Go Figure

Government numbers often don't add up.

The government depends on chains of data and numbers for many of its tasks, from calculating cost-of-living hikes to cobbling together international trade figures and collecting tariffs. And if a chain is only as strong as its weakest link, then the government's chains of reasoning are ready to snap at many junctures.

A recent series of studies has cast doubts on the accuracy of many varieties of government data, some with huge consequences and some only minor. All, however, raise troubling questions about the quality of information the government uses to make policy.

Perhaps the most important trouble spot is the Consumer Price Index, used to measure inflation and to set cost-of-living increases in federal benefits and inflation adjustments in federal income taxes. Analyses from the Congressional Budget Office and the Federal Reserve show the CPI overestimating inflation by anywhere from 0.2 to 1.5 percent. Such inaccuracies in the CPI can be expensive: If the CPI is overestimated by 0.5 percent annually, it will add $47.8 billion to the deficit over the next five years, according to Office of Management and Budget estimates.

Errors and deficiencies in U.S. trade statistics can make for unnecessary policy worries. In 1990, the General Accounting Office reports, there were fears of a capital drought in the United States, with a purported net $5 billion outflow in private capital. However, statistical discrepancies for that year's figures add up to $73 billion, making any conclusion about the real situation difficult. The GAO says that trade statistics very possibly undercount U.S. exports in general.

There are problems on the import side as well. The GAO went over reams of data on unit value of imports and found errors in more than half of the transactions they studied. In one case, for in stance, the GAO found a range in values for wood dowel rods in which the highest unit value was 952,250 times the value of the lowest.

The GAO also found, in a separate report on the Environmental Protection Agency's collection and management of its scientific data, that "many of EPA's scientific data sets are either incomplete, obsolete, or missing altogether" and that "the agency has long relied on outdated scientific information…some criteria (e.g., metals criteria) are overly stringent and based on outdated science."