Mix Effect: When the Median Price Moves and No Property Does

A median price that climbs three points does not say properties are worth three points more. It says the middle transaction, among those that took place that month, sits three points higher. Those are not the same claim, and the gap between them has a name: the mix effect. For the wider provincial picture, see our read of the Montreal single-family segment in July 2026.

What a median actually measures

A median ranks the transactions of a period and keeps the middle one. It weights nothing, corrects nothing, and compares no property to itself over time. It describes a basket: whatever happened to sell. Change the composition of that basket without touching a single price, and the median moves anyway. That is a property of the indicator, not a flaw in the data.

The minimal example takes a few lines. Picture a market selling condos at $400,000 and single-family homes at $600,000 in equal numbers, with no price moving from one month to the next. If condo sales halve the following month while single-family sales hold, the overall median climbs. No buyer paid more for a given property. No seller received more. The statistic still went up.

Two rates of decline in July 2026

The QPAREB release of August 6, 2026 supplies what makes the effect visible province-wide. In July 2026 Quebec recorded 7,407 sales, down 6% year over year. Single-family homes account for 5,141 of them, down only 1%. Those two rates of decline are not the same, and that alone is enough to shift the composition of the basket.

Consider what that implies arithmetically. If the whole retreats 6% while one of its segments retreats only 1%, that segment's share of the total necessarily rises. The rest of the market, taken together, is therefore falling faster than 6%. The provincial median is computed on a basket whose component weights differ between the two periods being compared.

Why this is not the price-volume divergence

There is a separate phenomenon often confused with this one: prices rising while sales fall. That one stems from the relationship between available supply and creditworthy demand, and it implies genuine pressure on values. The mix effect implies nothing of the sort. It moves a median at strictly constant values. You can therefore observe it in a perfectly flat market, which is not true of the price-volume divergence.

The distinction matters because the two call for opposite readings. A price-volume divergence tells you something about market conditions. A mix effect tells you nothing whatsoever about values: you have to remove it before you can hope to see anything else. Treating them alike means reading noise as signal.

Geographic recomposition, harder to spot

The effect does not operate only across property types. Holding the segment constant, it also operates across districts. A regional median aggregates neighbourhoods whose price levels sometimes differ twofold. If sales drift month to month toward the more expensive districts, or away from them, the regional median follows without any neighbourhood moving. That version is harder to catch, because the label on the statistic does not change: it is still the same region and the same property type.

How to strip the effect out

The method is to drop one level down whenever a doubt appears. Read the median by segment rather than overall. Then, inside the segment, by district. A median computed on a homogeneous set can no longer move through recomposition, only through a real price change. It is the same reasoning that says to band a sample by floor area before extracting a price per square foot.

One simple control goes with that reading: check whether the subset's transaction count stays comparable across periods. When it collapses, the subset median becomes unstable in its own right, and it is better to publish the individual values than to present a statistic the sample no longer supports.

What it changes for a seller

A regional median does not set a price. It frames a context, and only when read at the right level of aggregation. What sets a price is sales comparable to the specific property, in the specific district, over a recent window. A seller who learns the regional median rose a few points has learned nothing about their home. Nor has a buyer who learns the opposite. In both cases the useful information sits one or two levels further down.