Paired Sales - Supporting Adjustments

[prasercat]

Analysis B: I found one sale-resale - custom home - mountain neighborhood. Analysis A was a six month over six month unit price analysis within the subject's neighborhood, and analysis C was a six month over six month unit price analysis over a wider geographic area, which included the majority of the subject's small town. Analysis C was the one with lots of data - twenty-five units in the past six months and seventeen in the six months prior. All unique custom homes, of course - widely divergent everything - ages from new to many decades old, shacks to about 5,000 SF, poor to very good quality, 9,000 SF lots to acreage, it goes on and on like that. Sounds a bit like what you deal with.

In A, instead of a six month over six month, I do a full year to full year median analysis with as targeted an analysis as I can get, since there are more data points this way; so, like, 4-15-2009 to 4-16-2010 compared with 4-15-2010 to 4-16-2011. I had one recently that had good correlations, within a percent or two, between the same home resales (I had six and listed them in the report side by side), the full year to full year median and the normal 1004MC grid. The 1004MC if often too exaggerated, in the up or down, so the other approaches should be getting me closer to reality. 2 or 3 percent can make a considerable difference in reconciled value when you have to go back up to a year for comps due to lack of density and sales.

It's always a struggle and tug-of-war between targeting the searches as much as possible to keep out as much randomness as possible but keeping the data points at a sufficient level to make a meaningful median analysis. It normally takes me many iterations to strike the best balance.

If there were enough same home resales, I would give it the primary weight. When I apprenticed and prices were going up, volume was so high, there were lots of resales, so the job of calculating the increase was much easier and the lenders loved them in the report. While I could prove prices were rising fast, I couldn't prove it was a bubble, at least at that time, even with the cost approach. There were fundamentals behinds it, like recent mass discovery of the location amenities, cheap on the basis of comparisons with other resorts and scarcity. When the California money dried up, it had a big burst and some very expensive developments went bankrupt; however, I'm sure it will recover in a few years.

When markets go up too fast, speculators pile in and make things crazy. I wish appraisers had better data to predict when a market gets over extended like those techical analysis charts on Etrade; such as; "The market studies herein indicate a rise in prices of 20% over the past year, which is now the 5th year of double digit increases. Looking at the price and momentum chart, one can see momentum is waning, volume is drying up, although prices continue to rise. According to these charts, we are due for a correction in the market" I would put a "sell" recommendation on this neighborhood and to buy gold" Ok, I'm kidding!

 

[David Mescon]

I agree. I do year over year analyses too, depending upon the specifics of the assignment and the dataset. This appraisal was for a homeowner who wanted to know what her home was worth. No MC, no 1004, and no stupid FNMA rules - just a real appraisal. It was a pretty tough one, though. Not the nicest house in the neighborhood, but definitely one of the nicer ones.

Real estate markets have been cyclical - probably since time immemorial. For the most part, from the industrial revolution through the S&L "debacle" the cycles were exceedingly predictable. Since the bust that began in the later part of 1988, the period became unpredictable - but the cycle remains the same. Relative stability, followed by a shorter period of moderate appreciation, followed by a (usually shorter) period of rapid appreciation, followed by a bust. Periods of rapid appreciation used to last typically for one to three years. Whenever the market hits that rapid part of the cycle, we know we're nearing the end. Whenever we enter the rapid appreciation phase, the market immediately becomes "over extended" as you so eloquently put it, and the reset will usually drop to somewhere around that point, give or take. Got to go pick up wifeypoo - she rode her scooter to work and it's raining now.

 

[AnonApprsr]

Each appraisal is independent of another, I could value a FP different for each and every appraisal I did. It's my opinion of the market reaction to differences relative to the Subject. I don't care if I have no paired sales, it's what my opinion is. It is what I'm paid to do.

 

[David Mescon]

Each appraisal is independent of another, I could value a FP different for each and every appraisal I did. It's my opinion of the market reaction to differences relative to the Subject. I don't care if I have no paired sales, it's what my opinion is. It is what I'm paid to do.

Could not agree more. Nothing quite as fun as taking some statistical snob's spreadsheets and turning his premises inside out with just minor changes to datasets, sample rates, etc... It frustrates me greatly to hear over and over about how adjustments are meaningless without statistical proof - which of course is just plain foolishness. Sometimes things can be proven - sometimes they can't - and sometimes the correct answer is "because I'm the appraiser and I said so." Oftentimes an answer can be shown to be correct by the inability to disprove it.

 

[ZZGAMAZZ]

The subject of yesterday's assignment has an in-ground pool. I found 4 pool comps and made downward adjustments to 3 of them because they are standard sized pools and the subject pool is much smaller...ours is either a very soft science, or we're fooling ourselves about the need for residential appraisals...

 

[McPheeters]

Sometimes things can be proven - sometimes they can't - and sometimes the correct answer is "because I'm the appraiser and I said so." Oftentimes an answer can be shown to be correct by the inability to disprove it.

...I am quite sure that answer will not work out so well when defending onself in front of the Board (not our Board anyway!)

