« Federal Budgetary Comparisons: Canada and the United States | Main | Real Interest on Reserves »


Feed You can follow this conversation by subscribing to the comment feed for this post.

This analysis raises some interesting points. I wonder if it would remain consistent should we look at the net additions (completions - loss (demolition, conversion, etc))of period t compared to population growth in period t+1. Or perhaps backward as a function of demand pressures that have been observed.

As starts have no guarantee to be added to the market until they are completed, and not all starts end up in completions, they might not be a robust indicator.

Adjusting for population is, of course, necessary, but the correct adjustment is for households. The number of persons per household has also changed. With rising age of population and marriage, and smaller families, the population today may represent a larger number of households.

Creating even a very rough approximation of the household number time-series would be a great update to this post. Although, I am unfamiliar with the data, for recent years this number may be available. A linear transition from an approximation to 1868 numbers would be sufficient, but demographic pyramids for 1868 would be key.

Addendum to comment on households: The link to Statistics Canada show a chart of the relevant household numbers from 1851. The linearity assumption was indeed a good first approximation:

The shift to smaller households over the past century

Those are good points Tyler & David. I'm not sure if I have come across any data linking starts to completions at least not over a 150 year span. Having completions might also provide a way to estimate a stock of housing variable going back to Confederation. I suppose one could get an estimate of the housing stock from the 1871 census and work forward from that. The point about using households is also a good one and would provide an additional way of presenting the data. One does wonder if the number of households has been growing faster than population in recent decades.

A friend of mine who was a real estate lawyer in Toronto told me that there was evidence of falling household size at least in Toronto - at the time the number of households in the old city of Toronto had grown significantly in the previous 20 years while the population had not (though I expect that latter trend would no longer be true. It makes sense, between aging populations and the decline of marriage, that households would grow faster than population.

Out of curiosity, did anyone actually bother to look at the link in my previous comment? As both Livio and commenter Bob seem to muse about household numbers while the link provides Statistics Canada quantitative data on this. I'll save you the trouble, the people per household went from 3.5ppl to 2.5ppl since 1970 (a 29% decline) and it was 10 ppl/household a 100 years ago. The household growth rate is consistently 2%-4% higher than population growth rate.


I believe you are correct that the data does not go that far back, and regardless the methodology has likely changed multiple times. CMHC does provide data back to 1991 as seen here:


Though it would be worth noting that due to the long-horizon of housing construction, creating a lagged variable between starts and completions could be difficult.

CMHC does appear to have data that goes back to at least 1962 as found here:


Though it does not seem yet to be on their website.

The use of net additions with David`s approach to HH size could prove interesting in further analysis.

Thanks David. Given there are fewer people per household and the number of households is growing faster than population than a dwelling starts per household series would likely show we are adding even less to housing supply. At some point I would like to update the post but it is March...

What do we mean by 'housing starts'. For example, condo construction has increased over the last few decades, and is this considered 'one start', or is each unit in a condo a separate 'start' here?

@kapil: A 100 unit condo project would be 100 housing/dwelling starts.

The comments to this entry are closed.

Search this site

  • Google

Blog powered by Typepad