Wondering about yesterday's weather in Tallahassee? No problem, ask the Weather Bureau. Interested in who won last night's ball game? Just check the newspaper. Want to know how the economy did last month? Uh-oh.
The economic statistics that the government issues every week should come with a warning sticker: User beware. In the midst of the greatest information explosion in history, the government is pumping out a stream of statistics that are nothing but myths and misinformation. Most of the surging information economy--including software, telecommunications, and entertainment--is poorly covered by the data. While figures such as a 5.9% unemployment rate or a 3% inflation rate seem to have a reassuring solidity, in fact the connection between the government's statistics and the reality is getting more tenuous every year. Such official measures as gross domestic product (GDP), producer price index (PPI), or capacity utilization say far more about what's happening in old-line manufacturing industries than they do about such leading-edge companies as Microsoft, Disney, MCI, Fidelity, and Intel.
Based on BUSINESS WEEK analysis and recent economic research, it becomes clear that the real economy is vastly different from the one painted by the government's numbers. The true rate of inflation is at least a percentage point lower than the official consumer price index, which misses changes in consumer buying habits and gains from product improvements. Business investment in equipment, after adjusting for depreciation, is a full 30% bigger than the government statistics say, as the data only captures a fraction of the money companies spend on software and telecommunications equipment. And productivity growth is also twice the official rate, which omits rising productivity in most of the service sector. Taken together, this means that GDP growth is stronger than the numbers show, the nation's productive capacity is a lot higher, and the dangers of inflation are far lower.
NOT WORKING. A true picture of the economy would also show that Americans are right to be worried about jobs these days, despite the low official unemployment rate. Increasingly, the labor market is filled with surplus workers who are not being counted as unemployed. The rate of labor force participation--those working or looking for work--has dropped sharply for men since 1989. Estimated conservatively, some 1.1 million more prime-age male workers are out of the labor force compared with five years ago. Adding those workers back in would push the real unemployment rate to 6.8%, from its reported 5.9%. And there are at least 500,000 more workers with some college who have jobs but are underemployed compared to 5 years ago. That's an enormous pool of surplus labor available to fuel the economy's growth.
The bum numbers are costing the country a bundle. Wall Street, corporate executives, and Washington policymakers continue to make bad decisions based on the official--but incorrect--statistics. And the potential for harm is increasing as the statistics become ever less accurate.
The biggest danger right now is that the government's numbers are scaring the Federal Reserve and bond market vigilantes into raising interest rates. If overstated inflation numbers and misleading investment numbers induce the Fed to unnecessarily slow growth by 1%, that alone would take almost $70 billion off GDP.
OFF BY BILLIONS. Bad data hit the economy hard in other ways. Inaccurate readings of the economy's health may lead companies to unnecessarily cut back on production or to produce far more than is needed, at an average annual cost of some $5 billion. Because entitlements and income tax brackets are indexed to inflation, a 1% overstatement of the CPI raises the federal budget deficit by an estimated $5 billion each year. And an overstated CPI will lead workers and businesses to demand bigger wage and price increases, which will impose an additional $5 billion drag on the economy. This brings the toll from bad statistics to an average of some $15 billion a year in incorrect business and consumer decisions, according to BUSINESS WEEK estimates.
The shortcomings of the economic statistics also leave economists fumbling with major policy questions. Is real pay falling or rising? Is productivity improving or stagnating? Health care, trade, welfare reform, income inequality: All suffer from the same lack of accurate information. "You need the data to do real research on what interventions help and which ones don't," says Zvi Griliches, an economist at Harvard University. "The biggest problem is the system itself is not measuring the concepts people care about," adds Michael J. Boskin of Stanford University, head of the Council of Economic Advisers under George Bush. "And that divergence is growing over time."
Paradoxically, aggregate economic statistics are becoming more and more important even as they become less reliable. It used to be that a business could focus on a narrow set of markets. But as the economy gets more complex and interconnected, having good information about other markets and other parts of the country is essential. Businesses base their hiring and investment decisions, in part, on expectations of future sales and costs. Chrysler Corp., for example, tracks the different components of the producer price index to estimate what will happen to its cost of materials.
