Are Bad Statistics Causing Bad Policy?

Could be. Michael Keran, chief economist of the Prudential Insurance Company of America, says that the early statistics on Gross National Product are ``notorious'' for being substantially revised. For example, the GNP numbers released Wednesday by the Department of Commerce showed the output of goods and services growing at a 2.7 percent annual rate in the third quarter. That was revised up from the 2.5 percent rate reported a month ago. That GNP number could be revised again in another month, in mid-1990, and even later.

``We are all trapped with this data,'' says Dr. Keran, referring both to economic forecasters and to Federal Reserve policymakers. ``You can't rely on the current numbers.''

Keran often ignores the early GNP numbers, leaving his own forecasts untouched. He finds that the results of his mathematical forecasting model conform better to the final GNP numbers than to the early numbers. The same is the case for some other models of the economy.

Keran figures the economy has been running at close to zero growth in this last half of 1989. That implies either that GNP is plunging sharply this quarter, or that the numbers will be revised substantially, or both.

It also suggests that the Fed may have been fooled into following a tighter monetary policy than necessary. On Monday, the Fed sent a signal to participants in the credit markets that their guess of last week that monetary policy had been eased again was false. It pushed short-term interest rates back up to where they had been for the last few weeks - just under 8.5 percent for what is called the Fed funds rate.

Because the Fed has not eased as much as he anticipated, Keran is now expecting that the economy will be weak in the first half of 1990 as well as in the current period. He doesn't expect the ``nice bounce back'' he had earlier forecast.

If he's right - and if the Fed was misguided by bad statistics into following too tight a monetary policy - the nation will lose billions of dollars in output. The resulting reduction in federal revenues will far exceed what it would cost the government to gather more reliable statistics.

As an example of how far off statistics can be, Keran recalls December 1987 when economists on average, looking at then current GNP numbers, were forecasting growth (measured fourth quarter to fourth quarter) for the year of 3.2 percent. By the time GNP revisions were complete, the number came out to 5.5 percent - a change amounting to more than $100 billion.

Keran relies on statistics less subject to revision for his current relatively gloomy assessment of the economy. These include a decline in long-term interest rates and the monthly reports of purchasing agents showing a solid decline in business activity.

For what they are worth, the GNP numbers show consumer spending rising at a brisk annual rate of 6.2 percent, the strongest since the first quarter of 1988. That's up from a preliminary estimate of 5.8 percent.

One factor boosting GNP was the reduction in the trade deficit as a result of a weakness in imports. But slower import growth also meant that business inventories did not rise as much as the preliminary GNP numbers indicated. These two factors offset each other.

Inflation, as measured by a price index tied to GNP, climbed at a 2.9 percent annual rate in the fourth quarter. That is a dramatic improvement from the 5 percent rate in the April-June quarter. Keran anticipates further good inflation news in the current quarter and in the year ahead.

That's partially because he believes the productivity of labor and business have improved substantially. This would give the Fed more room to stimulate the economy - if it so chose - without reviving inflation.

If revised GNP numbers for the third quarter prove positive when they come out next summer, the expansion in the US will have just had its eighth birthday. But many economists are anticipating negative numbers for the present quarter. If the slump continues well into 1990, it will be defined as a recession.