National Average vs Local Market: Two Scales

CMHC expects a national average price of 675,200 dollars in 2026, down 0.6%. The same summer, all three median prices in the Montreal metropolitan area rise. Two figures moving in opposite directions, and yet not in contradiction. This article explains why, without reprising the volume-and-price mechanism covered in our piece on the revised 2026 forecast.

Two figures moving opposite ways in the same summer

On one side, CMHC's revised forecast of July 22, 2026: 457,200 sales across Canada in 2026, down 2.8% from 470,314 in 2025; a national average price of 675,200 dollars, down 0.6% from 679,543 dollars; 241,400 housing starts, down 6.8% from 259,028.

On the other, QPAREB's July 2026 statistics for the Montreal metropolitan area: median prices rose across all three segments, single-family homes by 4%, plexes by 6%, condominiums by 2%.

One price falling, several prices rising, published weeks apart. A quick read sees a contradiction and asks who is right. That is the wrong question: both statements are accurate, and they are not about the same thing.

What a national average gathers, and what it erases

It gathers markets whose only shared trait is sitting inside the same country. Regions rising and regions falling, large cities and rural municipalities, segments whose dynamics have no reason to coincide. The number that emerges is the result of that addition, and it resembles none of its components.

What it erases is precisely the spread that matters to someone buying or selling. A national average will never report that one segment of one region is rising while another falls elsewhere. It is built not to. That is not a flaw, it is what a national aggregate is for.

The practical consequence is straightforward. A national average tracks an overall direction over several years and compares a country to itself across time. It does not locate a property, and using it that way produces false conclusions from accurate figures.

An annual forecast and a monthly measurement do not compare

This is the second gap between the two figures, and it weighs as much as the first. CMHC's numbers are an expectation for the whole of 2026, produced in July, therefore covering in large part months that have not yet happened. QPAREB's numbers measure what happened in July 2026.

A forecast and a measurement share neither standing nor shelf life. The first gets revised, and this one was in fact revised downward against the February forecast. The second is not revised in the same sense: it describes a closed month.

It also bears repeating that a revised forecast is not a market event. It updates an expectation. It changes no transaction, no asking price, no rate. It comes from a body that observes rather than an actor that operates.

What July says about Quebec itself

CMHC states that Quebec should post more modest gains in a balanced market. That is a qualitative formulation covering a full year, and we draw no figure from it.

What July 2026 supplies as measurement, for the Montreal metropolitan area, is another kind of statement: 3,338 sales against 3,709 in July 2025, a fifth consecutive month of decline and the sharpest drop since February 2026; 19,790 active listings, up 17% year over year and 9% above the historical average for a month of July.

The two sets complete each other rather than settle each other. One gives an annual, national frame; the other a monthly, regional snapshot. Neither validates or invalidates the other, because neither measures what the other measures.

At what level a decision is made

Neither nationally nor regionally. At the level of recent comparable sales in your sector, for your property type. That is the set a buyer measures yours against, and the only one where a price is set.

Higher scales keep real value, provided they are given the right role. They explain why the context is shifting, they anticipate a general direction, they keep you from being surprised by a movement. They never answer the how-much question for a specific property.

The limits of this article

It compares no dollar amount across the two sources, deliberately: the regional median prices cited here are given as variations, not in dollars, and setting them against a national average in dollars would produce a gap that measures nothing.

It also does not address the mechanism linking sales volume to price levels, a subject covered elsewhere on this blog. And it takes no position on whether CMHC's forecast will prove right, which will be judged on data nobody holds today.