Do Government's Economic Numbers Add Up? Not if you look at them one by one as Wall Street does

July 17, 1988, Sunday, NASSAU AND SUFFOLK EDITION; BUSINESS; Pg. 62

By Al Gordon

ON APRIL 14, reporters gathered at the Commerce Department’s Pennsylvania Avenue headquarters in Washington to learn the agency’s findings on the U.S. trade deficit for February. Under the department’s tight security procedures, no one could breathe a word of the report - or even leave the fifth-floor room - until 8:30 a.m. Then, after officials gave the OK, the media flashed the word that there had been a $ 13.8-billion deficit that month, up from $ 12.4 billion in January.

Despite cautions from government experts against making too much out of any one month’s numbers, the part of the nation that appeared to be awaiting the data most anxiously - Wall Street - took the news very badly.

With the trade figures in wide circulation among traders long before the 9:30 opening bell, the Dow Jones average of 30 industrial stocks plunged 45 points in the early moments of New York Stock Exchange trading. By the close, the Dow had lost 101 points, one of the worst drops on record. 

It was a dramatic demonstration of how economic indicators can influence the financial markets. And it was also a dramatic indication of the perils this poses.

Economic indicators, said Commerce Undersecretary Robert Ortner, the department’s chief economist, “are the raw materials for economic decision-making,” influencing matters as weighty as the Federal Reserve Board’s monetary policies and as personal as the shopping you do in the supermarket.

But economic indicators, like any other statistics, are subject to error, temporary distortions, seasonal variations, and revision in the light of additional information. They can be misleading, or they can be misread. And that means Wall Street or Main Street must be very careful in using the indicators - more careful than they have sometimes been.

“I worry about the mischief [this] can create,” said Courtney Slater, a Washington, D.C. economic consultant who studied the U.S. statistical system for a congressional committee.

The U.S. government has been collecting statistics for as long as it has existed - the census, in fact, is written into the constitution - with the present economic indicator system taking shape during the 1930s in the wake of the Great Depression.

On the strength of what is shown by the dozens of indicators the government releases each month, Washington shapes economic policies; businesses make marketing and expansion decisions; cost-of-living adjustments are made in labor contracts, government benefits and tax rates, and billions can be made or lost in the markets.

According to Janet Norwood, commissioner of the Bureau of Labor Statistics, each 1-percent change in the consumer price index, a measure of inflation, has a direct $ 4.5-billion impact on the federal budget alone, through increases in Social Security and other benefits and adjustments to tax rates that are automatically triggered by hikes in the index.

Right now, such basic questions as what your job prospects will be, what kind of interest rates you’ll pay, and whether you’ll be living in an economy that is booming or heading into recession hinge largely on whether public and private policymakers correctly read the economic signposts. Only last week, Federal Reserve Chairman Alan Greenspan said he saw signs of inflation, and the next day, the nation’s major banks reacted by raising their prime lending rate, a benchmark for other loan rates.

Yet despite all the indicators’ impact, few Americans know where all these numbers come from and what their strengths and weaknesses are.

In a report earlier this year, Congress’ Joint Economic Committee praised the “high quality” of the government’s statistical programs, which it called “models for other nations.” Most economists, while finding fault with particular indicators, agree that overall, the economic statistics are of high quality.

The government’s statistical agencies also are credited by congressional experts and the economic community with keeping their work largely free of political influences. They have been hurt, however, by recent federal budget cuts. The Bureau of Labor Statistics, for example, earlier this year took a 4.26-percent cut in its $ 220-million annual budget. That forced the agency to reduce the sample size for many of its surveys and to eliminate such reports as the monthly employment data for New York City and Los Angeles.

The Joint Economic Committee calculates that the $ 498 million the government will spend on all its statistical programs this year is less in real terms than was spent in 1980.

But no matter how good they are, because of the prospect of error or statistical fluke, there are limits to what can be learned from any one indicator and any one month’s data. “One month does not a trend make,” cautioned Norwood at the Bureau of Labor Statistics.

To truly spot economic trends, experts say, it is essential to look at indicators over time - three months is the generally accepted minimum - and to look at each indicator in relation to the others and check the “fine print” in each figure.

But the financial markets often have ignored the caveats. In the case of those February trade figures, for instance, the Commerce Department would later obtain more information that led it to revise the deficit down to $ 12.99 billion - $ 800 million less than the number that rattled Wall Street.

