Canada’s sharp downward trend in fertility rates has appeared in this space before. For many reasons I see this as a looming crisis. Here, to refresh your memory, is how Canadian fertility rates per female have looked since 1991: The Audit is a reader-supported publication.
As Greg, Ross and Steve indicate, we have here a severe case of "missing variables." Simple correlation coefficients can be very misleading then. In a more complete model your intuition may very well turn out to be true.
The sharp changes in fertility rates over 1991-2023, up and down, would seem to be the first matter to be explained.
Great post. In Vancouver, where housing is super-expensive, Jens von Bergmann has drilled down to census-tract data. It turns out that the number of children (aged 0-15) is falling in low-density neighbourhoods, and rising in high-density neighbourhoods. In Vancouver, families can afford to live in multifamily housing, not expensive single-detached houses. https://morehousing.ca/children
Canada is in a transition where we no longer have affordable homes in our major cities. And in some cities it is extremely unaffordable, where two professional incomes won’t let you comfortably buy anything but a small condo without financial help.
But for some young couples it is very important to have kids, do they will do so even if living in a small condo because having kids is that important to them.
It makes total sense in that context that the statistics show no meaningful correlation.
Biologists have noted that in natural systems there are tipping points where reproduction falls. Food security, predation stresses, etc.
Taking that and looking at Canadians aged 25-40, many lack security in housing and employment, so there is that pressure on having kids. And then there is the doom about the future of the planet which isn’t really an encouraging topic for the youth today.
To focus on high rises vs homes seems to make as the forest for the trees.
I had never seriously considered the relationship that you posited. The fact that you have not (yet?) been able to see a correlation is irrelevant. What is relevant is that you thought of it and explored it.
I also would have thought that the relationship was as you thought. Anyway, keep up the enquiring mind.
Excellent essay David. By coincidence I have been wading through similar data for a related project. Using Cansim 34-10-0126-01 you can compute the % of new builds that are apartments by province over time and compare to provincial fertility rates. There are hints of a downward effect when the apartment fraction rises above 0.3 or 0.4 of total new builds. But it's noisy. One thing I haven't done is normalize fertility by the fraction in the eligible age range (around 18-40). BC's birth rate is catastrophically low (1.0) but that may partly be due to all the retirees moving from the prairies to Victoria.
A bigger problem is sorting out dual causality: if X causes Y and Y causes X a regression Y=a+bX yields biased and misleading slope coefficients. Like regressing crime rates on police spending, you could easily conclude more police cause more crime because you get the causality backwards. In this case developers might choose to build more apartments in regions where fertility rates are low (including, for example, areas with growing retiree populations). What we need is a way to identify future fertility intentions of young family-aged Canadians and regress that against housing stock characteristics. I think your intuition would be borne out but I don't think Statscan has data on fertility intentions.
"I identified the percentage of apartments in buildings that have five or more storeys and the average family size of couple-with-children economic families for each of Canada’s electoral districts."
At first glance, this doesn't seem like a particularly helpful comparison. Average density doesn't tell us much when families are likely to sort themselves into any available and affordable low density housing rather than staying in high density housing. I agree with the premise that a family can't/won't want to raise multiple children in a high rise if they can avoid it, but that comparison doesn't tell me whether they can avoid it or not.
Would they really need to move to another district just to find more family friendly housing than that? Is all the more family friendly housing at full occupancy with too little supply for demand so they can't move into it? Are high rises the sort of housing in Canada routinely filled with people who both want children and are too poor to move to more family friendly housing?
Just thinking aloud... If you're looking for possible impacts from housing on decisions to have children or not, it might be better to look at something like the average percentage of household income being used for housing (rent/mortgage) based on the housing type and overall economic bracket of the household (preferably also drilled down to filter on just households in the prime fertile age range so your data isn't distorted by University dorms, retirement communities, and other places generally too young or old to be considering children). That might show if housing costs are suppressing children by making it unaffordable for families to increase their living space as needed to add children (either if they can't afford to move at all or if the move itself would increase their housing costs too much to afford the other added expenses of adding a child).
Or, to put it another way, if we regard having a larger family as a luxury good of sorts, then the question becomes how much, both in percentage and absolute terms, of each household's "luxury" budget is available to afford children? What non-luxury budget items have been cutting into the luxury budget the most that government can efficiently and effectively impact? It wouldn't surprise me if housing costs, especially for family friendly housing, have been increasing as a percentage of household budgets for some time, raising the floor of fixed costs in household budgets out of reach for many, but I don't know the numbers for Canada.
Interesting ideas. I'm not sure I'd be able to drill down that deep using the census dataset. I may try to identify other data sources and see what comes out.
I confess that I instinctively assumed that the birth rate in high-density housing (aka ‘The Projects’ or, typically, ghettos) would tend to be higher than in neighbourhoods of single-family dwellings and so I am pleasantly surprised by the results of your analysis. But what to do about the dropping birth rate? We’ve seen what over-enthusiastic immigration has done to the economy and the nation’s social infrastructure.
Every Western country (plus China) has been struggling with this question for years now. Perhaps, as I mentioned, the key part of the solution is about changing attitudes rather than economics.
