The Trouble with Municipal-level Population Projections | Home: Free Sociology!

Skip to content

Co-authored by Jens von Bergmann and cross-posted at MountainMath

Are people liquids or solids?

Trick question: they’re kind of both. This matters in terms of how we track people and project their location forward in time. There are basic demographic methods that effectively take people as solids. We can see where they are now. We can see how they’ve been moving recently. We can age them forward in time, including adding new little people and imagining older people dying off. And we can project forward how many people we’ll have in the future.

But people are also liquid. They slosh around a bit, but they eventually tend to settle downhill into the places where there are containers for them. Here our best bet in terms of projecting people’s location forward in time is to figure out the lay of the land and where the most likely containers are going to be located.

Sometimes our liquid and solid projections match up ok. But other times they don’t. Let’s make this discussion a little more solid by zooming in to take a look at a potential divergence in projections right here in Metro Vancouver.

When – if ever – will suburban Surrey surpass the population of the City of Vancouver?

The solid answer to this question tends to be that it’s going to happen, and quite soon! The liquid answer is much less certain on this point, leaving it mostly contingent on how much housing gets built where.

Solid demographic models

We can see the answer to when Surrey will surpass the population of Vancouver in recently released StatCan population projections at the municipal (census subdivision) level. StatCan’s method, in a nutshell, treats people as solid, aging them forward in place (within CSDs) via a crude approximation from past trends and made to sum into earlier calculated projections across the broader region (CD). Past net migration enters into past trends by age group and becomes extrapolated into the future. In trying to avoid running into absurd scenarios like negative population or unrealistically large growth scenarios in select areas and subgroups they crudely extrapolate areas and age groups that historically have been growing linearly and those that have been declining exponentially (while using language that would make any high school math teacher tear their hair out, but never mind that). More holistically this could be dealt with by a mechanism that implements regression to the mean over time.

We can see a slightly different answer to when Surrey will surpass the population of Vancouver in more standard provincial projections released by BC Stats in British Columbia. BC Stats also treats people as solids, but applies more refined methods to projecting age cohorts forward based upon distinct components of change (fertility, mortality, and migration) estimated from the past with some adjustments for current policies and future expectations, especially concerning immigration.

Figure 1 shows both of these projections, for BC Stats, the cross-over where Surrey surpasses Vancouver is imminent, happening sometime just after the Census, by the end of 2026. For StatCan, it looks like it’s one year further away, happening by the end of 2027.

Figure 1: Population Estimates and Projections for Vancouver and Surrey

We can see the cross-over between Vancouver and Surrey is heavily driven by the projected decline in Vancouver’s population. This is striking. Where does it come from? As it turns out, temporary residents arriving in the metro area of Vancouver have more often landed in the City of Vancouver than the City of Surrey. Recent rollbacks in temporary residents due to Canadian immigration policy have been extrapolated forward, and have been built into the solid population models of both StatCan and BC Stats, where they enter in just slightly different ways. The assumption is that if you turn off this conveyor belt of temporary residents, then the City of Vancouver’s population will decline more than the City of Surrey’s. But is this correct? We could work with different assumptions. Let’s think about our conveyor belt as more like a stream, one of many feeding the pond of our metropolitan area. What happens to the deepest part of the pond – our central city – if this particular stream gets diverted?

Liquid demography

Where will people sloshing around Metro Vancouver land? As it turns out, they land mostly where there’s housing to contain them. If we believe this, then projecting in the future should be based mostly on where housing is likely to land rather than our cohort projections. To make this more concrete, let’s think about housing growth.

Overall, both Vancouver and Surrey have been adding housing in recent years. To better match housing to population we would want to look more broadly at housing services provided, not just number of housing units. We don’t have open data on square footage or number...