The limits of forecasting

One of my first jobs when I became a bank economist in Canada was to compile a medium term (5 and 10 year) forecast for the Canadian economy. This forecast was then used by the bank's Planning Department to develop asset and liability projections for the bank itself, which in turn drove plans for staffing, branch opening and so on.

The job became mine because the previous incumbent had not endeared himself to the planners. Whenever he presented his forecast to them, they would start to cough nervously and say that they couldn't live with the balance sheet implications of his economic forecasts. So one year, my predecessor took along a forecast written entirely in pencil. When the usual demurrals began, he whipped out a large eraser and poised it over the page, saying "OK, tell me what numbers you would like". The planners were aghast, though perhaps not as aghast as when they found out that I was the replacement forecaster.

The fact is that no economic forecast exists in a vacuum. It was quite legitimate for the bank's planners to hesitate if the balance sheet implications of the economic forecast looked unsustainable. Numbers thrown out by macroeconomic models don't have to be compatible with good financial management. (Problems at Northern Rock and elsewhere might have been avoided if management at those companies had kept this in mind). A forecast has to be based on a set of assumptions, then followed through logically to its conclusions.

This little diatribe is, of course, motivated by the response to the economic forecasts in the budget, which show the UK economy starting to grow again by year-end, showing a small positive result for 2010 and accelerating in 2011. These numbers in turn drive the borrowing projections. Lord knows these are scary enough, but the Chancellor stands accused of excessive optimism and even, by Vince Cable, of "reverse engineering" the forecast. That's to say, Vince believes that the Chancellor came up with the borrowing numbers first, then had his economists generate a macro forecast that would be consistent with the borrowing level.

I have to say I doubt this: it's hard to see why the Chancellor would have used a borrowing requirement of £175 billion as a starting point, if he had any choice in the matter. In any case, since the future is unknowable, the only real requirement is that the growth numbers and the budget numbers should be consistent. Anyone can then argue about the economic asumptions without impugning the Chancellor's honour: as he would not hesitate to admit, if the economy does worse than the government expects, the borrowing figures are likely to be higher.

The macro forecasts themselves look optimistic but not unachievable, and given the relatively good forecasting record of the Treasury, they deserve a bit more respect than they're getting. They can certainly be taken more seriously than some of the hot air emanating from the likes of the Institute for Fiscal Studies. This august body has seen fit to pour scorn on the Treasury's forecast for 2010 and 2011, while with a straight face publishing its own fiscal outlook for the period up to 2032! Supposedly it will take until February of that year for the public debt/GDP ratio to return to the "acceptable" 40% level.

Amazingly (well, actually not amazingly at all), this has been regurgitated by the media as if it had been handed down from on high on stone tablets. All I can say is that I hope that the IFS, like my erstwhile colleague, has a very large eraser.