Some 25 years ago, when I was in charge of forecasting at the Organisation for Economic Cooperation and Development (OECD) in Paris, a colleague and I conducted a detailed postmortem on the accuracy of economic forecasts. We wanted to understand why forecasters had got things so badly wrong after the great 1973-74 oil shock.
Our conclusion was that in most years economic forecasting is quite accurate—actual GDP growth generally comes out to within plus or minus one percentage point of estimates prepared a year ahead. However, forecasters get thrown when two conditions are present simultaneously: when economies are subjected to a shock that is large—which we took to be a shock of 1 per cent or more of GDP; and when that shock is novel—meaning that it involves transmission mechanisms that have not been observed before.
The oil shock of 1973-74 was a global exemplar. It was certainly large, at a stroke raising prices in OECD countries, and thereby reducing real incomes, by around two percentage points. And it was also novel. About 2 per cent of OECD GDP had been transferred to a handful of oil-producing countries, and no one knew how fast they would spend this money, where they would park it while they were deciding, and what and from whom they would buy when they ultimately did. In the event, it took the oil producers several years to spend their windfall; they parked their unspent balances largely in US treasury bonds, and the resulting increase in world savings slowed world GDP growth to a crawl for several years.
By contrast, the 1978-79 shock, while equally large, was not novel—because it was now understood how oil shocks transmitted themselves through economies. And while the consequences of the first shock were poorly predicted, forecasters got the second one right.
Such a framework provides a way in to considering the possible consequences of this present crisis. First, does the current shock qualify as large? On the twin assumptions that the global destruction of wealth arising from the sub-prime collapse is of the order of $5 trillion, and that people reduce their spending by around five cents for every dollar that their wealth decreases—roughly the pace at which they increased their expenditure on the way up—the resulting shock to spending might be around 0.5 per cent of world GDP. But that is almost certainly an underestimate. People who have been living on credit and are at their borrowing limit may well have to cut back more sharply than they splurged on the way up; and sharp increases in mortgage costs could also compel sharp expenditure reductions. Moreover, this shock is hitting a world economy that was already slowing. Much of the developed world had probably entered recession some time in the first half of 2008, the result of the US housing slowdown, the doubling of oil prices between 2004 and 2007, and their redoubling after that.
So a large shock, yes, but is it also novel? Some past crises in individual economies—Sweden and Norway in the 1990s come to mind—were similar in some respects, but these economies were too small to infect the rest of the world. Moreover, by virtue of being small, they could devalue and export their way out of recession in a way the world can't. Similarly, Japan's "lost decade" of the 1990s, which involved painful deleveraging after a long credit-driven property boom, was a national affair. The assets which have led to today's crunch were manufactured on an industrial scale by the US, and bought by institutions globally.
One other shock—the dotcom collapse in 2000—slowed OECD growth to 1 per cent or below. But neither this nor the oil shocks involved anything like the current drying up of liquidity and credit, and the uncertainties about who is holding what risk, that has been seen in most of the major economies in recent months.
In some ways, a closer parallel might perhaps be found in the oft-cited events of the 1920s and 1930s, which culminated in the great depression. But this does not represent as close a parallel as is sometimes suggested. One reason is that the size and role of governments is fundamentally different today. In the 1920s, government spending in the rich economies accounted for only 5-10 per cent of GDP. Today this (comparatively) stable component of spending is four to five times as large. Moreover, the so-called "automatic stabilisers," whereby government deficit spending rises as the economy weakens, damp down any shock in a way that was not the case then. And, most importantly, governments accept a responsibility for supporting the economy that simply was not recognised in the 1920s.
It is hard to escape the conclusion that the world economy has been hit by a shock which is both large and novel, not least in respect of the unprecedented sharp rise in inter-bank rates and borrowing costs for companies. This leaves the forecaster in uncharted territory, casting around for clues. One such clue is offered by the OECD, which recently estimated, for the US, that the tightening in financial conditions (including credit conditions) may subtract nearly 2 per cent from GDP over the coming four to six quarters.
If asked to give a best guess, I would conclude that not only is the world economy bound to slow over the next few years, but that world GDP may even fall—which, if it does, would be the first time in postwar history. However, any fall would seem likely to be markedly less severe than in the 1920s, when the peak-to-trough decline in the major economies averaged nearly 12 per cent, ranging from around 30 per cent in the US and Canada to somewhat under 10 per cent in Japan, Italy, and Britain.
If things turn out to be worse, it will probably be because policymakers make a set of fundamental mistakes (such as lapsing into protectionism as they did in the 1930s), or because the media have panicked consumers and companies into reducing their expenditure more than they need to. At the same time, however, we should not fool ourselves. Whatever the precise path of GDP, whether globally or nationally, it will feel awful.
