U.S. Economic Policy Held Skewed by Flawed Statistics
An 18-wheel truck laden with Detroit-made auto parts crosses into Canada near Buffalo, N.Y., bound for an assembly plant outside Toronto. It is night and raining, and the driver, running late, does not bother to deposit in the unattended drop box at the border the required report of his cargo.
A few weeks later, an auto-transport carrying cars largely assembled from those same parts crosses back into the United States at Port Huron, Mich., and a new batch of imported cars is duly reported by the Customs Service. As a result, the U.S. trade balance drops another notch, helping fuel protectionist passions in Congress--and the next estimate of the nation’s economic output is understated by at least one truckload of auto parts.
This is no isolated episode. Experts say that flawed statistics are plaguing the government’s efforts to measure activity not only at the Canadian border but throughout the nation’s $4-trillion economy. Consequently, government estimates of economic growth, inflation, unemployment, the trade deficit--all of them and more suffer from major distortions.
“We have been seriously misjudging the character of this economy,” said Joseph W. Duncan, who ran the statistical policy division of the Office of Management and Budget under Presidents Gerald R. Ford and Jimmy Carter and is now corporate economist and chief statistician with Dun & Bradstreet.
The cumulative impact of all the statistical quirks is far from academic.
Congress makes trade policy according to the latest data about the ballooning U.S. trade deficit. The government’s consumer price index directly determines annual Social Security cost-of-living increases and many private wage adjustments. Congress makes tax and budget policy according to whether the economy seems, from the latest data, to be growing or stagnating.
There is virtually no way to measure whether the errors in data have translated into errors in policies. But critics are convinced that they have--and that the mistaken policies have damaged the nation’s economy.
“America’s lack of information about itself has increasingly become an impediment to national economic growth,” the Joint Economic Committee warned last month after a series of hearings on government statistics.
Authorities lay part of the blame on budget cuts exacted by Presidents Carter and Reagan in the government’s data-gathering bureaucracy. But no amount of money would help the experts overcome their conceptual barriers to measuring today’s economy.
In a shopping mall of a Washington suburb, Gerald F. Donohoe, a senior Commerce Department economist, recently tried to replace his decrepit alarm clock but found nothing that would merely wake him up. Instead, he bought a model that included an AM-FM radio, a snooze control and a 24-hour digital display. It cost $15.
Fascinated, Donohoe, who is responsible for producing statistics on national income, spent part of his spare time for the next several weeks leafing through old Sears catalogues, tracking other commonplace products that have recently turned into low-priced technological marvels.
The Commerce Department’s Bureau of Economic Analysis treats such items as stable in price and static in quality, although the consumer today is obviously getting more value for his dollar from such products. Thus, the government underestimates economic growth and overestimates inflation, by amounts that can only be guessed at.
The government’s statisticians have not been able to keep pace with what Robert Ortner, the Commerce Department’s chief economist, called “the dynamic, flexible economy.”
Duncan said: “There is a conceptual lag, the failure of the system to respond to change.”
‘The Name of the Game’
Kenneth V. Dalton, who heads the office of prices and living conditions at the Bureau of Labor Statistics, contends that “quality adjustment is the problem. In terms of price indexes, it really is the name of the game.”
The government is making some progress. Under Ortner’s and Donohoe’s supervision, the Commerce Department recently developed a means of accounting for the accelerating quality improvements in mainframe computers and their 15-year record of 14%-a-year price declines.
But Ortner admits that only the surface of the statistical problem has been scratched as the economy plunges into a new age of proliferating services and ever more sophisticated techniques for producing goods.
Duncan estimates that overall economic performance may be understated by 5% to 15% because of what he believes is a growing inability of official government data collection to cope with accelerating structural changes in the economy.
That would mean that economic output this year will be $200 billion to $600 billion more than the officially estimated $4 trillion. “That is enough,” Duncan said, “to change this recovery from anemic to relatively strong.”
In countless government offices, specialists use word processors and centralized data banks to churn out up-to-the-minute reports, clear examples of increased productivity in the bureaucracy.
But, as far as Labor Department statistics are concerned, government bureaucrats are no more productive now than they have been for years. This is because the Labor Department measures the work output of the bureaucracy simply by counting the number of bureaucrats at each pay level.
Productivity is not so hard to measure in the economy’s manufacturing sector. “You have concrete outputs that can be measured against concrete labor inputs: goods produced against hours worked,” Ortner said.
In the service sector, by contrast, nobody knows how to measure the output of bankers or insurance adjusters or store clerks, much less of government bureaucrats. “Productivity, in particular, is probably understated everywhere, but especially in services,” Ortner said.
Workers Counted
So economists do for the entire service sector what they do for government workers: They assume that, if an office has twice as many workers as another, it does twice as much work. Consequently, the service sector has shown little or no improvement in officially measured productivity.
“Everybody agrees that this is wrong,” said John E. Cremeans, head of the Commerce Department’s office of business analysis. “And everybody agrees that no better way has been worked out.”
The problem hits particularly close to home for the nation’s private economists. Roger Brinner of Data Resources Inc., an economic forecasting firm that has used small and affordable minicomputers to vastly increase its analytical powers, said: “As far as the Labor Department is concerned, we haven’t improved our productivity since 1975.”
And automation is making productivity harder to measure even in the manufacturing sector. “Now that we are moving more and more toward total automation,” Ortner wondered, “how will we measure labor productivity if the factory is mostly laborless?”
The Labor Department measures total employment in two ways. It surveys a sample of 59,500 households each month to determine the unemployment rate, and it analyzes monthly reports submitted by 250,000 employers to estimate growth in the number of jobs. By the first measure, 111.4 million Americans had non-farm jobs in June. By the second, only 99.8 million did. Which should you believe? The Labor Department is not entirely sure.
This much is clear, however: Economists are increasingly uneasy with the Labor Department’s monthly employment report. Ortner says that the official unemployment rate is suspect because it ignores the changing profile of the U.S. labor force.
Many more Americans are in the labor force today--that is, they have jobs or are looking for jobs--than in the past. As women and young people have flooded into the labor force, the so-called labor participation rate has jumped from 60.8% of people 16 and older in 1969 to 65.8% today.
If the labor force had been as big a proportion of population in 1969 as it is today but there had been no more jobs, that year’s extraordinarily low 3.4% unemployment rate would have been 10.8%--far higher than today’s 7%. On the other hand, if today’s labor participation rate were as low as it was in 1969, there would be 585,000 more jobs than job-seekers.
“If you want to measure the strength of an economy, you should look at the total number of jobs created, not simply at the unemployment statistics,” Ortner said. But that approach has drawbacks of its own--the number of new jobs in the last year is 1.8 million, according to the household survey, but only 1.5 million, according to the employer survey.
