Down 13.1% or Down 2%: Two Numbers for the Same August

APCIQ published August figures on September 4, 2026. The Montreal census metropolitan area recorded 2,853 sales. The same month is being described two ways: down 13.1% from a year ago, and down 2% from July on a seasonally adjusted basis. Both numbers describe those same 2,853 transactions. For the monetary backdrop to this period, see our piece on tariffs, supply shocks and the policy rate.

One month, two benchmarks

A percentage change is never a property of the month it describes. It is the relationship between that month and another one. Change the other one and the percentage changes, without a single transaction having moved. That is the entire explanation for the gap between 13.1 and 2, and there is no other.

The first figure sets August 2026 against August 2025. The second sets August 2026 against July 2026, after a statistical treatment we unpack below. Neither is more correct than the other. They answer two separate questions, and the common mistake is to take one answer for the other.

What the 13.1% decline measures

A year-over-year comparison answers a question about level: where are we relative to this time last year. It carries a structural advantage worth naming, because it explains why it is the most quoted measure. By comparing August with August, it lines up two months that occupy the same position in the year, with the same vacation pattern, the same back-to-school dynamic, the same listing habits. The seasonal effect cancels itself out, with no processing required.

Its drawback mirrors that advantage. It is slow. A full year separates the two points, and everything that happened in between is compressed into a single number. A market that fell sharply early in the period and has since stabilised produces exactly the same annual change as one sliding steadily all year. Those two situations do not call for the same decision.

What the 2% seasonally adjusted decline measures

Comparing a month with the one before it answers a question about movement: is this speeding up or slowing down, right now. It is the freshest reading available, and that is where its value lies.

It is also unusable raw. Comparing August's sale count directly with July's would blend two different things: what changed in the market, and what changes between those two months every year regardless. That is exactly what seasonal adjustment exists to correct.

What the adjustment corrects

Housing activity follows a recurring rhythm through the year. That rhythm is not an opinion; it shows up in the series year after year. The statistical treatment estimates that regular profile from history, then removes it from the observed series. What remains is the part not explained by where the month sits in the calendar.

The goal is to make two consecutive months comparable, which they are not naturally. That is a legitimate and useful operation. We are not putting a number on the seasonal factor applied to August here: that factor belongs to the methodology of the organisation publishing the series, and inventing one would lend false precision to an explanation that does not need it.

What it leaves untouched

This is where the measure is most often misused. Seasonal adjustment removes a calendar effect and nothing else. It does not correct interest rate levels, household borrowing capacity, the stock of active listings, or the broader economy. A seasonally adjusted decline of 2% is still a 2% decline attributable to the market itself: the treatment did not manufacture it, it isolated it.

Two limits are worth knowing. First, a seasonally adjusted figure is not an observed figure. It is processed data whose result depends on an estimated profile, and that profile can be revised as the series lengthens. Second, over a single month the magnitude stays small and the room for interpretation is wide. A monthly change earns its meaning inside a run of monthly changes, not on its own.

Which one fits your decision

A seller setting a price compares their property with what has recently sold nearby. That is a question about level, and the annual comparison gives the right marker: the market is absorbing materially fewer transactions than a year ago, and that weighs on how long it will take to find a buyer.

A buyer weighing whether to wait is asking about movement. The seasonally adjusted monthly series is what informs that, and a 2% decline describes neither a collapse nor a turn: it is a market still drifting gently lower. The two figures sit together without contradiction. A level well below last year, and a small monthly move, describe the very same August.