How do you measure decline?

[Marco]

I use a semiannual method for some cases.

I really appreciate all your responses. I've read the articles you referred me to and I down loaded them for future reference.

One method I've used, and I'd like your opinion on it, is that I compare the average sales price and/or price per square foot of comparable properties sold in the six months prior to the effective date of the report with the six month perior prior to that. I think this might provide a more accurate read of what's happening in the current market than, say, comparing the past year to the year before that, or the past quarter to the same quarter one year ago.

The problem with this method is that there is not always a large enough number of comparable sales to make me comfortable. In that case I have to modify my method by go further back in time and/or wider in geographic scope. But I wonder what the price of homes in 2006 has to do with the price of homes today in terms of the current rate of decline.

But the big challenge is not to establish whether the market is declining but declining at what rate currently i.e., over the last 90-120 days.
If the average sales price of home sold in the past quarter was down by 9% compared to the average sales price of homes sold in the same period one year ago, should that annual rate of 9% be applied to a home that sold 90 days ago? Furthermore, using a rate based on average price per zip code would be way skewed in most markets in my area as there is such a wide variety of sub markets and housing types within any particular zip code.

Thoughts?

 

[Truett Neathery]

The house appraisal I did last week was a 1962 unremodelled 4 BR 2 BA on one acre (minimum lot size in the area, some solds were up to 5 - 10 - 20 acres, I threw those out, of course) - this left me with 3 sales - up to 4.5 miles away in a different ZIP, 1 was an REO. That's it, folks. That's all there are. there were 6 sales in the last year, and 3 listings.
Although I have ordered Anthony Young MAI's DVD on "Graphing in EXCEL', I really believe the sample size on this one is too small to graph. It won't even look "scientific", which this trade isn't, let's face it.
TN

 

[Metamorphic]

One method I've used, and I'd like your opinion on it, is that I compare the average sales price and/or price per square foot of comparable properties sold in the six months prior to the effective date of the report with the six month perior prior to that. I think this might provide a more accurate read of what's happening in the current market than, say, comparing the past year to the year before that, or the past quarter to the same quarter one year ago.

Price per square foot is a great metric for properties where GLA is the primary factor influencing value. Its great because the PSF calculation factors out differences in GLA which makes for less variation. Basically its great for conforming homes. But in those circumstances it seems like it doesn't really matter whether you graph price or PSF. You get the same results. However there 's good reason to graph raw price. For one, you get to see on the graph the range of prices in the neighborhood and the relative activity in higher-end vs. lower end properties.

I think there's a lot of value in several years of data rather than just the period that covers the age of your comps. For one you can detect seasonal variations. For another you provide the report reader with a larger scale perspective on the depreciation ride this particular neighborhood has taken.

The problem with this method is that there is not always a large enough number of comparable sales to make me comfortable. In that case I have to modify my method by go further back in time and/or wider in geographic scope. But I wonder what the price of homes in 2006 has to do with the price of homes today in terms of the current rate of decline.

Obviously your best data is always going to be the close, most recent, model match. The whole trick of appraising is to find the best compromise when the ideal situation doesn't present itself. Measuring decline is not different....you widen your search parameters till you get a data set that you believe is definitive and the best possible.

But the big challenge is not to establish whether the market is declining but declining at what rate currently i.e., over the last 90-120 days.
If the average sales price of home sold in the past quarter was down by 9% compared to the average sales price of homes sold in the same period one year ago, should that annual rate of 9% be applied to a home that sold 90 days ago? Furthermore, using a rate based on average price per zip code would be way skewed in most markets in my area as there is such a wide variety of sub markets and housing types within any particular zip code.

I dont think its appropriate to try to identify a declining market on that short of a time scale. When you look at how much data you're realistically going to have in a 2 or 3 month period that's germane to your subject, then when you realize that within that time frame you're going to have 30 or more days of error in terms of when the actual meeting of the minds of price occurred (because close date isn't really when the decision occurs and even contract date is not that precise). With all those accumulated uncertainties, I think the decline rate you might derive on that short a time line, will have a lot of variability that probably doesn't exist in the market.

I think you're a lot better off looking at what has happened over the last year, dividing by 12, and using that as your rate. If you want to fine tune that number look at the most recent quarter or 6 months worth of data and see if looks like its increasing or decreasing in rate of decline and tweaking your 12 month base number.

