Paired Sales Analysis

Paired sales analysis puts the question to the market and reads the answer. Two recent closed sales are found that are as alike as possible in every respect except the difference being measured; the difference in their prices is what buyers in that market paid for that difference. The method is direct, it is widely used by appraisers and other market participants, and its reliability depends chiefly on how close the pair is on everything else.

The requirement is similarity on every variable other than the one being isolated. Two houses in the same subdivision, the same style, the same size within a few percent, the same condition, closed within a few months of each other, differing only in garage configuration, produce a clean garage adjustment. The further the pair drifts on other variables, the more the price difference carries things other than the garage, and the less it can be relied on.

Perfect pairs are rare, and imperfect pairs are still worth having. A pair that differs on size as well as garage can be corrected for size first, using the living-area rate derived by the Ratterman Method, and the residual difference read as the garage. A pair that is close but not exact still shows whether the figure is $5,000 or $25,000, which is the difference between an analysis and a guess. Three or four imperfect pairs pointing to the same range are stronger evidence than one pair that appears perfect.

Finding a pair

A pair is built from the subject’s market area outward. The feature being measured is named first, and the search is drawn tightly around everything else: closed sales in the market area over the past twelve months, the same property type and style, a living-area band narrow enough that size barely matters, and the feature present in some sales and absent in others. The MLS search produces candidates; it does not produce pairs. Each candidate is confirmed against the record: the listing photographs and remarks for condition and finishes, the assessor’s field card for year built, lot size and outbuildings, the building department’s permits for recent work, and the closed listing for concessions and sale type. A candidate that was bank-owned, carried a large seller credit, or was confirmed as a sale between relatives (by the transfer documents or the agent on the sale, not assumed from a shared surname) is set aside, because its price carries something other than the feature.

A worked pair: the two-car garage. In one subdivision, two colonials of the same style, 1,850 and 1,880 square feet, both in good condition, went under contract within a few weeks of each other. One has no garage and sold for $450,000. One has a two-car attached garage and sold for $470,000. The 30-square-foot difference is not adjusted, because a difference that small is too small to quantify within the pair and does not change what the pair shows. The market paid $20,000 for the garage. Where the subject has a two-car garage and a comparable has none, the comparable is adjusted upward by $20,000.

A worked pair: the second full bath. Two nearly identical capes in the same market area: one with two full baths sold for $440,000; one with a single full bath sold for $410,000. The market paid $30,000 for the second bath. Where the subject has two full baths and a comparable has one, the comparable is adjusted upward by $30,000.

A worked pair with a correction. Two ranches in the same market area, one 1,400 square feet with an updated kitchen that sold for $392,000, one 1,520 square feet with an original kitchen that sold for $378,000. The size difference favors the second house by 120 square feet; at the segment’s living-area rate of $55 per square foot, that is $6,600. Correcting for size, the second house indicates $371,400 for a 1,400-square-foot ranch with an original kitchen, and the market paid roughly $20,600 for the updated kitchen. The figure is approximate, and it is far more useful than the $45,000 the kitchen cost.

Bringing both sales to the same date

Two sales that went under contract months apart are corrected to a common date before their prices are compared; otherwise the market’s movement between them is read as the feature. In a segment rising about 0.4 percent a month, a colonial without a garage that went under contract in February at $440,000 and an otherwise matching colonial with a two-car garage that went under contract in July at $468,000 are five months apart. The February sale, brought forward five months at 0.4 percent a month, or 2 percent, indicates about $448,800 in July’s market. The garage accounts for about $19,200, not $28,000; the remaining $8,800 is the market. The contract date governs because the price was agreed on that date. The contract date is the date the contract was signed by both the buyer and the seller. The listing agent records that date in the contract date field of the MLS record; the date the listing’s status was changed to pending is not the same thing, because the status change can be made days later. Where the MLS record does not show a contract date, the closing date is used and the analysis notes that it was.

Several pairs, one figure

A single pair is one reading. The figure used in a grid comes from several, set side by side after each has been corrected for time and for any size difference.

Three of the four pairs fall between $17,800 and $21,500. The fourth stands apart, and the record explains it: a permit search showed permits for a new roof and heating system on the house with the garage in that pair, work its listing did not describe, so the pair likely measures the garage and the systems together. It is set aside with that reason written down. The three remaining pairs average $19,500 and support a two-car garage adjustment of about $20,000 in this segment. The figure is rounded because the evidence supports an amount in that vicinity, not an amount to the dollar.