Moreover, subsequent monthly reports demonstrated that Wall Street had overreacted to a phantom problem. The trade deficit has been steadily, albeit slowly, narrowing this year. Using a three-month average - a methodology the department introduced with the April trade figures to track long-term trends - the average monthly deficit for April declined to $ 12.0 billion from $ 12.9 billion in January.

The markets always have reacted to economic news. But now, many experts think, the extent and purpose of their surveillance is markedly different.

Irwin Kellner, chief economist of Manufacturers Hanover Bank, said “a whole cottage industry has grown up” on Wall Street of forecasters who not only try to guess the economy’s direction but guess what every upcoming indicator will show.

Yet, economists say, these guesses typically go awry and serve little purpose other than to make an already volatile market even more so.

For example, the consensus forecast was that the June unemployment rate would go no lower than 5.4 percent and nonfarm payrolls would increase by no more than 300,000 employees. In fact, the actual unemployment rate was 5.3 percent and nonfarm payrolls grew by 346,000 workers.

“Once again, the majority of forecasters were faked out,” said Marshall Front, an economist with the Chicago investment firm of Stein, Roe & Farnham. “You would think that with their rate of failure, they’d give it up.”

The problem is compounded, Kellner said, by Wall Street’s “Alice in Wonderland” mentality: Good news is bad news, and vice versa. Thus, a glowing report on employment is seen by the markets as a source of gloom on inflation.

Slater accuses Wall Street of “making a game” out of economic indicators. If so, some would argue it’s a game with very high stakes. The April market plunge was only one instance of indicator-triggered market volatility. Perhaps the most notable example came last Oct. 14 when another trade report Wall Street didn’t like led to a 95-point Dow decline, which, experts later agreed, helped set the stage for the market’s crash five days later.

Being right or wrong about the economy, however, may no longer be the point.

“The game is to guess if the change in the indicators is enough to cause the Federal Reserve to change its behavior,” said Jerry Jordan, a former member of President Ronald Reagan’s Council of Economic Advisers and now chief economist for First Interstate Bank in Los Angeles.

A kind of dog-chasing-its-tail effect may be taking place, some economists believe, because while the markets are basing their actions on expectations about the Fed, the Fed may be tailoring its actions to avoid upsetting the markets.

Wall Street always has been guided by the presumption that if the Fed senses an inflationary burst, it will raise interest rates. And when interest rates rise, among other things, bonds become a more attractive investment than stocks.  Trade has joined inflation as a major obession because of the theory that if the trade deficit widens, the dollar will fall and the Fed will be forced to raise interest rates to keep foreign investment coming into the United States.

As David Wyss, head of Data Resources Inc., the Lexington, Mass.-based private forecasting firm, put it: “The squeaky statistic gets the oil.”

The key point, adds Jordan, is that an indicator is important to Wall Street if the markets think it is important to the Fed, regardless of its actual importance to the economy. “If the markets thought policy depended on whether the Fed’s governors eat green peas for lunch on Tuesday, they would be monitoring garbage cans and delivery trucks at the Fed,” he said.

Wall Streeters dispute the notion that they are playing mischievous games, arguing that they are only protecting their legitimate interests.

Alan Sinai, chief economist for Boston Co., a New York investment firm, said the markets have to look closely at the indicators because the economy’s behavior, and the central bank’s actions that could affect that behavior, are crucial to the investment climate.

Besides, says Alice Rivlin, former chief of the Congressional Budget Office, trying to outguess other traders is simply “the nature of Wall Street.”

For the Bureau of Labor Statistics, the Commerce Department’s Bureau of Economic Analysis, the Census Bureau and other agencies that give their readings on the U.S. economy’s vital signs several times a month, this all adds up to a headache.

The indicator game “puts an extra burden on the numbers,” said Chuck Waite, an associate director of the Census Bureau who is in charge of the unit that compiles the trade statistics.

“We wish the stock market didn’t hang so on every number we put out,” Norwood said. “It makes us worry more about being certain we are explaining things as well as we should.”

Because of their influence on the economy and on world financial markets, the numbers are closely guarded secrets until their release. After the raw data are compiled in the field and fed into agency computers, only a few select employees have access to the overall findings and strict release times are imposed on the news media.

There are no recent known instances of leaks outside the agencies, but the Commerce Department recently fired some employees caught using their inside information about the indicators to make trades on Wall Street.

For all the attention paid to the indicators by Wall Street - and by politicians and the news media as well - most Americans have only a hazy idea about how the data are gathered and what they really show.