 

[Metamorphic]

The house appraisal I did last week was a 1962 unremodelled 4 BR 2 BA on one acre (minimum lot size in the area, some solds were up to 5 - 10 - 20 acres, I threw those out, of course) - this left me with 3 sales - up to 4.5 miles away in a different ZIP, 1 was an REO. That's it, folks. That's all there are. there were 6 sales in the last year, and 3 listings.
Although I have ordered Anthony Young MAI's DVD on "Graphing in EXCEL', I really believe the sample size on this one is too small to graph. It won't even look "scientific", which this trade isn't, let's face it.
TN

Will this work for you?

 

[Truett Neathery]

Thanks, looks good but isn't!

There are yoo many dissimilar properties in the Meadow Vista - Applegate - Colfax- Christian valley market to cast such a wide net. Now, we're back to 6 sales, 3 listings in a year. Certainly not what's depicted in the graph. Nice artwork, though.

 

[Truett Neathery]

Spelling bee

I really can spell TOO, but I can't always hit the right letter

 

[Metamorphic]

There are yoo many dissimilar properties in the Meadow Vista - Applegate - Colfax- Christian valley market to cast such a wide net. Now, we're back to 6 sales, 3 listings in a year. Certainly not what's depicted in the graph. Nice artwork, though.

For purposes of identifying market decline the technique is not as sensitive to the specific area/neighborhood as you think it is.

This graph is the same 286 sale data set (less about dozen that were in a zip/MLS zone that only had a handfull of sales with the GLA/lot size/age restrictions). I split out the MLS zones and zip codes (which overlap obviously). As you can see they all track together, strongly suggesting that, at least for purposes of market decline they are undergoing the same level of downward pressure. Basically 4%/quarter. Exactly that what you measure on the first graph I posted.

However, when you break down to the smaller data sets you see more variability due to the outlier sales and you have to go to graphing by full year rather than the quarter basis that the larger data set will support. There's a $90/ft sale in 95722 that drags down the 2008 numbers and there's a high dollar sale in 95602 in 2005 that boosts that one up and in both cases (looking at a partial year) there's not enough normal sales to average them out.

 

[Gregory Beck]

By pairing the last closed sale with the lowest price alternative substitute(s), for me, seems to be more of a common sense and timely method. Why make it so complicated? If I can buy the same, or similar property for $300,000, why would I pay $325,000, all things being more or less equal?

HUH What did you just say??

 

[Steven R. Smith]

Truett, as kindly and gently as I can say this, Anthony Young is NOT an MAI. He only just joined the AI as an Associate this summer, after much prodding over a two year period, and has yet to take one course. 10 Courses, 5-years of Peer Reviewed Reports and a Thesis are required, along with a Degree.

September of last year, he took his first Stats class in Palm Springs. He is self taught with regard to Excel and has had great success promoting his DVD and Webinar's. Until the Stats course, with George Dell, MAI, SRA, last year he had never heard of Gnumeric.

He deserves to be commended for the masterful way he has been able to develop and promote his seminar and DVD. Which is mutually exclusive from having any credentials.

Those who do hold profesional designations, worked hard for them. Both George Dell and I have tirelessly promoted advanced education and done all we can to try and point the way for the residential appraisal community to increase their skillsets. I have created two new seminars this year to help the residential appraiser in this regard.

Documenting Time Adjustments and supporting Market Data Adjustments is probably done by less than 1% of the appraisers.

Although all licensed appraisers certify compliance with good appraisal procedures in their USPAP Certs. Most have their forms programs on Automatic, using what ever they were told to use when they started. And they use the same Adjustment factors in ever Location, Price Range, Size Range and Quality Range.

If one were to simply think about that for a minute, it is intuitively wrong, just wrong.

In my office we do Market Segmentation for what ever product we are appraising in the Location we are in. We have found differening Time Adjustments within the same Zip Code, as there are different Neighborhoods, and Market Segments within.

Then we compare and contrast what we measure, with what is published by Zip Code, City, County. By doing so it is our goal to bring context to the Time Adjustment we are using and bring strength to our analysis of why and what we are using.

So far, we have never been questioned about our Time Adjustment. We have used up to -10% Per Month, for June, 2008.

We are now measuring Weekly Time Adjustments.

 

[Blue1]

To me, it is more useful to graph not only sale prices and dates but number of sales per month. This produces (for me) a more reliable trend line.