Consumers, too, worry about future income when deciding how much to buy or save. "People's expectations don't come out of thin air," says Thomas Juster, an economist at the Institute for Social Research at the University of Michigan at Ann Arbor. "Information that relates to job security and inflation-rate prospects has a significant cumulative impact on people's decisions."
Take inflation, for example. The CPI is used to set cost-of-living increases for union contracts and Social Security payments, while the producer price index is often used to index commercial contracts. Moreover, corporations and workers often use the published inflation rate as a benchmark for determining what constitutes a fair pay increase, even in the absence of an explicit contract. Some 76% of corporations use the CPI to help determine how much to increase their salary budget, according to a recent survey by Hewitt Associates. "The CPI helps determine people's ideas of what kind of raises that they ought to get," says Larry Ball, an inflation expert at Johns Hopkins University.
As a result, if the official CPI is stuck around 3%, that level of inflation can actually become a self-fulfilling prophecy, since workers want to keep up with inflation. But if the official number reflected the reality of a 2% inflation rate, that might encourage companies and workers to hold down wage and price increases, thus lowering the true rate of inflation. "If there is upward bias in the CPI and you get rid of it," says Ball, "people might be happier with smaller wage increases."
An inaccurate CPI number does other damage, too. Federal Reserve Board Chairman Alan Greenspan has argued that higher inflation slows productivity growth by making it more difficult for businesses to plan ahead. An overstated CPI, by raising the actual inflation rate, could cut one- or two-tenths of a percentage point off of productivity growth. That might not seem like much, but it is worth $5 billion to $10 billion of GDP.
As much as 10% of the ups and downs of industrial production is a reaction to economic statistics that are later revised, according to a recent study by Michael Waldman of Cornell University and Seonghwan Oh of Seoul National University. For example, says Waldman, a series of pessimistic economic data in the first half of 1989 helped persuade businesses and consumers to cut back on spending. The statistics were later revised up--but by then, the damage had been done, costing the economy at least $10 billion in lost production.
"PREHISTORIC." Companies are also concerned with the accuracy of the more detailed statistics that the government publishes about individual industries. Texas Instruments Inc., for example, uses industrial production numbers issued by the Federal Reserve to help project revenues for its various businesses. But Vladi Catto, chief economist for TI, has reservations about these numbers, which are calculated in part by measuring electricity consumption. "That's prehistoric," says Catto.
Catto also worries about the trade statistics, especially since the government no longer classifies exports and imports by whether they are for final sale or just for assembly purposes. So now it's not clear whether chips shipped to Korea will just be plugged into a VCR for sale in Topeka, Kan., or in Seoul. "It's much more difficult to trace exports and imports," says Catto.
The stream of economic statistics, whether accurate or not, helps shape the political climate in which policy is made. Take the debate over the 1988 trade law, which provided U.S. industries with new weapons to protect themselves against foreign competition. At the time, the merchandise trade deficit for 1987 was reported as $171 billion. Since then, revisions and the inclusion of services in the trade numbers have pushed the trade gap for 1987 down to $151 billion. "The debates over the trade bill would have been less vitriolic if people had had the real data," says Boskin.
Certainly, the problems in the numbers are affecting Fed policy. Not because Greenspan is ignorant of the potential biases: Indeed, he is the ultimate sophisticated consumer of economic statistics. As he recently testified before Congress: "The list of shortcomings in U.S. economic data is depressingly long." Publicly, Greenspan has often said the CPI probably overstates inflation, while in private he has been very critical of GDP figures, especially the quarterly figures. Instead, the Fed, when setting monetary policy, now relies heavily on anecdotal evidence of inflation, hiring, and other economic trends collected from the 12 regional Fed banks, and on reports from industrial, service, and financial companies.