Lotsa luck!

 

[residentialguy]

...I am quite sure that answer will not work out so well when defending onself in front of the Board (not our Board anyway!)

Lotsa luck!

You act as if you do something different. Not all adjustments in the Sales Comparison Approach can be directly extracted or supported by the available market data with a high degree of accuracy. Some adjustments have an element of subjectivity and professional judgment which the appraiser has applied based on prior observations of the reactions of typical/knowledgeable buyers' and sellers' in the marketplace. This method is a standard and well accepted practice within the appraisal industry.

Appraising Residential Properties, 4th Edition, Appraisal Institute, Page 342, "Limitations of Paired Data Analysis" states: "...This brief discussion of paired data analysis may seem to suggest that identifying the effects of property differences from market data is a straightforward procedure that can produce accurate, complete mathematical results in all appraisals. Such an impression would be misleading. Appraisers develop an opinion of market value by applying their judgment to the analysis and interpretation of data. Paired data analysis is a tool that an appraiser can apply to market data in some circumstances. When used in conjunction with other analytical tools, this type of analysis supports and guides the appraiser's judgment, but it does not take its place. Perfect sets of comparables that vary in a single, identifiable respect are rarely found. Because properties that are sufficiently similar to the subject are usually limited in number, the decision to apply paired data analysis in a given situation is a matter of judgment. Often the sampling size may not be larger enough to provide a solid statistical foundation for the appraiser's conclusions. Nevertheless, paired data procedures are important valuation tools that appraisers should use whenever possible. Identifying matched data sets and isolating the effects of variables is a practical methodology for studying market data, even if a comprehensive paired data analysis cannot be performed. When only a narrow sampling of market data is available, which would not lend itself to statistical analysis, paired data analysis can be used to test the results of other analytical procedures..."

 

[David Mescon]

...I am quite sure that answer will not work out so well when defending onself in front of the Board (not our Board anyway!)

Lotsa luck!

Let's see... I create four statistical models to support adjustment "A." All of them are mathematically correct and accurate. Each of the four datasets is comprised of equally reliable data. The datasets are the same, except, say, for chronological sample periods, which are one month, one quarter, semi-annual, and annual. Each yields a vastly different result. Which one is correct? Which one is "right?" Prove it. It usually still comes down to a judgment call. What are you gonna do "in front of the board" when I come in and present another equally, (or possibly even more thorough), analysis that completely contradicts yours? What if my analysis is more thorough than yours, contains more and better data, and on the surface, would appear more credible? Now, what if you were absolutely right originally, and my seemingly "well supported analysis" is wrong? What would you say? Would you abandon your judgment call simply because I presented you with some statistical model that supported a different opinion? I sure hope not. We used to be paid to give estimates of value. Now we're paid to opine value. Funny how around the same time we went from estimate to opinion our liability increased exponentially.

By the way, feel free to disagree with my "opinion" any day.

 

[23Degrees]

You act as if you do something different. Not all adjustments in the Sales Comparison Approach can be directly extracted or supported by the available market data with a high degree of accuracy. Some adjustments have an element of subjectivity and professional judgment which the appraiser has applied based on prior observations of the reactions of typical/knowledgeable buyers' and sellers' in the marketplace. This method is a standard and well accepted practice within the appraisal industry.

Appraising Residential Properties, 4th Edition, Appraisal Institute, Page 342, "Limitations of Paired Data Analysis" states: "...This brief discussion of paired data analysis may seem to suggest that identifying the effects of property differences from market data is a straightforward procedure that can produce accurate, complete mathematical results in all appraisals. Such an impression would be misleading. Appraisers develop an opinion of market value by applying their judgment to the analysis and interpretation of data. Paired data analysis is a tool that an appraiser can apply to market data in some circumstances. When used in conjunction with other analytical tools, this type of analysis supports and guides the appraiser's judgment, but it does not take its place. Perfect sets of comparables that vary in a single, identifiable respect are rarely found. Because properties that are sufficiently similar to the subject are usually limited in number, the decision to apply paired data analysis in a given situation is a matter of judgment. Often the sampling size may not be larger enough to provide a solid statistical foundation for the appraiser's conclusions. Nevertheless, paired data procedures are important valuation tools that appraisers should use whenever possible. Identifying matched data sets and isolating the effects of variables is a practical methodology for studying market data, even if a comprehensive paired data analysis cannot be performed. When only a narrow sampling of market data is available, which would not lend itself to statistical analysis, paired data analysis can be used to test the results of other analytical procedures..."

Anyone else find that the regression models themselves provide ample support for the above?

Based on some limited experience working with one of the regression programs it seems possible that what is actually being shown is the inherent nebulosity tied to the contributory value of any particular parameter. The ranges provided by the program for these adjustments are almost always large to the point of being meaningless - and this is with reasonably sized data sets. Barring reliance on the often rare and anecdotal "paired sale" are these well designed statistical models displaying that the tying of a precise number to a particular parameter without indicating for the sake of clarity that "professional judgment" played a large if not dominant role in the final determination is, for lack of a better word, "misleading"?

