Down 4%: The Aggregate That Hides Three Opposite Movements

In July 2026 the Quebec City census metropolitan area recorded 709 sales, down 4% year over year. The figure looks unremarkable. It is anything but, because none of the region's three submarkets did that: one rose, two fell, and one of the two fell hard. The deeper mechanism in the sharpest-falling area is handled separately, in our piece on the northern periphery and commuting cost; here the subject is the aggregate number itself.

Three numbers, three directions

The QPAREB release of August 6, 2026 splits the Quebec City metropolitan area into three submarkets. The South Shore is up 11% year over year. The agglomeration is down 6%. The northern periphery is down 17%. The whole comes in at down 4%.

Sit with the first and third of those for a moment. They differ not merely in size but in sign. One describes a market where more properties changed hands than last year, the other a market where markedly fewer did. These are not two degrees of one phenomenon. They are opposite phenomena that occurred in the same month, a few dozen kilometres apart.

What an aggregate does to opposite signs

A total adds. That is its job and also its limit. When its components move the same way, it summarises them faithfully: the sum keeps their shared direction and only the intensity is averaged. When they move opposite ways, addition makes them cancel, and the result stops being a summary and becomes a residue.

The region's minus 4% is therefore not a state of the market. It is what remains after a rise and two declines partly annulled each other. It is arithmetically exact and descriptively empty. It matches the situation of no buyer and no seller in the region, because nobody lives in the average of three submarkets.

A spread in size versus a split in sign

Two situations get conflated readily. In the first, every submarket falls, one by 2% and another by 12%. The aggregate stays usable: it points to a direction all of them share, and the spread in size is a refinement layered on afterward. A hurried reader who stopped at the total would get the direction right and only the degree wrong.

In the second, the signs differ, and the question of whether the market is rising or falling loses its regional answer. Not for want of data, but because the question is malformed: there is not one market here, there are three, and one of them does the opposite of the others. This time the hurried reader gets the direction wrong, which is not an error of degree but of orientation.

The check to run on any aggregate figure

Two questions do it, and they take seconds. What are this total's components, and do they all carry the same sign? If the answer to the second is no, discard the aggregate and substitute the component that covers you. This is not methodological polish, it is the condition for the number to mean anything in your file.

The test reaches well past geography. It applies to property types, to price bands, to time periods. Anywhere a published statistic results from an addition, check that the addition has not neutralised opposing moves before leaning on it.

Two mirror errors, one cause

A seller in the rising submarket who believes they sit in a market down 4% is about to concede, on price or on terms, what nothing in their own area's data justifies. They will pay for that reading in real money, on a single transaction, without ever learning they paid it.

A seller in the sharply falling area making the same read commits the mirror error: they believe their position is roughly a quarter as difficult as it is, price accordingly, and discover the truth through a lengthening time on market. The two errors are symmetric and come from one move, mistaking a residue for a description.

What the regional figure still does

Minus 4% is not to be thrown away, it is to be filed correctly. It measures the region's activity volume accurately, which is what a macroeconomic observer, an administration or a statistical service needs. It also places the region in time: at 709 sales the month remains 19% above the ten-year average, a level reading that suffers no cancellation between submarkets.

What it does not do is ground an individual decision. A private buyer or seller does not transact in a region, they transact in an area, on a street even. Confusing the scale of the measurement with the scale of the decision is the most common error in reading housing statistics, and the costliest, because it is invisible to whoever makes it.