Bad Policy Starts With Bad Numbers

A constant theme of the Clinton Administration has been "change." But how can policymakers take the nation in a new direction when they don't have a good grasp on current reality?

The problem is statistics. The government is still trying to measure a 1930s, industrial-style economy instead of a late-20th century information economy. It's good at tracking metal-bending stuff but miserable at discerning trends in the service sector.

So Washington collects as much data on mitten makers as on software writers. Worse, it does a terrible job measuring intellectual property. On computer software, for example, the government calculates its export value by pricing only the instruction manuals and the blank disks on which the programs are written. The actual software itself simply disappears. Ridiculous!

What does this mean? While President Clinton is discussing the "weak" dollar at the Group of Seven meeting in Naples, the seven heads of state will be looking at figures showing the U.S. merchandise trade deficit running at an unnerving $150 billion. But the reality is that exports are a lot higher and the deficit a lot lower. A $300 software program exported to Europe is listed at about $5 on the U.S. trade books. Total software exports probably alone amount to $4 or $5 billion, not counted in the official stats.

The government's lousy numbers cause real damage when it comes to measuring inflation. Practically the entire economics profession is unanimous in agreeing that the CPI, the consumer price index, generally overstates inflation by 1/2 to 11/2 percentage points. Officially, prices are rising at 2.4% this year--and so are expensive government cost-of-living allowances. If inflation is really going up at 11/2%, not 21/2%, Washington could save $50 billion over the next five years, a tidy sum to help pay for health-care reform.

What can be done to fix this mess? First, Congress and the President should take seriously statistics and their economic and financial impact. At the moment, statistical agencies are largely orphaned. Paying attention means finally hiring a director of the Census Bureau and spending the dollars needed to get government statistics up to date. A few million could go a long way.

Second, data collection is now divided among scores of offices and agencies throughout the government. In trade, for instance, the U.S. Customs Service collects the basic information, but the Census Bureau and Bureau of Economic Analysis share responsibility for reporting and analyzing it. This kind of bureaucratic feudalism is silly. At the very least, agencies should be allowed to share company-level information--cooperation that is now illegal in many instances.

Beyond that, Washington should consider a wholesale restructuring of its statistical agencies. Some of the work, such as collecting market and price data for some commodities, could be privatized. The rest should be turned over to a single government agency, much like Statistics Canada, a model system that in many respects puts the U.S. data collection and analysis to shame.

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