Why they crashed too

By P.W. | LONDON

WHY did no one see it coming, asked the Queen at the height of the financial crisis in 2008. Implied in her question was another: why did economic models fail to anticipate it and why did they fail when it occurred? The answer, say two leading econometricians, is that such models crash along with everything else in crises. David Hendry and Grayham Mizon explain that this occurs above all because the “law of iterated expectations” fails. To the uninitiated that is cryptic: what do they mean by that?

The standard models used by policymakers in central banks go by the unlovely term of DSGEs (short for dynamic stochastic general equilibrium). These involve forecasting future economic developments on the basis of what has happened in the past. Naturally these forecasts take into account past trends and variability, but crucially they assume that future variability will continue to be around the same historical averages. There are no unanticipated “location shifts”, say the econometricians, meaning that the averages of the underlying probability distributions do not alter.

This approach works as long as the structure of the economy remains stable. But it breaks down when fundamental shifts in the economy occur, such as the outbreak of a financial crisis. The models did not anticipate such a crisis in highly developed economies, let alone one of its severity, which came close to toppling entire banking systems. That in turn meant that the economic forecasts were based on information that the financial crisis had rendered redundant. That is why the “law of iterated expectations”, which uses probability distributions based on previous experience to predict future outcomes, broke down.

One way of thinking about the models is to imagine a sailing boat being drawn along by an ocean current (trend) but with no set wind direction, so that the gusts (shocks) buffet the vessel but never fundamentally push it off course. That is a stable system and the models can handle that, providing useful help to the ship’s navigators in the central banks. But if the boat encounters turbulence caused for example by an unexpected powerful cross-current then the crew will be let down at the very moment that they most need reliable guidance.

Financial crises are one example of such structural breaks, but they can occur for other reasons. The econometricians show how the British labour market went through at least four main phases in the past 150 years. Between 1860 and 1914 unemployment moved up and down sharply but the fluctuations were centred around a stable long-run average of about 4%. That era was quite distinct from the interwar period of mass unemployment, peaking in the early 1930s when Britain was forced off the gold standard. Another structural break occurred during and after the second world war, ushering in a long period of low unemployment. A further one followed the oil-price shocks of the 1970s and the return to high jobless rates followed by their gradual descent until the crisis of 2008.

Put simply, the models for all their bristling mathematics cannot cope with what Donald Rumsfeld dubbed the “unknown unknowns”; or what the econometricians call “extrinsic unpredictability”. But should this have come as such a surprise for policymakers? Almost a century ago Frank Knight highlighted the distinction between risk, which can be calibrated in probability distributions, and uncertainty, which is more elusive and cannot be so neatly captured. Moreover, central banks, of all institutions, should not have forgotten their own history: the Bank of England spent much of the past three centuries fighting financial crises while the Fed was set up to counter them. The models may have failed but it was their users who vested too much faith in them.