GLA adjustment per square foot

It is not price per square foot

A $500,000 sale of a 2,300 sq ft home has an average price of about $217 per square foot. That number includes the land, the site improvements, the garage, the kitchen, and everything else. It is not what the market pays for one additional square foot of living area, which is the only figure a GLA adjustment should reflect. Contributory value at the margin is typically a fraction of average price per square foot, and using the average produces adjustments that are far too large.

Deriving the marginal rate

Two defensible routes, both starting from closed sales in the subject's competitive market area over a bounded time window:

1. Paired sales. Find sales alike in every material respect except size, divide the price difference by the GLA difference, and repeat across several pairs.
2. Regression. Fit sale price against GLA and the other characteristics that matter locally; the GLA coefficient is the marginal rate with those other elements held constant.

Reconcile the two. Where they agree, the rate is well supported. Where they disagree, the disagreement usually points at a data problem worth finding before the report goes out.

Applying it in the grid

The adjustment is the GLA difference measured toward the subject, multiplied by the rate. With a subject of 2,180 sq ft, a comparable of 2,310 sq ft, and a derived rate of $45 per square foot:

(2,180 − 2,310) × $45 = -$5,850

The comparable is larger than the subject, so the adjustment is negative: the comparable is superior in this element and its sale price is adjusted downward. Measure GLA consistently — above-grade finished area only, per ANSI or the convention your client requires — because a rate derived one way and applied another is not supported at all.

Documenting the support

State where the rate came from, over what sales and what period, and why it is reasonable for this market — not just the number in the grid. Try the adjustment calculator with your own rate, or read how regression derives it.