Matched Pair Analysis
The key is to find data sets where one property is highly similar to another property. They don't have to be identical, however. Note that at least one major variable SHOULD be different.
Using this analysis technique is a measurement of WHAT THE MARKET INDICATES, not what the appraiser thinks. The best way to communicate the concept is by example.
In this example, you have located the sale of a house in a large subdivision; we notice it is the House A floorplan. You will be comparing House A with the sale of a House B floorplan. In this subdivision, shall we say, there are only 6 floorplans (A through F), all built by the same builder. In this subdivision, all the houses are new or relatively new.
The lots in the S/D are all identical in size, frontage, etc., thus the lot value of all homes sold can be assumed to be highly similar. The homes are all built by the same builder and have highly similar construction details and interiors. All the homes have been sold over a 12 month period in a highly stable market. The two sales both occured 6 months ago. All of the House A and House B plans are in the 3BR 2Ba genre. Both House A and House B have a dbl garage and a fireplace. In fact, this builder does not like to deviate much from house to house. Differences in his homes are mostly cosmetic.
Importantly, all of the homes are sold with highly similar financing and concessions packages (more about this later). For now, let us say that all homes have sold using FHA financing with 3.5% down and with the builder paying $2,500 toward closing costs. Thus, financing and concessions for all sales is highly similar.
Thus, we can see that these two house plans, while they look a little different as to roofline, porches, front doors and paint colors, are highly similar in almost every respect. They also have similar financing and concessions.
However, they are not the same size. The House A has 2,000 SF while House B has 2,400 SF.
The question, then, is how much does size affect the price of the two homes.
Home A sold at a price of $200,000. Home B sold at a price of $240,000.
There is a 400 SF difference in size and a $40,000 diffeence in price. Dividing $40,000 by 400 SF = $100 per SF. Basic and simple.
Since there are no other variables affecting the price, this is your size adjustment, as indicated by the analysis of the two homes. It is a mathematical proof and, in this example, is virtually irrefutable.
To make the case stronger, however, you need more than one example. As Dick West once said, "one sale does not a market make".
You now want to compare House A with another House B floorplan located down the street. As it turns out, the second instance of the House B floorplan (ie another address) as shown in the MLS indicates that it also sold at $240,000. Thus, the result is the same. A size adjustment of $100/SF is proven a second time.
Yet, being a stubborn and hard to satisfy professional, you demand yet further corroboration of your research results.
You decide to test it against one of the House C floorplans. You check the C size and it is 2,700 SF. In fact there is one that sold 3 doors down. It had a sale price of $260,000. Uh oh, this one only brought $96.30/SF. Shouldn't it have been $100.00/SF too?
Concerned that your theory is weakening, you test House A against another House C floorplan. Sure enough, this house sold recently, as confirmed by the MLS, at $260,000. This, also, reflects $96.30/SF.
Does the latter analysis disprove the first? You search for several other House A versus House B floorplan comparisons and find 5 more via MLS right in the same subdivision that support $100/SF.
That makes 7 paired-sales-analysis comparisons at $100/SF. The statistical strength of this figure then being reasonably accurate is very high, assuming the subdivision is itself competitive with other similar subdivisions in the overall broader neighborhood.
But what about the House A versus House C analysis. How is that differential explained? This is where the seasoned professional differs from the newb.
Typically, as size increases in homes the price-per-SF should tend to lessen so long as "wet areas" (ie kitchens, bathrooms, laundry rooms) remain proportional to the overall size. This is due in part to the well-known concept of "economy of scale". But it also relates to the fact that the large rooms in a home can be expanded in size more cheaply than wet and dense areas in a home. These are the more expensive areas to build. Wet = kitchens, bathrooms, laundry rooms, etc; Dense areas = areas with cabinets, bookcases, custom floors, wood moldings, specializations, etc. That is to say, it is far cheaper for the developer to enlarge a bedroom or family room than add kitchen/bath space. One could also demonstrate the impact of fixed costs to the builder versus variable costs.
Thus, as the differential size of the comparison increases, the market is typically going to reflect this factor. The House A versus House C comparison reveals this: a diminution of the size adjustment as the size increases, assuming quality of construction remains similar and certain construction ratios are observed.
That is why it is best to stay as close to the same size as possible when appraising homes. It also demonstrates, oddly enough, that size adjustments may vary from comp to comp (though few software makers and reviewers seem to know this). Of course, they might argue that if the size adjustment did in fact differ, then that particular comp should not have been selected in the first place. Too much variation in the size differential may distort accuracy.
Which brings us to an even higher level discussion about curves. Almost all variable data sets in real estate move on curved lines when graphed. The size adjustment derived from a very large neighborhood data set would most likely emerge on a curve. The curve in the example would start at $100/SF and then trend lower as a function of size. Thus, as the size increased within the large data set (assuming once again equivalent lots, construction, etc.) the curve moves down to $96.30/SF, then to $95.00/SF, then to $94.00/SF and so forth until you leave the data set's capacity to measure the change.
Thus, size adjustments actually follow a gradual curve, since size is highly variable among residences. We sometimes do not (cannot) realize there is a curve because the data sets are not large enough and the sales volume is not active enough.
