Why Grexit odds are probably 99% wrong

What is the probability of Greece exiting the eurozone? The fluctuating survival chances of the eurozone’s weakest economy are hugely important for global financial markets. Its continuing membership of Europe’s monetary union would avert a disruptive market correction, at the very least.

Statisticians would say the odds are impossible to calculate. “Grexit” — Greece’s exit — would follow a sequence of events, decisions and emotional choices by voters and politicians, and unpredictable economic and financial forces. Even if point estimates were possible, this week’s news flow shows they would change from hour to hour.

But that has not stopped economists and strategists attaching quite precise numbers to Grexit risks. On Monday, Berenberg Bank raised the possibility of Grexit from 40 per cent to 55 per cent. BNP Paribas reckoned it had increased only to about 20 per cent. But Standard & Poor’s rating agency saw a 50 per cent likelihood. Writing for the Financial Times, Mohamed El-Erian, economic adviser to Allianz, saw an 85 per cent probability. Nobody was specific about the timing.

Are such estimates of any use? Possibly, as a shorthand way of expressing levels of concern — or grabbing headlines. But the range of estimates highlights their spurious accuracy (are you sure it is 55 per cent and not 62 per cent?), which in turn reveals markets’ inability to price political risks. In fact, a widespread view among traders recently has been that Grexit risks are impossible to trade, which helps explain why share and bond price movements have been relatively limited as events have unfolded in Athens.

The danger, however, is that assigning numerical probabilities creates the false impression that markets have superior knowledge about the future; the risk is of being caught out badly when they are wrong.

The idea of attaching probabilities to economic events comes from corporate high-yield (or “junk”) bond markets, the riskier end of the corporate debt spectrum. Here, historical data can be used to estimate the probability of a company defaulting and the amounts bondholders are likely to recover if it does. Prices can then be set accordingly.

The temptation to use a similar approach with European government bond markets is understandable. Perceived “credit risks” — the risk of default — have risen since the post-2007 crises. Moreover, in an era of ultra-low interest rates and “quantitative easing” by the world’s main central banks, government debt prices have become more sensitive to political risks and the whims of policy makers. Being able to price such factors would create some great moneymaking opportunities.

Unfortunately, that is not easy. There is no past history of countries exiting the eurozone. Greece would (presumably) also only be thrown out of the eurozone once, so there will be no track record. Even if Grexit possibilities were calculable, nobody could say what the cost would be to investors in, say, Italian bonds. While Grexit might fuel fears of broader eurozone break-up risks, it would perhaps be met by a forceful policy response by the European Central Bank, which might even push Italian debt prices higher.

So why are guesstimates made of Grexit probabilities? The best explanation I have heard this week is that by quoting numbers, strategists can indicate their conviction levels. If you say there is a 90 per cent chance of something happening, you are pretty sure it will. If you see just a 10 per cent chance, then you do not think it will. According to this way of thinking, a “50 per cent chance” is the ashamed analyst’s way of saying “I don’t know”.

What is left unsaid is that there will be no way of checking in the future whether the numbers made any sense. Even if you predict just a 10 per cent chance of Grexit and it happens, you will still be able to boast: “I did warn you.” One strategist to whom I spoke admitted: “The best thing is that ex-post you’re never wrong, so long as you avoid assigning 100 per cent or zero per cent.”

The benefit to strategists is they give the impression they have special insights. Whether they are of much value is questionable, but here is a prediction: the investment community will continue attaching spurious numbers to Grexit and similar possible events. Of that I am 95 per cent sure. Maybe even 96 per cent.

ralph.atkins@ft.com