The Insoluble Perils of Prediction

Statistically, there are usually only a handful of ways that a prediction can come true – but a myriad, even infinite, ways that it can go wrong. We should almost never expect a particular prediction to hold true.

Predictions have always been fraught with perils. Gauging the future is tricky business, a business filled with well-meaning and serious commentators drowned out by pranksters and ideologically committed imbeciles. Despite being wholly unequipped for the task of divining the future, most political pundits and financial commentators are revered as saints for their extraordinary abilities to misjudge the future. 

We ought to treat most predictions with a large chunk of skepticism. We should reject vaguely phrased and imprecise ones, not because they couldn’t carry some valuable insight, but because they cannot be assessed. 

If I forecast a future recession, but not what will cause it or trigger it, how long it will last, how bad it will be, and where it will strike, I’m not making much of a prediction; I’m not providing any information beyond extrapolating the known past. Recessions, like earthquakes and shark attacks, occasionally happen, and saying that they will happen again is trivial and does not mean others ought to admire me for my fantastic foresight.  

Statistically, too, there are usually only a handful of ways that a prediction can come true – but a myriad, even infinite, ways that it can go wrong. If all these possible scenarios are equally likely (which they of course aren’t), we should almost never expect a particular prediction to hold true. 

The challenge is, as popular economist Tim Harford stated before interviewing Philip Tetlock, author of the celebrated Superforcasting: The Art and Science of Prediction: in a myriad of predictions of varying quality, how do you find the Cassandras amidst the Chicken Littles? 

Well, let’s track them, said Tetlock after noting that political “experts” were repeatedly asked for their conjectures even after having seen their previous prediction gone entirely wrong. Let’s see how often they are right. For decades, Tetlock has run this research program, measuring forecasters’ success at predicting things.

In regards to environmentalists’ outrageous claims about the future, Toby Young at the Spectator said refreshingly that rather than get demoralized about misinformation and hyperbole, “we should be grateful that these gloomsters make such oddly precise predictions.” That lets us measure the predictions, assess them, and subsequently discard the gloomsters’ opinions when enough of their predictions have gone the opposite way: “It’s like putting a sell-by date on their credibility,” rejoices Young. 

From global famines in the 1970s to the hilarious projection that Spain by 2020 will be uninhabitably hot and ridden with malaria, to cataclysmic sea-level rises and mass extinctions and climate change deaths in our century, environmental doomsday prophets have made this into a sport that we should delight in playing. 

Let me play prediction follow-up with another gloomster, the NYU economics professor Nouriel Roubini, known in the financial press as Dr. Doom for having predicted the housing crash in 2007 – or, more precisely, having called doom-and-gloom enough times to accidentally be vindicated by future events.  

Iran, Oil Prices, and the Black Coronavirus Swan

In January, Dr. Roubini gave a series of talks at the Norwegian asset-management company Skagen’s prestigious annual conference. At the time, his lecture seemed very thoughtful and nicely put together. To remind you of the incessant noise that is global news, in the background were escalating conflicts with Iran, worry about the impacts of oil prices on a frail economy, and U.S. trade wars.

True to his reputation, Roubini painted a dire macro picture of the new decade: Purchasing Managers’ Index data and other forward-looking indicators for the major economies kept coming in lukewarm at best. The trade deal with China that we have now all but forgotten about looked bleak. Hovering in the background was Brexit, with its many possible and dire outcomes, and in January the telenovela that is the presidential election seemed like it could go in absolutely any direction. 

Roubini very neatly tied together three different topics and how they were likely to impact each other: Iran, oil prices, and global growth. It was a plausible story: oil prices matter for global growth as oil is the master industry that powers every other industry, and what happens politically and militarily with Iran can certainly move oil prices a lot. 

Here’s my stylized summary of Roubini’s argument:

ScenarioGlobal GrowthIran-U.S. ConflictOil PricesRoubini’s likelihood
1Expansion (3.4%)Return to normalBelow $705%
2Slowdown (3.0%)Military escalation$70-8045%
3Further Slowdown (2.7%)Direct Military Exchange$80-10030%
4Global Recession (<2.5%)Full-scale war>$10020%

At the time the projections were sound and serious; these were the issues of the day. Yes, his causal mechanism and assessment of Iran’s political options seemed sensible. Perhaps we could quibble over the exact estimates – were we really facing a one-in-five chance of full-scale war with Iran? – but overall, it was a perfectly reasonable story. 

Enter reality. coronavirus scare, pandemic chaos, draconian government actions, and an outbreak in Iran that wholly re-arranged everyone’s priorities. In Iran, as in most countries, the focus of politics swiftly changed to epidemics, health care, and travel restrictions rather than political grandstanding over energy, influence, or oil. Governments left and right went bananas, enforcing one centrally planned authoritarian restriction after another.

Enter reality round two: the Saudi Arabia–Russia oil price war meant that global oil prices fell rapidly and massively when the Saudis ordered large increases of production – oil producers are “flooding the market simultaneously in a game of chicken,” wrote Wall Street Journal’s Spencer Jakab

While the news is still fresh and the precise reasons are not yet known, efforts seem to have been aimed at squeezing Russian and American oil producers in order to improve its own market share. The move from $55 per barrel, where crude had been trading during February ($58 when Roubini spoke on January 17) to below $30 in intra-day trading on Monday made Roubini’s prediction about oil prices entirely obsolete. 

Prices of less than half of Roubini’s lowest (and least likely) scenario were associated with U.S-Iran peace and a thriving world economy. Instead we’re looking at a paralyzed Iran with upwards of 10,000 people infected, a country too busy with its medical emergency to heed the prediction of some gloomy economics professor; we’re not looking at war with Iran, but global production at a standstill, ruthless travel and commercial restrictions in many countries, and a possible global recession. 

Not that it prevented Roubini from publicly claiming that he predicted this and that the coronavirus outbreak and oil shocks are “a true signal of upcoming global recession.” I thought war with Iran and rapidly increasing oil prices were his main risk factors for global recession…?

We can invoke the black swan, Nassim Taleb’s almost tired analogy for sudden and unexpected tail-risk events. Just a few weeks ago, Roubini’s careful analysis seemed prudent and probable. Then the world happened, and his impressive predictions were worth nothing. 

Naturally, it is too much to expect Dr. Roubini to have foreseen the impact and domino effect of this pandemic. The point here isn’t that things can’t change and that predictions made on outdated information must hold, but how shaky forecasts are in the face of market reality, human psychology, politics, and black swans. 

Stuff happens, and your carefully laid plans and sensibly argued projections are quickly overthrown. 

The lesson here is twofold: the reality of our world includes sudden events that cause all kinds of unforeseen turmoil. And puffed-up soothsayers, however lucky their forecasts might have been in the past, are usually completely wrong. When some of them are right, as the law of large numbers and the existence of overconfident know-it-alls guarantee, we should probably not credit them too much. 

Lottery winners are, after all, not impressively divining seers.



Post on Facebook


Post on X


Print Article