started at $450k and ended at $465k, so that is 3.3%/year.
But what it means if sales are evenly distributed along the time line is, The median 9 months ago is $450k, the median 1.5 months ago is $465k. That is a $15k increase in 7.5 months; if you annualized that, you'll get something above 5%. And this assumes the data is evenly distributed; load the data toward month 7 in the first bucket and month 3 in the last bucket, and data skews to a higher appreciation rate.

My point? I wouldn't use this as my sole basis for making market-condition adjustments; as an originator or as a reviewer.

 

[DARREN GODFREY]

Which somewhat goes back to my original post (seems like years ago! ).

If the adjustment is supported, then it is supported. If the adjustment is supported, then it might support adjustments across the board.
If the adjustment is not supported, then it is not supported. If independent analysis by the reviewer concludes something else, then that independent analysis becomes the reviewer's basis for (a) disagreement with the analysis and (b) possibly a different value conclusion if an appraisal-review value is part of the review SOW.

What is not a basis is "I wouldn't do it this way, so another shouldn't" when all the recognized texts regarding market condition adjustments state the same thing: If they are supported, apply them; and if they are supported for all the sales in the grid, apply them to all the sales in the grid.

Now, separate from this specific appraisal under review, but applicable in general:
I find it hard to believe (but not impossible that such would exist) that a single data point at the front-end of the 1004mc would be such an outlier that it would impact the market conditions analysis. It would seem to me if that is the case, there simply isn't enough data in the entire analysis to make credible market-trend conclusions (and, I think that is one of the biggest complaints most raise concerning the 1004mc; by itself, it isn't always reliable). Further, it may not be a true comparable in the first place. I do think excluding it would be the appropriate thing to do if it was a flip/renovation, etc.

Second, if one is going to rely solely on the 1004mc for market conditions, one better understand its weaknesses.

I've posted this one before- I hope someone can point out where I have a flaw in this example because I cannot see it:

There are three buckets that contain data from different-sized periods and refine the trend to a data-point that reflects a median price. It may not be (and likely rarely is) the annualized price-trend many use it for.

Let's make this easy: Let's assume the distribution of sales is equal along the timeline within each bucket's period (and how often does that occur with a small data set?):
The first bucket is six-months (prior 7-12 months). Assume it has 9 sales and the median price is $450k. That means the median price is sale #5 as ranked by price. But it reflects all sales within that period, so the point-in-time may not be 12 months ago but on average would be approximately at the end of month 9.
The second buck is 3 months (prior 4-6 months). Assume it has 5 sales and the median price is $460k. That means the median price is sale #3 as ranked by price. But it reflects all sales within that period, so the point-in-time may not be 6 months ago but approximately 5.5 months ago.
The last bucket is the last 3-months (current to 3-months. Assume it has 5 sales and the median price is $465k. That means the median price is sale #3 as ranked by price. But it reflects all sales within that period, so the point-in-time may not be current but 1.5 months ago.
Here is how that data lays out:
9 months ago $450k.
5.5 months ago $460k.
1.5 months ago $465k.

What is the annualized appreciation? Most would say it started at $450k and ended at $465k, so that is 3.3%/year.
But what it means if sales are evenly distributed along the time line is, The median 9 months ago is $450k, the median 1.5 months ago is $465k. That is a $15k increase in 7.5 months; if you annualized that, you'll get something above 5%. And this assumes the data is evenly distributed; load the data toward month 7 in the first bucket and month 3 in the last bucket, and data skews to a higher appreciation rate.

My point? I wouldn't use this as my sole basis for making market-condition adjustments; as an originator or as a reviewer.

EXACTLY! We have better tools like paired sales analysis ...which is much more specific. You actually compare apples to apples. Thank You!

 

[Riick]

Just an FYI: Fannie has expressed concerned that we are *not* making time adjustments.

I recognize that the OP believes that the OA use of the two outlier sales he's pointed out in the 1004mc invalidates the time adjustment.

IMHO, the real problem is that the 1004mc, itself, is misleading, and, that (most probably) using sales from just one subdivision is a purely rotten statistical methodology. What sample size was there for the 1004mc to draw from? 20-30 sales, or 100, 200 ? Sample size had to be pretty small if 2 low sales at the start of the period skewed the data from what the OP considers the "actual" +1 percent/yr to the apparently +6 percent/yr that the original appraiser developed.

If the OA [or the OP] doesn't know enough of the theory/practice of statistics to recognize that a too-small sample size, or a badly drawn sample, makes any conclusion or prediction (such as time adj) drawn from the data very very very "inconclusive", he has no business developing statistics for an appraisal. Stop and take a course.

AFAIC, the 1004mc is a set of junk statistics, piled on top of a hill of bull excrement.
Having said that, we still have to give the client what they ordered, and they are insistent on getting a 1004mc.
My solution is to do the 1004mc, and *notify* the client why the 1004mc is PROBABLY GOING TO BE MISLEADING.
Then, I develop my own stats, maybe a graph of sales that might be sale prices in 1-mile radius over last 2 or 3 years, and graph of same sales as price/sq.ft. - now look at the two trendlines.
Yes, that may not be the best analysis of price trends in the area, but it beats a trend developed from 24 sales over the last 12 months.
.