As Greg, Ross and Steve indicate, we have here a severe case of "missing variables." Simple correlation coefficients can be very misleading then. In a more complete model your intuition may very well turn out to be true.
The sharp changes in fertility rates over 1991-2023, up and down, would seem to be the first matter to be explained.
Great post. In Vancouver, where housing is super-expensive, Jens von Bergmann has drilled down to census-tract data. It turns out that the number of children (aged 0-15) is falling in low-density neighbourhoods, and rising in high-density neighbourhoods. In Vancouver, families can afford to live in multifamily housing, not expensive single-detached houses. https://morehousing.ca/children
You don't need to tell me about these problems, I live in Toronto. My older kids simply gave up and left the country.
Canada is in a transition where we no longer have affordable homes in our major cities. And in some cities it is extremely unaffordable, where two professional incomes won’t let you comfortably buy anything but a small condo without financial help.
But for some young couples it is very important to have kids, do they will do so even if living in a small condo because having kids is that important to them.
It makes total sense in that context that the statistics show no meaningful correlation.
Biologists have noted that in natural systems there are tipping points where reproduction falls. Food security, predation stresses, etc.
Taking that and looking at Canadians aged 25-40, many lack security in housing and employment, so there is that pressure on having kids. And then there is the doom about the future of the planet which isn’t really an encouraging topic for the youth today.
To focus on high rises vs homes seems to make as the forest for the trees.
Fascinating thesis, Sir!
I had never seriously considered the relationship that you posited. The fact that you have not (yet?) been able to see a correlation is irrelevant. What is relevant is that you thought of it and explored it.
I also would have thought that the relationship was as you thought. Anyway, keep up the enquiring mind.
Excellent essay David. By coincidence I have been wading through similar data for a related project. Using Cansim 34-10-0126-01 you can compute the % of new builds that are apartments by province over time and compare to provincial fertility rates. There are hints of a downward effect when the apartment fraction rises above 0.3 or 0.4 of total new builds. But it's noisy. One thing I haven't done is normalize fertility by the fraction in the eligible age range (around 18-40). BC's birth rate is catastrophically low (1.0) but that may partly be due to all the retirees moving from the prairies to Victoria.
A bigger problem is sorting out dual causality: if X causes Y and Y causes X a regression Y=a+bX yields biased and misleading slope coefficients. Like regressing crime rates on police spending, you could easily conclude more police cause more crime because you get the causality backwards. In this case developers might choose to build more apartments in regions where fertility rates are low (including, for example, areas with growing retiree populations). What we need is a way to identify future fertility intentions of young family-aged Canadians and regress that against housing stock characteristics. I think your intuition would be borne out but I don't think Statscan has data on fertility intentions.
Thanks for those thoughts. I've obviously got a lot more work to do on this general topic.
An important first step might be finding a good proxy for fertility intentions - tubal ligation rates might be one possibility.
I have a feeling I'll be back for more sooner rather than later.
"I identified the percentage of apartments in buildings that have five or more storeys and the average family size of couple-with-children economic families for each of Canada’s electoral districts."
At first glance, this doesn't seem like a particularly helpful comparison. Average density doesn't tell us much when families are likely to sort themselves into any available and affordable low density housing rather than staying in high density housing. I agree with the premise that a family can't/won't want to raise multiple children in a high rise if they can avoid it, but that comparison doesn't tell me whether they can avoid it or not.
Would they really need to move to another district just to find more family friendly housing than that? Is all the more family friendly housing at full occupancy with too little supply for demand so they can't move into it? Are high rises the sort of housing in Canada routinely filled with people who both want children and are too poor to move to more family friendly housing?
Just thinking aloud... If you're looking for possible impacts from housing on decisions to have children or not, it might be better to look at something like the average percentage of household income being used for housing (rent/mortgage) based on the housing type and overall economic bracket of the household (preferably also drilled down to filter on just households in the prime fertile age range so your data isn't distorted by University dorms, retirement communities, and other places generally too young or old to be considering children). That might show if housing costs are suppressing children by making it unaffordable for families to increase their living space as needed to add children (either if they can't afford to move at all or if the move itself would increase their housing costs too much to afford the other added expenses of adding a child).
Or, to put it another way, if we regard having a larger family as a luxury good of sorts, then the question becomes how much, both in percentage and absolute terms, of each household's "luxury" budget is available to afford children? What non-luxury budget items have been cutting into the luxury budget the most that government can efficiently and effectively impact? It wouldn't surprise me if housing costs, especially for family friendly housing, have been increasing as a percentage of household budgets for some time, raising the floor of fixed costs in household budgets out of reach for many, but I don't know the numbers for Canada.
Interesting ideas. I'm not sure I'd be able to drill down that deep using the census dataset. I may try to identify other data sources and see what comes out.
I confess that I instinctively assumed that the birth rate in high-density housing (aka ‘The Projects’ or, typically, ghettos) would tend to be higher than in neighbourhoods of single-family dwellings and so I am pleasantly surprised by the results of your analysis. But what to do about the dropping birth rate? We’ve seen what over-enthusiastic immigration has done to the economy and the nation’s social infrastructure.
Every Western country (plus China) has been struggling with this question for years now. Perhaps, as I mentioned, the key part of the solution is about changing attitudes rather than economics.