Economists making spectacularly bad forecasts, 1961 edition.

Back in the 1960s, the Ontario government's Department of Economics dutifully cranked out annual population forecasts. What is remarkable about these forecasts is how far wrong they were. The economists completely failed to predict the demographic changes that were about to hit Ontario.

Here are the 1961 projections side by side with the actual 1976 population numbers for Ontario, downloaded from Statistics Canada's CANSIM table 051-0001. (An image taken from the Ontario Economic Survey can be found at the end of the post):

The Ontario Department of Economics projections were most spectacularly wrong for children under 10. The economists failed to anticipate the end of the baby boom. They also, it seems, underestimated the improvements in life expectancy that would be enjoyed by older Canadians, and the amount of migration into Ontario.

Now one could argue that the early 1960s collapse in birth rates was triggered by something completely outside the economists' demographic models - a unforeseen and unforeseeable technological change. 

Yet the fall in birth rates was predictable, even in 1961. With the brief exception of the 1950s, Canadian birth rates had been falling for most of the previous century - see the picture below borrowed from here:

Indeed, by 1961 the number of births in Canada had already declined from its 1959 peak. The birth control pill had been in the market since 1957 - although not marketed as such - and had already been approved for contraceptive use in the US. Second-wave feminism was beginning. For example, US President John F Kennedy signed an Executive Order calling for a Presidential Commission on the Status of Women in 1961.

If the forecasters in the Department of Economics had talked to historians, medical researchers, or even their wives, they might have come across some information that would have caused them to question their forecast of continuing high birth rates.

Now one might say - so what? Why does it matter that the Ontario Department of Economics got the demographic forecasts spectacularly wrong?

It matters because governments, firms, and other economic actors rely on demographic forecasts to make investment decisions. If the demographic forecasts are wrong, schools get built unnecessarily. Medical students train as obstetricians, when they would be better off becoming orthopaedic surgeons. Real estate developers build homes for large families, not small ones. Inaccurate demographic forecasts are unlikely to be good economic ones.

This historical incident matters because it says something about why long-term economic forecasting is hard and how forecasts could be improved. What I take from this story is that sometimes questioning the parameters of a model, and bringing in additional information from novel sources, may be of greater value than refining and tweaking a business-as-usual model.

For those who are interested, here is an image showing the 1961 Ontario population forecasts:

Comments

[Donald A. Coffin]

Having once(1976-1980) been peripherally involved in the population forecasts for one state (Indiana) and its subdivisions (counties) in the US, I can attest that making population forecasts is about as much fun as a root canal. Obviously, by 1976, the people actually doing the forecasts knew that the baby boon had busted. And the forecasts were for dramatically slower population growth than Indiana had experienced from 1900 to 1960. When we traveled around the state presenting the forecasts, the local governments and Chambers of Commerce were essentially unanimous in rejecting them--they thought that the forecasts were unreasonably low. [The forecasts were wrong...they were somewhat (but not a whole lot) too high.] Forecasts of low population growth were unpopular because all the community leaders were invested in the concept of growth (or populations, of everything) as an ultimate good.

[Frances Woolley]

Donald - fascinating! Thanks for sharing that. Do you remember Nick's old blog post on university crashes http://worthwhile.typepad.com/worthwhile_canadian_initi/2011/01/university-crashes.html? That explained why growth makes governance easier.

[Jacques René Giguère]

The pill didn't much to do. with it. It's merely another,maybe easier, mean of regulating your fertlity that you wished to regulate in the first place. (It's a family blog, so we'll skip the details...)
In my own town, the college is situated at the edge of the city on a stupefyingly large site surrounded by a forest. It was supposed to be many times larger and would have been in the downtown area had the population reached the 300 K that was envisionned (we're merely 25K...)
It also means problems with stagnating water in the main pipes as the flow is too slow in the too large pipes. It also has some other oddities such as if you go up a few kmona desolate rocky strreet, you suddenly find a large unpaved road six lane wide. It was supposed to link a luxury suburb of 150 K inhabitants to aforementionned downtown. Never paved as the auburbs wasn't built. Probably the world's largest forest trail.

[Frances Woolley]

Jacques René - love that story! Course that story is also about urbanization as well as declining birth rates.

Martha Bailey has been doing some fantastic work on the birth control pill, e.g. here. Her argument is that what made the birth control pill special was the certainty it provided about the timing of births. With pre-pill contraception methods, it was possible to avoid pregnancy 90% or 95% of the time. But that still meant unplanned pregnancies would happen 5 or 10% of the time - and that 5 or 10 percent chance of pregnancy was enough to make, e.g., investing in a professional degree extremely risky.

[AureliuszKleks]

You probably have already seen this, Japanese demographic forecasts were also bad: https://flowingdata.com/2015/01/08/japan-fertility-rate-forecasts-versus-reality/

[Peter]

It's been a while since I've been here, but I am now in Alaska and dare not touch a population projection. I think that even the though the "technology" (birth control pill, discovery of oil in Alaska, Western political shift to nationalism) may be known the impact of that technology is unknown until it becomes obvious in hindsight. In Vancouver the bridge to North Vancouver opened up real estate to be within commuting distance and the population grew. There is land next to Anchorage that some people believe will attract population growth - even though there are a lot of signs that population overall will decrease (at least in the short run). What is fundamentally different that says one bridge will be a catalyst for growth and the other will not? What makes a major infrastructure project an economic driver in one community and a drain in another? I impressed with any 15 year forecasts anywhere in North America has the sign correct for each demographic cohort.(I.e. Did the model accurately predict growth/decline) - let alone the magnitude.

[Lewis]

I'd just make one point about what actual data might have been available to the demographers way back in 1961. It might have been a mid-1950s total fertility rate, probably not a late 1950s one and certainly not an early 1960s one. In other words, there's a lag in the data (e.g. the most recent TFR from Statistics Canada is 2013). So for the demographers doing the projections released in 1961, this would mean they would be looking at an upward trend that had been underway since the late 1930s or early 1940s. And many demographic projections are pretty much trend-based with a little bit of judgement added. To predict the sharp drop in the TFR, the demographers would have to add a lot of judgement to overrule the trend analysis ... and then have a lot of explaining to do with their bosses.