With that in mind, here is a closer look at four of the most closely watched monthly indicators:

Merchandise trade. “Probably the worst figure the government publishes,” Kellner asserts.  Even Ortner isn’t totally satisified with it, although he says the problems “have been largely repaired.” The trade data are being “criticized without foundation,” he says, but “we’re still not quite finished with it. There is still room for improvement.”

Ironically, the trade figures theoretically should be among the most exact. Unlike most indicators, which are generated by surveys of a select sample, the trade data are an accounting of the nation’s actual total import and export activity as counted by U.S. Customs officials. The Census Bureau is respon- sible for compiling and analyzing the customs reports.

So what’s the problem? “Customs is not a statistical agency,” Waite said. Customs agents are chiefly responsible for collecting tariffs, ferreting out smugglers and other such duties, he said, and “statistics are not at the top of their priority order.” Moreover, the sheer volume of imports has grown faster than the ability of Customs to process its records.

As a result, paperwork from Customs was slow getting to the Census Bureau and, by 1985, as much as 55 percent of any given month’s trade data was actually delayed reports on activity that occurred in other months. Because the carryover was “big, variable and inconsistent,” Waite said, the month-to-month changes were essentially meaningless.

As a result of administrative changes made since 1986 by the Customs Service and Commerce, including a delay in the release date for the indicator, the carryover rate was cut to less than 3 percent per month last year, he said.

A second problem with the trade figure is that it tends to undercount exports. Customs services are set up around the world to control what comes into their borders, not what goes out.

This problem was particularly severe concerning U.S. exports to Canada, our No. 1 trading partner. At many border points, the U.S. export-reporting system essentially calls for truck drivers to leave off a form at an unmonitored checkpoint - a method that results in something less than total compliance.

In 1986, Waite said, the Census Bureau estimated that $ 10 billion of U.S. exports to Canada - about 10 percent - went uncounted. To solve the problem, the United States reached an agreement with Canada to swap data.

Showing confidence that these problems have been resolved, Commerce last month resumed a seasonal adjustment of the indicator after a two-year suspension.

Notwithstanding what Slater calls the agency’s “heroic effort” to improve the figure, most economists remain leery of it because of its high volatility. A large order of a “big ticket” item such as an airplane, or shipping delays or a host of other factors can distort any month’s results.

Even on a seasonally adjusted basis, the size of the trade deficit has changed by more than 20 percent most months this year. That’s why the government is now computing the three-month averages for the trade figure; the computation smooths out the monthly zig-zags.

Waite cautions, however, that there is no single “best” figure on trade. The three-month average is a guide to long-term trends, but “if you want to know how much we need to borrow to finance our trade deficit,” he said, the raw, unadjusted data are what count. “You don’t borrow on a seasonally adjusted basis,” he said.

Leading indicators. Maligned by some economists, valued by others, Commerce’s index of leading indicators is the government’s principal economic forecasting tool.

It is a composite of 11 other indicators culled from information on: average workweek, average weekly initial unemployment-insurance claims, new orders for manufactured goods, vendor deliveries, plant and equipment orders, building permits, changes in inventories, changes in sensitive materials prices, stock prices, money supply and changes in business and consumer borrowing.

“Leading” is the key word here. James Stock, an economics professor at Harvard’s Kennedy School of Government who is researching the index, said that to be included in the index, economic numbers must do more than represent an important part of the economy. To be a leading indicator, the number must be one of the first to detect speedups or slowdowns.

A buildup of inventories, for example, tends to foreshadow a slowdown, because companies normally reduce orders of new goods - leading to production cutbacks at their suppliers’ factories - until the inventories are reduced.

What counts is the index’s direction, with three consecutive months of decline generally accepted as a signal of recession. Historically, no correlation exists between the extent of the decline and the severity of the recession.

But says Professor Donald Ratajczak of Georgia State University, by the time the government gets through revising the measure, “the index isn’t really leading anymore.”

The inventory and credit components are not available when the initial index is released. So, the agency later releases a revised index that includes those two components and later figures on the other nine. As with most indicators, the tentative figures are released so that the report will be timely, but the price of timeliness is inaccuracy. Revision of the leading indicators often results in a substantial change; sometimes even a total reversal of direction.

The December index, for example, initially was reported as a 0.2-percent decline, but the revisions showed it actually gained 0.4 percent. Moreover, since the index declined in November and January, th revision had the effect of breaking a string of three consecutive declines.

