Trying to Nail the Economy Down With Rubbery Statistics

The U.S. economy cruised ahead in the fall, lost power in the winter and clicked back into first gear last spring.

No, scratch that. The U.S. economy streaked ahead in the fall, slammed on the brakes after New Year’s and picked up modestly in the spring.

Or did it, perhaps, do something different altogether?

Strangely, no one is quite sure. The official statistics that shape the nation’s view of its economy, influence policy decisions, dominate the evening news and electrify Wall Street traders often miss the mark--sometimes dramatically.

Highly publicized reports are routinely revised. Just Tuesday, new figures from the Commerce Department suggested that the 1990-91 recession was milder and the recovery more vigorous than depicted last year. The next day, the White House slashed an earlier forecast for growth this year.

Consumer inflation reports, by contrast, typically aren’t corrected, although many analysts believe misleading U.S. price data lured the Federal Reserve Board toward an errant boost in interest rates a few months ago.

From specialized gauges of home sales and exports to overarching measures of jobs and growth, the story is much the same.

“We not only have a hard time telling where the economy is going, but we have a hard time telling where it is--and we’re not sure where we were a few months ago,” laments Michael J. Boskin, who was chairman of President George Bush’s White House Council of Economic Advisers.

In large part, the culprit is the $6-trillion economy itself, a relentlessly evolving mishmash of goods and services that challenges the most capable of bean counters.

Small enterprises come and go, invisible as minnows in a murky sea. Services and technologies now flourish that data crunchers never dreamed of. Globalization, advances in quality, changing business practices and the far-reaching underground economy all conspire to throw off the government’s best guess of what’s going on.

“Hundreds of billions of dollars get based on these statistics, which--for those of us who know how unreliable they are--is a joke,” said James F. Smith, an economist at the University of North Carolina, Chapel Hill.

Actually, the Commerce Department, the Bureau of Labor Statistics and the Census Bureau, which track many economic vital signs, are widely credited with doing a first-rate job--within the limits of their budgets. The government spends more than $600 million a year to take the economy’s pulse, according to the White House Office of Management and Budget.

Nonetheless, the job is harder than ever, and the zigzags of statistics add a note of doubt to what’s happening in the real world of factories, offices, shops and services, many analysts agree.

For instance, Bush was widely accused last year of overselling the national recovery, a charge buttressed by some of the data available at the time. But Tuesday’s findings suggest the economy was picking up steam throughout 1992 and racing forward by autumn--more in line with Bush’s claims.

“If I were George Bush, I’d demand a recount,” declares Ross DeVol, an economist at the WEFA Group in Bala-Cynwyd, Pa.

Virtually every week, the media turn a new statistic into a prominent news story. When the revisions come later, even whopping changes may get little notice. To cite just one example, a tepid gain in May retail sales ballooned sevenfold by the time the final data came out--in an obscure footnote to July’s report.

By design, initial estimates often are a hybrid of sketchy evidence and projection; the fuller, more-textured picture becomes available in the following months. News stories rarely emphasize such limitations, however, and an army of private economists offers authoritative-sounding commentary on the preliminary reports.

“It’s very disconcerting to see people take a single piece of data and assume it means much more than the people who produced the data thought it means,” said Janet L. Norwood, former commissioner of the Bureau of Labor Statistics (BLS) and now a senior fellow at the Urban Institute, a Washington think tank.

Many people don’t want to wait for more detailed evidence, however. Public interest in economic statistics has soared in recent years, along with an expanding investment industry and the spread of electronic news services that offer a rapid-fire volley of the latest numbers--all against the backdrop of growing concerns about the U.S. economy.

“I’d like to see my own publication and others report the numbers a little more skeptically,” allowed Stephen H. Wildstrom, senior news editor in Business Week’s Washington Bureau. “But you can’t stop reporting them just because they may be wrong.”

Revisions, he added, deserve much more attention in news coverage than they often receive.

When the Commerce Department asked analysts if it should delay unveiling its advance reports on retail sales to make them more reliable, the answer was no, recalled Irving J. True, chief of the Census Bureau’s monthly retail sales branch, which collects the data. For all the doubts about the advance report, “they want to keep getting the earlier numbers,” True said.

Speed can be risky. Until early this year, officials had underestimated new home sales in the initial, monthly report for at least 14 straight months. (Number crunchers, trained to focus on building permits, were led astray when builders started selling more houses before obtaining permits.)

In light of such mishaps, experts say the safest way to interpret a new number is to view it as one fragment of a larger jigsaw puzzle. Only as varied pieces of evidence are fitted together does a reliable picture gradually emerge.

“It’s always risky to take these figures at face value,” said Michael Metz, chief investment strategist at Oppenheimer & Co. “The revisions are often astonishing.”

