Bad Data Can Lead To Bad Forecasts And Grim Results

The welter of economic statistics that emanate from Washington each month provide vital information to business and the public. But because many such initial reports are based on incomplete data and subject to heavy revision, they inevitably entail a risk that economic decisions will be based on faulty information. And if a protracted series of reports are biased in one direction, the impact on the economy could be substantial.

That, in fact, is exactly what seems to have occurred in the period prior to the start of the current recession, according to economists Michael Waldman of Cornell University and Seonghwan Oh of the University of California at Los Angeles. The two researchers examined the impact of a string of eight erroneous leading indicator reports, starting in January, 1989, that proved to be too low after final revisions were in.

Waldman and Oh found that these reports had a significant effect on the views of economic forecasters, accounting for a quarter of the shifts in their monthly forecasts. More important, their econometric model shows that the reports had a substantial impact on the level of manufacturing output two to four quarters after they were issued.

Specifically, the economists estimate that factory output in the third quarter of 1990, when the current recession began, was some $10 billion lower than it would have been if the reports issued in 1989 had been accurate. "Although the damage to economic perceptions caused by faulty initial reports may not have been enough to tip the economy into recession," says Waldman, "it clearly worsened the downturn."

Before it's here, it's on the Bloomberg Terminal.