Which of these ideas would you say was the more ridiculous? That economists might be able to predict recessions by analysing pregnancy rates or that we look to prevent swimming pool-related drownings in the US by politely asking Nicolas Cage to give up acting?
Given the former possibility was recently a BBC headline, you may take it more seriously but the reality is that both are founded on the same misconception: correlation = causation.
Pregnancy rates predict recession?
The BBC article referenced an academic paper, Is Fertility a Leading Economic Indicator?, whose authors tracked more than 100 million births in the US between 1989 and 2016 and concluded that conception rates drop several months before other signs of a recession become visible.
As such, they suggested, fertility rates might help predict economic downturns and were in fact “more accurate than some traditional indicators”.
“Economists are frequently criticised for failing to accurately predict the direction of economic growth,” the BBC notes.
“Increasingly they are looking beyond traditional measures such as manufacturing output, retail spending and house prices to help build a more complex and accurate picture.”
And, in the paper’s authors’ eyes: “The factors behind the last three recessions also had a profound and rapid effect on fertility decisions.
“In fact,” they added, “these factors seem to have impacted fertility decisions before large parts of the economy.”
Glossing over how three recessions hardly makes for a rigorous statistical sample, it all sounds plausible enough and yet, in reality, the foundations on which the paper’s conclusions are based are as solid as that pool-safety plan necessitating Nicolas Cage’s early retirement.
Where's the causation?
For while there is a perfectly good graph showing a high correlation between the number of people who have drowned after falling into a pool since 1999 and the number of films in which Cage has appeared over the same period – just as there is between US conceptions and the last three recessions – no correlation has any statistical significance unless there is also a causal link.
It is a point we have made before, here on The Value Perspective – for example, in Tangled up with blue, where we picked up on the high correlation (and zero causal link) between the per capita consumption of cheese in the US in the first decade of this century and the number of people who have died after becoming entangled in their bedsheets over the same period.
While both the ‘Drowning v Cage’ and ‘Cheese v Bedsheet’ correlations come from Tyler Viglen’s excellent Spurious Correlations website, however, the ‘Recession v Conceptions’ one has been considered worthy of extensive media coverage.
“We were surprised when we saw [the correlation],” one of the report’s authors told the BBC, “and then we were surprised no-one had noticed it before.”
It's a trap often fallen into
It is of course entirely possible someone had indeed spotted it before but understood confusing correlation with causation is a trap human beings – from academics to investors – fall into time and again.
What is more, there are currently plenty of investment strategies on the market that have been back-tested over shorter periods of time and demonstrated lower correlations than our two ‘Spurious’ examples.
In Tangled up in blue, we concluded: “While people may be happy to entrust their money into such options regardless of the lack of causation, we prefer to recommend strategies that are not only able to demonstrate outperformance but can also explain why they outperform.
This understanding is the only protection from being whipsawed by the short-term underperformance that will inevitably affect all strategies at some point.”
And Nicolas Cage can go back to working on the sequel to The Croods with a clear conscience.