But professional investors have their own uses for the data, which may have little to do with economic forecasting. “It’s gamesmanship,” said Metz. “You’re trying to outsmart the other traders.”

All of which can lead to the peculiar Wall Street exercise of calculating how others are calculating a statistic’s effect on the economy, however tentative the statistic may seem.

“Even if you don’t believe the number, you’re guessing how other people are going to respond to it,” explained Gary Schlossberg, an economist at Wells Fargo Bank in San Francisco who advises bank traders each morning.

In extreme cases, the twisting and turning of economic statistics can alter the view of recent history.

It happened in June, when the BLS formally announced that a stunning decline of 640,000 jobs it had reported from April, 1990, to March, 1991, never happened.

A key to the mystery was a change in business practice. Up to half a million U.S. employers turned to private payroll processors in the 1980s. In many cases, however, data transmitted to the government contained accidental overcounts of employees.

In 1991, when the payroll firms refined their computer software, the overcounting apparently stopped. The new, more accurate count, however, registered as a drastic decline from the earlier, inflated count.

Confusing? It took BLS officials several months, in cooperation with the American Statistical Assn., to unravel the puzzle. And some analysts still aren’t satisfied.

“It certainly was a very difficult problem to explain to the public--even to sophisticated data users,” said George S. Werking, a BLS division chief. But, he maintains, BLS learned more about its data sources through the painful exercise.

Separately, the bureau has been introducing an electronic call-in system for employers to use in its monthly payroll estimates, instead of relying on them to mail in forms. Over the long haul, officials are counting on such innovations to give a better reading of what’s going on.

The broadest gauge of all--the gross domestic product, a quarterly calculation of the value of goods and services produced by the nation--suffers from limited information about the economy it seeks to measure.

Key facts trickle in from hundreds of sources. There is no comprehensive survey of the growing service sector. Hospital and utility revenues, for example, are received two to three months after the initial growth estimate is made public.

Beyond that, statistical booby-traps abound.

It was easy to count the bushels of wheat, tons of coal and other tangible products that once were the nation’s stock-in-trade. Today’s high-tech goods and services present new problems. Price declines--especially in high tech--often reflect gains in efficiency that can be hard for statisticians to compute properly. Similarly, a price hike might be misinterpreted as inflation, when it really reflects a quality improvement.

“Counting the boxes doesn’t do very well,” said Allan H. Young, chief statistician at the Commerce Department’s Bureau of Economic Analysis.

Services, from electronic financial information to white-collar enterprises, present other challenges. “How do you measure the output of a lawyer?” Young asked.

The reason all this matters to more than statistics junkies is that misleading data can lead to mistakes in national policy.

Decisions that affect virtually everybody--interest rate levels, public works spending, taxes--are influenced by the official take on the economy.

“If somebody says you have a tummy ache and you have appendicitis, you have a problem,” said Richard B. Hoey, chief economist at the Dreyfus Corp. With more precise data, he believes, policy-makers would have better appreciated the “credit crunch” that hit the economy in the late 1980s as lending dried up.

Slippery statistics have obscured public debate on a range of other issues, from trade to saving habits. U.S. exports are widely believed to be undercounted, perhaps by tens of billions of dollars. The U.S. savings rate once was revised upward by more than 50% for 1985, 1986 and 1987.

Inflation is the latest example of the link between policy and dubious data. Earlier this year, after the consumer price index showed a pattern of increases, the Federal Reserve Board signaled that it was close to hiking interest rates. As a result, public worries mounted, investors hoarded gold and business executives braced for increased borrowing costs.

Now, though, some analysts believe the official price data led the Fed to the verge of a dangerous, unwarranted policy.

Their argument: The U.S. economy no longer resembles the world as envisioned in the consumer price index, where people generally pay retail prices and stick to a consistent “market basket” of purchases.

Rather, discount buying is pervasive, they argue, and consumers can easily substitute one type of fruit juice or item of clothing for another if they want a lower price. Some also maintain that quality increases erroneously add about a percentage point to the CPI.

“It would have been a tragedy if the Fed raised interest rates, with the economy as weak as it is, to fight this phantom inflation,” said Nancy J. Kimelman, chief economist at Technical Data, a Boston financial advisory firm.

With these sorts of problems in mind, Bush adviser Boskin pushed for funding to improve data gathering, and Congress approved several extra millions of dollars in 1991 and 1992. But continuing bids for more money, more recently by the Clinton Administration, have stalled.

Economic statistics, said one Administration economic official, is an orphan with lots of “friends” but few real champions when it competes with other budget needs.

“Our citizens deserve an accurate picture of where the economy is relative to other economies and relative to its own past history,” maintains Boskin, now at the American Enterprise Institute in Washington. “We’ve got a long way to go.”