Summary
A falling market thins your data at exactly the moment it asks more of you. Lead with the indicators that turn first. Be clear that a negative adjustment measures what has happened, not what is coming. And when the trend reversed inside your window, don't average across the turn. Measure the two stretches separately and use the current one.
Why a negative adjustment is harder to support
The math is the same in either direction, but everything around it changes.
Volume drops in a falling market, so you have fewer sales to work with. Sellers withdraw rather than take less, which removes the very transactions that would have shown the decline and leaves what remains looking stronger than the market is. Concessions rise, so a closed price increasingly overstates what the seller actually netted. And today's closings were contracted in a market that no longer exists, so the lag between contract and settlement goes from a nuisance to your biggest source of error.
Put together, the data understates the decline during the exact period everyone is asking about it.
Read the indicators that turn first
Closed prices move last. By the time the median has fallen, the turn is months old. The leading indicators matter more here than in any other market.
- Sale to list ratio. Falls as soon as buyers start negotiating again, well ahead of closed prices.
- Days on market. Rises first. It rises for the properties that eventually sell, which understates the change, because it leaves out the ones that never sold at all.
- Months of supply. The clearest single measure of the balance shifting, and it moves on listings rather than settlements.
- Price reductions. How many active listings have cut price, and by how much. Often the earliest signal you can see.
- Withdrawn and expired listings. A rising share means your closed sales are being filtered by what sellers were willing to accept, not by value alone.
None of these becomes the adjustment. They establish that a decline is real and roughly when it began, which tells you what period to measure the closed sales over.
Concessions hide a decline inside a flat price
When a market softens, the terms usually give before the price does. A seller who won't cut fifteen thousand off the list price will pay fifteen thousand toward the buyer's costs, and the sale records at full price. Measure the trend on closed prices alone and you'll read that market as flat while effective prices are falling.
If your data reports concessions, run the trend on the concession-adjusted price. If it does not, look at how common and how large concessions were over the period, and say plainly that your measured trend is a floor rather than a precise figure.
When the market turned inside your window
This is the case that produces the worst adjustments, because the usual method fails without any warning sign. Fit one rate across a period that rose for eight months and fell for four, and you get a small positive number that describes neither half and adjusts every comparable wrongly.
Do this instead.
- Find the turn using the leading indicators rather than closed prices. They will put it earlier and show it more clearly.
- Measure the two stretches separately. Two rates over two periods, each with its date range stated.
- Apply the stretch each comparable belongs to, or chain the periods into an index so a comparable that spans the turn is adjusted through both. An index handles this far better than a single monthly percentage, and this is one of the few times the extra step clearly pays for itself.
- Weight toward the current stretch. A comparable contracted before the turn describes a market that has since changed, and it deserves less weight in reconciliation even after adjustment.
Say what it is and what it isn't
A negative market conditions adjustment measures change that has already happened and shows up in the data. It isn't a projection. Describe it as one and you've invited a challenge you can't win, because a forecast isn't something an appraisal can support.
The wording that survives review stays narrow and factual. This data set, over this period, moved this much, measured this way, and here is the range. What the market will do next has no place in that sentence.
The range test matters more here
In a thin market it is tempting to widen the geography or the time window until enough sales turn up. That trade isn't free. A set assembled to reach a count may no longer be one market, and a trend fitted to it is describing a mixture.
So check whether your adjusted indications tightened. If a decline you're confident about does not bring the comparables into better agreement, the likelier explanation is that your set covers two segments, not that the decline is wrong. Reporting a wider supported range with an explanation is a stronger position than a precise number the data can't carry. It's the same reasoning behind a documented zero.
Common questions
Should I apply a negative adjustment if only the leading indicators have turned?
Not as a measured rate. Closed sales haven't moved yet, so there is nothing to measure. What the leading indicators support is a discussion of market conditions and, where the evidence is strong, more weight on your most recent comparables. Adjusting closed prices for a decline that hasn't reached closed prices is forecasting.
How do I handle a comparable that closed above the current market?
Adjust it for time from its contract date like any other, then see whether it still sits outside the adjusted range. If it does, the real question is whether it belongs in the set at all. An outlier is a candidate for exclusion with a stated reason, not for a bigger adjustment that drags it into line.
Does a declining market change how the physical adjustments are derived?
The methods stay the same, but the data gets thinner and pairs get harder to find, so the supportable ranges widen. Deriving physical rates from a period that spans the turn also mixes two markets, so time-adjust your sales before you extract physical rates from them.
