The United Nations recently revised its forecast of world population for the end of the century to 11 billion people, up 800 million or about 8 per cent from its previous projection in 2011. Well, what’s an error of a few hundred million people between friends?
The UN quickly explained that the revision originated in “improved fertility forecasting methods.” Note how it immediately framed the change in terms of its ‘improved’ statistical models, not the inadequacy of its previous models. After all, missing the population count by 800 million human beings – more than the populations of North America or Europe today – is not exactly rounding error. All this repeats mistakes that have plagued demographic forecasting back to the days of Thomas Malthus and his prediction that population levels were condemned to stagnate since agricultural output could never rise beyond a subsistence level.
The fundamental error subsequent generations of demographers have made is their underlying assumption that populations change so slowly and predictably that they are relatively easy to forecast, at least compared with economic or social trends (and don’t even try anticipating technological changes; all the best minds in the country assembled in Little Rock after the 1992 election by Bill Clinton for a deep think about the future failed to mention the Internet). This mistaken belief that demographic change is glacial is why population was the only variable for which Statistics Canada issued official long-term forecasts, although recently it branched out into forecasts for the labour force, an even harder variable to anticipate.
There are many examples of sudden shifts in population and labour force that have mocked official forecasts over the years. In 1938, President Franklin Roosevelt convened America’s leading population experts, who unanimously predicted that the U.S. population would peak at around 140 million in the 1940s and then slowly decline. The U.S. population today is 314 million.
Legendary business thinker Peter Drucker wrote in his book Innovation and Entrepreneurship that erroneous forecasts of population are so common that they represent “a rewarding opportunity for entrepreneurs” since decision makers “cling to the assumption that demographics do not change – or don’t change fast.” The baby boom after the Second World War was completely unexpected, and the baby bust of the 1960s was equally surprising to demographers.
And if population is unpredictable, labour force participation (the share of the population willing to work) is unfathomable. The post-war surge in women joining the labour force mystified forecasters, as did the recent increase in the labour force participation of older workers. Looking forward, no one can anticipate how long older workers will remain in the labour force as longevity rises, medical technology improves, and labour market conditions tighten further.
In the future, if the UN or Statcan insists on producing forecasts of population or the labour force, they should be obliged to include an easily-understood summary comparing their past forecasts and the actual outcomes, so people can assess for themselves the credibility of these exercises. It would also force statisticians and modellers to confront the flaws in their forecasts, something they seem blissfully unaware of and unaccountable for.
Philip Cross is a Research Fellow at the Macdonald Laurier Institute and the former Chief Economic Analyst at Statistics Canada