But this dependence on anecdotes rather than statistics raises two problems. First, policy necessarily gets less precise when less information is available. Second, as psychologists and lawyers well know, anecdotes can be biased and colored by the viewpoint of the questioner. If the Fed is looking for examples of inflation and labor shortages, it will certainly be able to find them.
But no matter how many anecdotes the Fed says point to shortages, the evidence is overwhelming that the U.S. is not suffering from tight labor markets and other bottlenecks. True, the unemployment rate has now fallen to 1988 levels. But the index of help-wanted ads is down 25% compared with then, and jobs seem much scarcer. Just ask new college graduates. "Students are working a lot harder at the job search," says Laurie Paul, director of career services at Drew University in Madison, N.J. "They'll take a little bit less job and hope they can move up. These students are a lot more realist-
Indeed, a look behind the numbers explains the change in attitude. After a decade of stability, the labor force participation of men has fallen off sharply, and the erosion is continuing this year, despite the apparent surge in new jobs. At the same time, underemployment is on the rise for workers with some college or more. In 1988, 46.2% of these educated workers held executive or professional jobs. By 1993, the percentage holding good jobs had dropped to 43.5%, with the decline much larger for men than for women. Adds Wayne Wallace, director of the career resource center at the University of Florida: "Getting a job is not the issue. The real issue: Is it an appropriate job?"
And even the payroll employment numbers, which have been growing at a rate of about 300,000 a month, are suspect. Out of that growth, about 50% is coming from what BLS calls the "bias adjustment," which is simply a fixed number added to the published number each month to take account of the fact that the payroll number historically gets revised upward each year. So even if job creation were to slow sharply, it would not show up in the bias adjustment for months. "That's a little unsettling," admits Katharine Abraham, BLS commissioner.
RELIABILITY FACTOR. The situation with the CPI is, if anything, even worse. By now, almost everyone from Greenspan to the Congressional Budget Office to the BLS agrees that CPI growth is too high. "I think the inflation rate is overstated by 1%," says Boskin. The reasons for upward bias in the CPI include technical problems with the calculation of the index and sluggishness in picking up changes in consumer shopping behavior. But the biggest problem with the CPI is that it doesn't reflect the improved quality of many products. New automobiles, for example, are far more reliable than they were a few years ago, yet this increase in quality--which pays off in fewer visits to the repair shop--is not accounted for in the CPI.
The producer price index suffers from the same problem. The BLS reports that the price of such telecommunications equipment as fax machines, video-conferencing gear, and modems has gone up in recent years. But that's not the experience of anyone who has gone shopping for these products. Adjusting for the tremendous gain in power and speed, true prices have actually dropped substantially for most types of communications equipment. In the case of pharmaceuticals, the inability of the BLS to properly account for generics and new drugs means the PPI for prescription drugs is overstated by some 50%. And there is no price index for software at a time when software is the single largest nonlabor expense for some companies.
Similarly, high-tech investments such as telecommunications equipment and software are either undercounted or not counted at all in the official investment figures. To put it another way, the 10 largest software companies have a total market capitalization of $80 billion--yet, according to the government, none of them produces any investment goods.
If the government is undercounting investment, it's not surprising that the capacity-utilization numbers are skewed as well. In the past, companies added to capacity by building new plants, which the government could easily track. But now Corporate America is trying to boost capacity by reorganizing production, something that the official statistics have no way of measuring. By reducing cycle time and improving manufacturing yields, "in just the first half of this year we `created' an `invisible factory' the size of one of our $400 million fabs--without any new brick and mortar," says Texas Instruments Chief Financial Officer William A. Aylesworth. "That kind of thing is probably happening throughout the chip industry, and throughout Corporate America." TI's goal is to free up another factory's worth of capacity in the next year.
The globalization of many industries is also distorting the capacity-utilization figures, since the Federal Reserve's figures include only domestic factories. But increasingly, corporations move production back and forth across national boundaries. Consider GE's Electrical Distribution & Control business in Plainville, Conn., which makes circuit breakers, panel boxes, and other industrial gear. According to President and CEO Lloyd Trotter, back in 1989 his foreign customers were mainly supplied from U.S. factories. But now the percentage has reversed. Exports account for only 20% of overseas sales, while the rest come from foreign factories in places such as Singapore--which are not counted in U.S. capacity.