If one wishes to be a professional appraiser in the modern world, you must gain an understanding of logical proofs. These proofs stem from the study of geometry and algebra. Many modern students are skipping geometry in high school and then take no further geometry in college. Without it, you will never be the appraiser you could be. The same applies to the study of statistics. This subject is not always taught in high school as a full course. Of similar value is the study of functions in mathematics courses often termed pre-calculus. Higher levels of mathematics are of course better, but not absolutely necessary.
If one has not had at lease one course, preferably two, in each of these areas, you are at a real disadvantage in the modern world of real estate appraising, whether it be residential or commercial. Additional classes are usually available at the local college.
That being said, we can move back to paired sales analysis.
What about homes with two variables that are different?
Well that's what makes it interesting. Especially since there might be 3, 4 or more variables that differ in the real world.
The key to understanding the approach to multiple variables is to simply establish one at a time, and then chain the "known" or "proven" adjustments.
Back to the example.
In the example used above, the builder has added a new addition to the subdivision. He now has decided to add a 3-car garage to one of his old floorplans because he thinks it will sell well. His B floorplan was his best seller so he asks his draftsman to enlarge the garage from 2 cars to 3 cars, but keep everything else highly similar.
The appraiser has already proven that the movement from House A to House B results in a size adjustment of $100.00 per SF. And after 2 more data samples from the new addition, sure enough, the House B plans are still selling at $240,000.
The new House G plan is the one with the 3 car garage. It is selling, at 2,400 SF but with a 3-car garage. A quick review if the MLS reveals that there are 3 sales of the House G plan, all at $247,500. Thus, it is obvious that the adjustment for the extra garage size and door is $7,500 (roughly $30/SF for day 250 SF).
Now it is possible fro the appraiser to use both House B and House G plans as comparables in his appraiser of a House A floorplan. Adjust $100/SF for size and $7,500/SF for the 3-car garage (versus 2 car garage). Or, any combination of the three plans. The mathematical power of paired-sale analysis can be extremely compelling.
That being said, there must be a note of caution thrown in. There are many caveats that apply in paired sale analysis. In the above "textbook" subdivision, the data is highly standardized and the builder has a predilection for order and repetition. In the real world, that is rarely the case.
Of special note is the financing and concession consideration. If this is not adjusted properly,
it will throw most if not all of your paired-sale conclusions off. The reason is one of apples versus oranges.
In our example, the House A plans have sold consistently at $200,000 with highly similar financing and closing costs for over a year. However, the builder's sale staff is complaining that they are losing sales to the subdivision down the street because he is offering a highly similar product but is offering to pay $5,000 in closing costs. Although the other builder prices the homes at $202,500, to make up the extra money paid out (same net), the buyers are willing to pay it because that means less cash at closing and a small difference in house payment. So, to keep peace, the builder does the same thing. He now offers House A (and all others) at an extra $2,500 and now pays $5,000 closing costs instead of $2,500. Thus, all the sales that occur after this decision, which has nothing to do with the real estate itself, are at $202,500 (oranges) instead of $200,000 (apples). If one did not know better, one would assume it was a simple price increase or that the market had changed for the better. On the contrary, the upward price movement was simply a manipulation of the concessions offered.
Unfortunately, most residential lenders and underwriters don't understand financing and concession adjustments, or, better said, don't want to see them. Maddeningly, many Realtors do not report them accurately in the MLS and most appraisers simply leave them off the form entirely. This partly stems from the fact that most justifiable financing and concession adjustments are negative to the sale price. Including a negative number in this part of the form will almost always provoke dismay on the part of the lender and the home owner, especially if your appraisal is below a sales price. What will happen to the poor appraiser if he refuses to bow to pressure to remove the concession adjustment is left to the imagination. Since the overwhelming majority of appraisers never make the adjustment, if one is required, the appraiser attempting to make the adjustment is deemed "outside the main stream" and therefore wrong.
When mathematically demonstrating adjustments using paired sales analysis, it is IMPERATIVE that the appraiser, first and foremost, appropriately obtain accurate data and then analyze the impact of financing and concessions on the sale price. This must be done prior to any other analysis. Otherwise you have a garbage in/garbage out scenario.
A further word of caution applies: many MLS sources do not display accurate or meaningful financing and concession data. Almost always, the Realtor is inputting this data themselves and since they frequently do not understand its import or impact, it is inaccurately stated or is so convoluted as to be unintelligible.
The best way to avoid this is to get a copy of the HUD closing statement or have it read to you. The appraiser can then compare with accuracy the sale price and any concessions paid by the seller. Unfortunately, in an era of privacy constraints, this is often not available. Thus, obviously, paired sale analysis derived strictly from the MLS, unless the MLS is meticulous in its reporting standards, is subject to being tainted.
Once again, the public and the government want good appraisals, built on good data and analysis, but throw up road blocks to the data. Appraisers are being held to a high standard but get very little governmental or policy support that would enable us to do the best job possible. I will have more to say about this in later submissions.
Jeffrey T. Couch