As for what a state board will say - that could be a function of who is on the state board. After reading this forum on the subject of adjustments over the years it seems that there are those that claim the ability to tie these numbers down to narrow dollar precision and provide undeniable levels of numerical support (which presumably would be supported by some level of statistics as is the case for most numerical pursuits that make such claims) and would probably expect the same of others.

Others might recognize that the answers are often fluid, open to subjective interpretation, and would therefore be likely to follow the guidelines of "credible" and "reasonable" when sitting in judgment. The determination of just what exactly is "proper support" for the adjustments in question might vary substantially between the two views. And both would probably have increased expectations in the case of the "across the board" adjustment.

Although the movement toward a purely qualitative analysis is not in the cards, the Reynolds Paper is always worth a citation whenever this discussion pops up:

http://appraisalnewsonline.typepad....-the-market-qualitative-vs-quantitative-.html

 

[Joyce Potts]

Paired Sales is the optimum solution, but in the absence of paired sales, where's what go in ALL of my appraisals. I've posted this before. THINK ABOUT IT -- and how it may very well keep you out of trouble. Many make fun of my BOILERPLATED EXTENDED SOW -- but so far it's worked for me (not yet had a FREAB complaint, but only a matter of time, I'm sure) and for several others whom I've helped successfully defend regulatory complaints. DO NOT MISLEAD THE CLIENT OR THE INTENDED USERS. SPELL IT OUT.

DISCLOSURE AND DISCUSSION OF PAIRED DATA ANALYSES AND ADJUSTMENTS:

Not all adjustments in the Sales Comparison Approach can be directly extracted or supported by the available market data with a high degree of accuracy. Some adjustments have an element of subjectivity and professional judgement, which the appraiser has applied, based on prior observations of the reactions of typical/knowledgeable buyers and sellers' in the marketplace. This method is a standard and well-accepted practice within the appraisal industry. All interested parties are encouraged to have an understanding of basic valuation practices when appraising atypical or complex properties; or where there is an extreme absence of like elements of comparison; or in instances where the market data is inconsistent on which to base better supported adjustments and/or overall value conclusions. Individual adjustments can not be relied on independently.
Appraising Residential Properties, 4th Edition, Appraisal Institute, Page 342, "Limitations of Paired Data Analyses" states:

"...This brief discussion of paired data analysis may seem to suggest that identifying the effects of property differences from market data is a straightforward procedure that can produce accurate, complete mathematical results in all appraisals. Such an impression would be misleading. Appraisers develop an opinion of market value by applying their judgement to the analysis and interpretation of data. Paired data analysis is a tool that an appraiser can apply to market data in some circumstances. When used in conjunction with other analytical tools, this type of analysis supports and guides the appraiser's judgement, but it does not take its place.

Perfect sets of comparables that vary in a single, identifiable respect are rarely found. Because properties that are sufficiently similar to the subject are usually limited in number, the decision to apply paired data analysis in a given situation is a matter of judgement. Often the sampling size may not be larger enough to provide a solid statistical foundation for the appraiser's conclusions.

Nevertheless, paired data procedures are important valuation tools that appraisers should use whenever possible. Identifying matched data sets and isolating the effects of variables is a practical methodology for studying market data, even if a comprehensive paired data analysis cannot be performed. When only a narrow sampling of market data is available, which would not lend itself to statistical analysis, paired data analysis can be used to test the results of other analytical procedures..."

For example, say an appraiser finds three sales that are nearly identical in a tract/production built subdivision, with the sole exception being that one sale has a three-car garage, while the others have two-car garages, yet their was no selling price differences.

Another example is when the available data has multiple dissimilarities with different prices, it often becomes impossible to determine how much each individual variable has contributed to the price differences. In theory and in perfect markets, using a market grid and paring the sales should result in a relatively close value range after the extracted and applied adjustments. --Factor in transitional market trends and it often becomes virtually impossible to derive identifiable individual units of comparison.

The Appraising Residential Properties, 4th Edition, Appraisal Institute, Page 344, "Other Quantitative Adjustment Techniques further states that in " instances where paired sales analysis is not conclusive, the appraiser may apply judgement to resolve the problem." The adjustments resulting from the appraiser's judgement is based on a study and understanding of historic or past buyer preferences. It further suggests that cost and depreciated cost data may be used with the appraiser arriving at the value contribution (not cost new) of certain amenities. In effect, the sales grid found in any pre-printed residential reporting forms afford every appraiser a convenient 'paired sale format' and should make paring the sales very easy. Yet, has anyone ever compared two appraisals done by two different appraisers for the same property where all adjustments were all the same?

When the market data affords such information, multiple paired sales in support of any and all adjustments is the optimum scenario. Unfortunately, 'the real world seldom affords consistency. One only need to look at the limited number of closed sales, inconsistent pricing, the oversupply of active listings, average days on market and price reductions within many sub-markets to gain a realistic insight into current market trends and the inherent challenges from a valuation perspective.