When a Difference Is Not Adjusted

Over-adjusting introduces error as surely as under-adjusting. An adjustment is made when the market evidence shows that buyers pay for the difference, and its amount is what that evidence supports. Where the evidence shows no price response, there is no adjustment.

No price response. A difference that buyers in the market do not pay for is not adjusted. The tenth of an acre in a subdivision of similar lots, the difference between a 1968 and a 1971 build date, a half-bath on a different floor: where the sales show no response, the analysis records the difference and leaves it unadjusted.

The absence of a response is itself read from the sales. In a subdivision of 1990s colonials, six sales over a year sit on lots between 0.41 and 0.54 acres. Once each price is corrected for time and living area, the six indications show no pattern by lot size; the two largest lots sold in the middle of the group. Buyers in that subdivision are not paying for lot size within that band, and a 0.48-acre subject and a 0.42-acre comparable are compared without an adjustment for the difference. The same six sales, set beside a sale on a 1.5-acre lot at the edge of the subdivision, might show a response that the narrow band does not, and the conclusion applies only to the band the sales cover.

No measurement available. Where a difference plainly matters but the market has produced no pair of sales from which to measure it, the analysis says so and leaves the difference unquantified, rather than assigning a figure that has no support. In that case, comparables are chosen on both sides of the subject for the feature, so that the subject’s position is bounded by the sales rather than calculated.

The size of the figure is not the test. Differences between the subject and the comparable sales are generally adjusted, except when the difference is too small to quantify or the sales show no market reaction to it. Where the segment’s sales show buyers paying $60 a square foot for living area, a comparable 60 square feet smaller than the subject is adjusted upward by $3,600, because that is what the evidence supports. Where the sales show no price response to a difference, as with the lot sizes in the band above, the adjustment is zero, however large the figure would have been. The decision rests on what the market evidence shows about the difference, not on the size of the resulting figure.

Present in every comparable. Where every sale in the analysis has a two-car garage and so does the subject, the garage is already in every price and there is nothing to adjust. The reverse case is different. Where every comparable has a two-car garage and the subject has none, every comparable needs the adjustment, and the sales in hand cannot supply it, because none of them lacks a garage. The figure comes from pairs elsewhere in the market area, or a sale without a garage is added to the set.

Already counted in another adjustment. A difference that another adjustment has already captured is not adjusted a second time. Where the condition adjustment was measured between renovated and original houses, and the renovation included the kitchen, a separate kitchen adjustment counts the kitchen twice. Where a comparable has a fourth bedroom because a family room was partitioned, the living area is unchanged and the adjustment is for the configuration alone; where the fourth bedroom was added as new space, the size adjustment already accounts for much of what the buyer paid, and any further bedroom adjustment is limited to what pairs show the configuration itself to be worth.

Beyond those four cases (no response, no measurement, present in every sale, already counted), the pattern of adjustments on a single comparable is itself information. Several large or material adjustments on one sale are a warning that the comparable may be weaker evidence than another available sale, and they call for reconsidering its selection. No count of adjustments settles the question; what settles it is whether a more similar sale exists. Where one does, even an older one, the relationship between the heavily adjusted sale’s price and the subject is too indirect for its adjusted figure to carry much weight, and the analysis is stronger with the more similar sale in its place. This is the point at which selection and adjustment meet: a comparable that needs much adjustment is a question about its selection.

The subject is a 1,750-square-foot ranch in average condition with a one-car garage and an unfinished basement. Two candidate sales are measured against it with figures from the market area’s pairs.

The two indications are within $2,000 of each other, and they are not equally reliable. Sale Y reached the subject through two small corrections, each supported by the segment’s own data. Sale X reached it through $86,000 of correction on four separate differences, and a modest error in any one of those four figures moves its indication by several thousand dollars. Sale X also differs in style and is well outside the subject’s size band, so it was a weak candidate before any adjustment was made. Sale Y is the better evidence despite its age, and Sale X belongs out of the set.

How it applies. Two patterns in a grid answer to this lesson. A grid dense with small adjustments for differences the market does not pay for has been built to look thorough; a grid in which one sale carries several large adjustments has kept a sale whose selection needs a second look against the more similar sales available. Either pattern is visible at a glance, and either produces a range that is less reliable than it appears.