The Politics and Sociology Of Economic Forecasting

This is an extended version of an article published in The New Zealand Herald July 7, 1998.

Keywords: Macroeconomics & Money;

Last week parliamentary leader of ACT, Richard Prebble, called for the resignation of the Secretary of the Treasury, Alan Bollard, because of inadequate Treasury economic forecasts. The call is so political that it may be dismissed as silly, but it also raises wider questions about economic forecasting in general.

Just over decade ago, as it happens when Mr Prebble was a Treasury minister, I had some concerns about the sort of political interference which Mr Prebble alleges. Discrete enquiries elicited that legally the Treasury forecasts are the Treasury’s and not its ministers, as is indicated by the its Secretary’s signature on the released forecasts.

Could the Treasurer or, as he was then, the Minister of Finance illegally interfere with the forecasts by “nobbling” the Secretary? As an ex-minister and long time parliamentarian, Mr Prebble would be knowledgeable about this, and would report that it is impossible if the public servant is of the highest integrity. I have known Dr Bollard for many years, and am confident he is. Moreover, given his research background and, later, his quasi-judicial position as Commerce Commission Chairmanship, he comes from a culture where improper interference would be stoutly resisted.

The Treasury forecast might be incompetent. But it would be difficult to blame that on Dr Bollard, for it is done by a team within Treasury. That team was built up by previous Treasury Secretaries. It is unreasonable to expect a Treasury Secretary in the job for only five months to have reconstructed the forecasting team.

But were the forecasts incompetent? First, forecasting cannot be judged by a single figure. Economic forecasting is not like political forecasting where party leaders make outrageous claims for voter share based upon unfulfilled hopes. It involves forecasting separate components, which algebraically lock together underpinned by a theory.

Second, there is data inaccuracy. The recently reported GDP fall of .9 percent for the March quarter is subject to error and to revision. This is not a criticism of Statistics New Zealand, for data uncertainty is inevitable for the earliest estimate (based on incomplete data) in a small economy (where the law of large numbers is not as applicable). All forecasters work from an incomplete and inaccurate data set, even if by the time the data reaches the papers and politicians it is treated as perfect.

Despite these cautions, there is a problem about the recent forecasts of the Treasury and others, as three simple points from the sociology of forecasting explains:

1. Economic forecasting basically involves making a long-term assessment of economic growth, gauging the business cycle around it. If the long-term assessment is wrong so will be the cyclical assessment. The New Zealand consensus has been for much stronger growth performance than has actually occurred, and we have suffered from over-optimistic forecasts. (The reason my own forecasting record is better than average is because I have been more pessimistic about the long-term growth rate.)

2. The most difficult time to forecast is at the beginning of the recession phase of the business cycle (where we are currently) and the beginning of the recovery phase. Typically forecasters at first underestimate the strength of the contraction (and of the expansion). This suggests that most forecasters (not only the Treasury) will be further lowering their growth predictions. The current recession will probably be deeper and longer than they are forecasting.

3. A Treasury forecast has to be near the centre of all forecasts. Variations between forecasts reflect a variety of theoretical and personal differences of the forecasters involved. An official forecast, minimizing these differences, will inevitably be in the middle rather than at an extreme.

Given that few of our forecasters have been doing well because of the first two points, the implication of the last point is that we must expect problems with the Treasury forecasts too.

Yet we cannot entirely discount Treasury responsibility for the forecasting problems. But it is not the Treasury Dr Bollard heads. Over a decade ago (when Mr Prebble was minister, although I have no idea whether he was involved in the decisions) the Treasury and the Reserve Bank cut back on the public funding of macroeconomic research (especially to anybody who disagreed with them). The result has been a steady deterioration in the economics profession’s competence to forecast the economy. Forecasting teams have broken up, the skills, knowledge and memory lost. The losses cannot be refurbished without a significant research program to respond to evolving circumstances. Ironically one of those who suffered from the cuts was the NZIER, where Dr Bollard was the director for seven years. He will be aware more than most of the inadequacies of macroeconomic research. When he has settled into the job, he will no doubt address them.

My final point is that Mr Prebble called for the resignation of the current Secretary of the Treasury on the basis of one problematic forecast. But the Treasury forecasts involved even larger errors when he was minister. Why did Mr Prebble not call for the sacking of the Treasury Secretaries then? If he had, perhaps we would not be in quite the economic difficulties we are today.