Forecasting is part art, and part science. Unfortunately, far more art than science gets applied - you’ll see a stunning example shortly.
The most dangerous of all forecasting techniques is the simplest - extrapolation. To many planners and forecasters, taking an historical trend and extending it in straight line from past to future is like looking at a traffic accident. No matter how grisly the scene may be, you just can’t take your eyes off it.
It matters little whether the straight line is derived from a linear regression. The underlying assumption regarding historical events - that is, that conditions which occurred in the past to produce events will occur in the future - is a risky assumption.
Ask anyone who has been in either the stock or the housing markets the past few years. Averages, and the comfort of narrow and orderly deviations from the mean, rarely align to produce predictable outcomes.
Population Projections
In the eighteenth century, The Reverend Thomas Malthus, a British political economist, understood the fallacy of drawing straight lines from past events. He explained the fallacy in his 1798 work, An Essay on the Principle of Population.
In put forth his theory that, as populations grew to the point of outstripping the capacity of the environment to sustain them, that war, famine, pestilence and disease would keep populations in check. The work considerably influenced social policy leading, in part, to the creation of the census poll.
It certainly stirred up more than a few arguments in is day - both in support, and in refute of his theory. It’s interesting to note that, 200 years later, the arguments still persist.
In 2007 the liberal-leaning New York Times claimed that man’s inventiveness would free him from the bondage of a Malthusian catastrophe, whereas the conservative Wall Street Journal devoted a front page article supporting the view that Malthus had his mojo working for him when he published his work.
Who’s right? Who’s to know! Thank goodness for politics and the miracle of elections.
A far more sensible view of the future that entirely avoids straight-line extrapolation comes courtesy of a 240-page United Nations study published in 2004, suitably titled, World Population to 2300. It contains this graph, showing world population projections through to 2100.
The analysts who prepared the report know better than to draw a tempting continuation of the blue line representing Actual worldwide population. Instead, their projection consists of three flavors: each is dependent on varying assumptions about the future.
The U.N. has certainly covered its bets: at the high end the population could double from its 2004 level, and worst case could decline to 1990 levels. There’s a projection in there to suit just about every glass half-empty or half-full outlook.
Kidding aside, this is not an exercise in wiggle room. The U.N. has rightly shown enormous variation in outcomes for one simple reason: it is impossible to predict what will happen to conditions that affect population growth with any certainty.
Will the Last Woman on Earth be a Hot-blooded Brazilian?
Which brings us to the stunning example I promised.
This week, The Economist published this article on declining fertility rates in Asia. Whereas three decades ago China imposed strict measures to control the number of children couples had (presumably, Communist Party members support the Malthusian view) there is a different problem today. In a period of unexpected increases in wealth and living standards across Asia, it turns out that fewer people are getting married. Hence, there are fewer children being born.
In its blog, Daily chart, The Economist had some fun with what it admitted are “back-of-the-envelope calculations”. It took U.N. data on fertility rates, and simply extended them in a straight line to see what would happen. The end product is this graph projecting the end of human civilization as the last child-bearing female leaves this world.
Don’t for a minute think that the editors of The Economist buys into this. They don’t. They are simply having “what if” fun, and end the blog by stating the condition, “if present trends continued unbroken,” to assure its readers of the unlikelihood of such events.
(That said, I’m intrigued by the Canadian fertility rates relative to other countries. Seems it is a very different country than the one I left in 1994.)
Takeaway
Most projections we see span no more than 5 - 10 years. It’s easy to take them as gospel.
Next time you see a forecast (projection) that piques your interest ask yourself this: what would the projection look like if it extended out 40, 50 or 100 years?
If that outcome seems absurd, then it would be prudent to examine the underlying assumptions used by the forecaster. Especially if the forecast concerns something you would bet your retirement or career on.
