Optimists and pessimists alike line up predictions for the year

How accurate are economic forecasts? Our Golden Guru table reveals all, as Diane Coyle reports

Jokes about economics tend to be well-worn old favourites. One of the classics is a piece of advice about forecasting: say where the economy is going, or when it will get there, but never predict both at the same time.

Professional economic forecasters have no choice but to commit themselves to both the where and the when, and the range of predictions is unsurprisingly wide.

This year the Independent is reviving its Golden Guru award for the most accurate forecast of the economy in 1995.

Last year was one in which the slowdown in the pace of growth surprised economists, but unemployment nevertheless kept on falling. Britain's inflation performance was good despite a surge in commodity prices and a weak pound.

The Government's target measure of inflation - the retail price index excluding mortgage interest payments - was up 2.9 per cent in the year to the fourth quarter of 1995. GDP grew 2.6 per cent in the year, according to the preliminary estimate. Unemployment was 2.25 million, equivalent to a rate a little over 8 per cent.

Taking these three aspects of economic performance alone, the closest forecast for 1995 at the end of the previous year was made by the National Institute of Economic and Social Research, the independent research body. The forecasting team was led by former director Andrew Britton and Gary Young. The two modestly put their success down to luck. The National Institute was closely followed by the economics team at the City of London investment bank Lehman Brothers.

Any measure of forecasting performance is bound to be arbitrary. Our guru index is the absolute value in percentage points of the difference between the actual and forecast values for only three variables - growth, inflation (on the target measure) and the unemployment rate. The forecasts were made between November 1994 and January 1995, published in the Treasury's monthly comparison of forecasts for the UK economy.

Success in other aspects, such as predicting the balance of payments deficit or the government borrowing requirement, wins only an honourable mention. And any forecaster who finds that subsequent revisions to the official figures take their prediction to the top of the guru league will have to settle for moral satisfaction this year.

One of the facts of life in forecasting is that it is hard to beat the average. It has been proven repeatedly that the average of a group of economic forecasts is more accurate than most of the individual forecasts that make up the average. Forecasts tend to cluster around the right numbers for the different variables whose values are being predicted while not often being spot-on for every value themselves. And even if an individual makes an ace forecast one year, he is unlikely to repeat it the next.

The 42 independent forecasts in this sample are no exception. The average of the forecasts came sixth in the guru league.

A new working paper* from the National Bureau of Economic Research in Cambridge, Massachusetts, sheds more light on this phenomenon. The author, Owen Lamont, looked into the question of why any forecasters ever bother to publish an outlook that is not close to the average.

The answer, he finds, lies in the forecasters' commercial incentives. On the one hand, their clients would quite like an accurate set of predictions. On the other hand, offering a forecast that stands out from the crowd is about the only way an economist can build a distinctive reputation, and hence win new clients.

Mr Lamont compared 133 forecasts of GNP, consumer price inflation and the unemployment rate in the US for the years 1971-1992. The predictions were less accurate if made by people who owned their own firms, by economists working in new firms or by older economists. Youngsters lack the confidence to stick their necks out.

The explanation, as you might expect for a bunch of economists, is that most forecasters are not paid according to their accuracy. If they ignore the conventional wisdom and turn out to be right they will get press attention and professional rewards. If they are wrong, nobody will really notice.

The results for the UK forecasts last year corroborates Mr Lamont's explanation. Two out of the top five forecasts were made by academically inclined teams, at NIESR and the London Business School. These have their commercial imperatives, but they are likely to be directly linked to forecasting accuracy.

The bottom five were all made by respected City economists, some of whom have built their reputation on an untypical analysis of how the economy works. In other years their bias would have propelled them to the top of the list.

Either way, the economics talent at the 25 City firms in the group costs their employers, at a conservative estimate, pounds 5m in salaries - and probably closer to pounds 10m.

This year one of the forecasting groups that is sticking its neck out is the Treasury, motivated perhaps by the Chancellor's more-optimistic- than-the-average view about potential growth thanks to his government's management of the economy.

For those who would rather rely on the average, GDP is expected to grow 2.4 per cent this year, target inflation to be 2.8 per cent by the fourth quarter, and unemployment to have fallen to 2.13 million. The Treasury predicts growth at 3 per cent and inflation at 2.5 per cent. (It does not publish its unemployment forecast, disqualifying itself from the guru stakes.)

However, there is one 1996 forecast even more optimistic than Mr Clarke's. Richard Jeffrey at Charterhouse predicts a 3.3 per cent rise in GDP, taking unemployment below the 2 million level, with inflation still well behaved at 2.7 per cent by the third quarter.

At the other extreme, forecasters at two City investment banks, Goldman Sachs and Robert Fleming, put growth at only 1.7 per cent, with inflation on the target measure at 3 per cent and 2.5 per cent respectively.

The range for 1996 inflation forecasts runs from 1.5 per cent (Patrick Minford's team at Liverpool University, which predicts 2.9 per cent growth) to 3.5 per cent (SBC Warburg, which reckons growth will be 2.6 per cent).

It is too early in the year to have any idea whether the optimists or pessimists will be proved right. But economists at Kleinwort Benson, one of the most accurate 1995 forecast teams, have recently published an analysis of the ''surprises'' in monthly economic data so far this year. Looking at the actual monthly figures for key activity indicators such as retail sales, money supply and industrial output, Kleinwort Benson economist Leo Doyle has found that nearly all have been stronger than the average expected by the financial markets.

Mr Doyle admits that: "The history of economic forecasting is not a proud one. Nearly all major turning points in economic activity have been totally missed by nearly all forecasters." His "surprise index" is designed to add the missing insight. It was positive during most of 1994, a high growth year, and negative for most of last year. In 1996 it has been positive again.

Research by Jane Memmler.

*NBER Working Paper No 5284, Macroeconomic Forecasts and Microeconomic Forecasters by Owen Lamont.

Economic forecasts made for 1995

GDP Inflation Unemployment Absolute

growth RPI excl mortgage rate error

interest

% % % % points

Top Five

NIESR 2.7 2.8 7.87 0.37

Lehman Brothers 2.8 2.8 8.12 0.38

London Business School 2.7 3.1 8.30 0.56

Kleinwort Benson 2.7 3.3 8.23 0.69

Credit Lyonnais 2.8 2.5 7.87 0.77

Bottom Five

Salomon Brothers 3.7 2.1 7.51 2.43

Robert Fleming 2.2 1.8 9.30 2.76

SG Warburg 4.5 2.8 7.15 2.89

UBS 4.3 3.3 7.15 2.99

Societe Generale 4.2 4.2 7.15 3.79

Average Forecast 3.2 2.8 8.00 0.78

Actual 2.6 2.9 8.04 n.a

Absolute error = sum of absolute errors on each component