A location pair

Location differences are measured the same way, and they are often larger than feature adjustments. In one market area, two 1,500-square-foot capes in similar condition, under contract within six weeks of each other, sold for $362,000 on a state route carrying steady through traffic and $391,000 on a residential street three blocks away. A second pair, two raised ranches of 1,700 and 1,720 square feet, sold for $398,000 on the same route and $425,000 on a side street; the 20-square-foot difference in that pair is too small to quantify within the pair and is not adjusted. The pairs indicate $29,000 and $27,000. The discount for frontage on the route is about $28,000 in this tier, and where the subject fronts on the route, a comparable on a residential street is adjusted downward by that amount.

When a feature travels with other differences

Some features arrive bundled. In many market areas the houses with three-car garages are also the larger houses on the larger lots, and the houses with in-ground pools are often the ones on level, private sites. A pair that differs in the feature also differs in what comes with it, and the price difference measures the bundle. Each other difference is removed first at its own market-derived figure, and the residual is read as the feature. Where the bundle cannot be separated, the pair is used for what it measures, for example the difference between a level, private half-acre with a pool and a sloped lot without one, and it is described that way in the analysis.

Pairs in other settings

Condominium pairs are often the cleanest available, because units in one complex share a site, a builder and an association; two matching second-floor units, one with a deeded garage space and one without, measure the garage space directly. Acreage markets produce fewer pairs, and the search goes back in time, widens within the town, or both, toward sales that a buyer of the subject would have considered. There is no fixed order; the direction is the one that yields that kind of sale. New-construction pairs come from closed sales in the same development, where a lot premium for a wooded rear line or a cul-de-sac position appears as the difference between the same model on different lots, once the options included in each sale are accounted for.

How many pairs are enough

Three or more pairs that agree within a few thousand dollars support a figure well. Two pairs that agree support it reasonably, and the analysis says that the figure rests on two pairs. A single pair is a reading, not a rate; it is used with that limitation stated and checked against a second kind of evidence, such as a group comparison of the segment’s sales with and without the feature, or a sensitivity test showing how much the range depends on the figure. Where the pairs disagree widely and the record does not explain why, the market has not settled on a figure for that feature in that segment, and the difference is better handled by choosing comparables on both sides of the subject for it than by an averaged number.

What a wrong figure does to the range

The subject has a two-car garage, and two of the three comparables have none. With the garage measured at $20,000, the three comparables indicate $466,000, $472,000 and $469,000 after all adjustments. Had the garage been entered at $35,000, the cost a contractor quoted for building one, the two comparables without garages would indicate $481,000 and $487,000, and the range would run from $469,000 to $487,000 instead of $466,000 to $472,000. One unsupported figure, applied consistently, lifts most of the range by $15,000 and widens it, and every other adjustment in the grid can be correct while that happens.

Explaining a pair to a seller

A pair is also the clearest way to answer a seller who expects a feature to be priced at what it cost. A seller who spent $60,000 on a kitchen two years ago can be shown two sales in the neighborhood: a cape with a kitchen of similar quality and a matching cape with the original kitchen, under contract a month apart, $24,000 apart in price. The pair shows what buyers in that neighborhood paid for the difference. It does not tell the seller the renovation was a poor decision; it shows how the market weighs a kitchen against the rest of the house, and it places the seller’s kitchen in the price range on the market evidence the neighborhood’s sales provide.

Every figure derived this way is specific to the market, the price tier and the period. A garage in a $400,000 subdivision and a garage in a $1,200,000 neighborhood are not the same adjustment. In a moving market the pairs are re-run each season, because the market’s payment for a feature moves with the market.

How it applies. A grid with a $10,000 garage adjustment on it either has a pair behind that figure or it does not. The pair is recorded beside the figure: the feature measured, the two addresses, the contract and closing dates, the two prices, each correction made with its amount, and the indicated figure, along with a line of explanation for any pair set aside. Without that record, the figure is likely to have been carried over from somewhere else. An office that keeps its pairs for a market area in one place has, within a year, measured figures for garages, baths, finished basements, location factors and condition, each with the sales behind it.