And what about the country's productivity statistics, which purport to measure how fast output per worker is growing? They are "a horror show," says Juster of the University of Michigan. "They're widely regarded as the worst kind of numbers that have ever existed." The government has no good way of measuring output in a whole range of industries, including banking, software, legal services, wholesale trade, and communications--all of which have invested heavily in information technology. And without a good number for output, the official productivity numbers for these industries are suspect. Most economists now believe that productivity growth in these industries is substantially understated by the government figures. As a result, overall productivity growth for the economy is understated by "something around the order of one-half to one percentage point a year," says W. Erwin Diewert, an economist at the University of British Columbia.
And the list of problem statistics goes on and on. The Fed no longer relies on M2 as a good indicator of where the economy is going. Nobody knows how much currency is held abroad. Even the savings rate is "terribly unreliable," says David Bradford, an economist at Princeton University. What the U.S. should be interested in, he says, is not savings but the creation of wealth--which the government statistics system is simply not designed to measure.
STOPGAP SOLUTION. To be sure, the primary agencies that generate national statistics have seen a moderate budget increase in recent years. That meant, for example, that the Bureau of Labor Statistics was able to update the monthly survey that generates the unemployment numbers. The BLS also has started collecting its monthly payroll employment figures from companies by letting them punch in the data by phone, which has drastically decreased the number of large revisions. The Census Bureau and Bureau of Economic Analysis were able to add service exports and imports to the trade numbers. And on Oct. 21, the BLS announced that it would revise the CPI starting in January, 1995.
But these initiatives, however necessary, address only a fraction of the problems with the statistics. The agencies do not have the funds needed to deal with an information-based economy. Indeed, they never caught up with the rapid growth of the service sector in the 1970s and 1980s. Even the upcoming CPI revision is only a stopgap solution--it will better track food prices but won't come close to capturing the gains from rapid technological progress. And it's coming far too late. "The CPI should have been redesigned years ago." says Janet Norwood, long-time commissioner of the BLS and now a fellow at the Urban Institute. "The American people don't want to spend money on statistics."
Still, there's a lot that could be done to make the official statistics more accurate. One easy change would be to reallocate statistics money from the Agriculture Dept., which gets far more than its share. Then the BLS, for example, could update the PPI more quickly. "We don't have a pot of new money for developing new indices in the service sector," says the bureau's Abraham. "If we had a few extra millions we could use it fruitfully."
But ultimately it will take hundreds of millions of dollars to get an accurate picture of today's economy. "It's one thing to collect data from huge factories," says Norwood. "It's another thing to go to a bunch of service establishments, which tend to be much smaller." And the problem is compounded in the information economy, where it's hard to even figure out how to value the product.
But it's well worth doing. American companies and workers are going through wrenching readjustments as they gear up to compete in the new world economy. Without good statistics, we don't know whether what we're doing is working. Better economic data will mean better economic policymaking by government, better decisions by investors and corporations and, ultimately, a higher standard of living for everybody--and we'll even be able to measure it.
Where the Statistics Dollar Goes FISCAL YEAR 1994* MILLIONS OF DOLLARS AGRICULTURE $294 DEPT. BUREAU OF LABOR 280 STATISTICS CENSUS 276 BUREAU NATIONAL CENTER FOR 135 EDUCATION STATISTICS ENVIRONMENTAL POLICY 107 ADMINISTRATION ENERGY INFORMATION 89 ADMINISTRATION NATIONAL CENTER FOR 88 HEALTH STATISTICS BUREAU OF ECONOMIC 44 ANALYSIS FUNDING FOR OTHER 1,346 FEDERAL STATISTICS TOTAL FEDERAL FUNDING 2,659 FOR STATISTICS PERCENT OF FEDERAL 0.2% BUDGET *Estimate DATA: OFFICE OF MANAGEMENT & BUDGET