Thus, there is an inherent five-month time lag in the index. If the signs of a downturn first appeared in January, say, it would not be until May - when the revised March index would be released to confirm a three-month decline - that one could confidently conclude that the index had warned of a downturn. But by then, economists say, it’s probably too late for policymakers to do anything about it.

Columbia economics Professor Geoffrey Moore, a nationally recognized authority on leading indicators, said that in addition to the time lag, the index also suffers from a failure to keep up with changes in the economy. “The economy is not the same today as it was 10 years ago,” Moore said.

He is developing a revised, 15-item index that he believes more accurately signals movement in today’s economy. He would add four new components - bond prices, new layoffs, business starts and failures, and a ratio of prices to labor costs.

In addition, Moore would turn to new public and private sources for data on inventories, vendor deliveries, sensitive materials prices, housing and business and consumer credit. These are nearly as accurate measures as the old ones but are much quicker, Moore said, reducing the lag time.

One problem that Moore said can’t be easily fixed is the index’s focus on goods-producing indicators. Services now constitute about 70 percent of the U.S. gross national product and are too important to ignore, he said. But service industries simply don’t show the same kind of cyclical changes as manufacturing and thus don’t fit into leading indicator calculations. To solve the problem, he said, it may be necessary to develop a separate index for services.

Moore’s work is not a mere academic exercise. Ortner said Commerce is looking closely at the Columbia research and is considering some revisions of its own. The expectation in economic circles is that it’s just a matter of time before an overhauled index of leading indicators is adopted.

Consumer price index. This indicator is the ultimate exercise in comparison shopping.

Each month, the Bureau of Labor Statistics representatives check prices at about 19,000 retail establishments - department stores, supermarkets, gas stations and the like - in 85 urban areas, as well as monitor rents for about 57,000 housing units.

The New York metropolitan area is surveyed and an index for the region computed each month. Smaller urban areas are rotated in the sample.

They are checking for prices on what the agency calls its “fixed market basket” of goods and services the average consumer might purchase. Essentially, this is the world’s most detailed shopping list.

It consists of 72 major categories (fresh fruits, for example, or furniture and bedding.) Each is then broken down into subcategories. Thus, the fresh fruits group is broken into apples, bananas, oranges and other.

Each of these items is then given a specific weight - the average price of bananas, thus, makes up 0.059 percent of the index. The weights are determined by a separate survey of U.S. households’ buying habits, which usually is taken once every 10 years.

This complex breakdown makes it possible to determine if prices are rising or falling across the board or only in a specific area - in recent years, the biggest price changes have involved energy or food. Without looking at this detail, one can’t fully assess the extent of inflationary pressures and what needs to be done about them.

The total average price of all the items on the shopping list is computed as a comparison to the average price levels during a set time period, currently 1982-84. (That period was chosen because the latest survey of consumer buying patterns was for the 1982-84 period.) Thus, the May CPI of 117.5 means that the shopping list for May, 1988, was 17.5 percent more expensive than the same market basket cost in 1982-84.

Although limited to urban areas, the CPI covers about 80 percent of the U.S. population and is generally one of the most highly regarded indicators.

But nothing is perfect. The revision of the index to reflect the 1982-84 buying patterns was not completed until last year. Many economists, and members of Congress, complain that the index isn’t being changed fast enough to provide current information on the cost of living.

“If you’re not continually updating it, there will be distortion,” Ratajczak said. Kellner complains that the fixed market-basket methodology means that the index can’t detect signs of consumer price resistance that might moderate price hikes. For example, if the price of beef goes up, supermarket shoppers might buy less of it, turning to something cheaper, chicken, for instance. But the CPI would continue to be computed as if beef consumption were unchanged, thereby indicating that Americans were spending more at the grocery story than they actually were.

Norwood said the fixed market basket technique is intentionally designed “to separate price changes from changes in the standard of living,” so constantly revising the market basket would blur data on prices.

Still, “recognizing that the world keeps changing,” Labor Statistics has decided to constantly monitor buying patterns to see if it needs to revise the market basket sooner than in 10 years, Norwood said.

Beyond the question of market-basket adjustments, the major controversy surrounding the CPI involves not the indicator itself but rather the uses to which it has been put.

In addition to Social Security and federal programs that are linked to the index, thousands of workers in the private sector are covered by contracts tied to the indicator, as are a wide array of deals ranging from apartment rents to child-support payments.

Although the moderation of inflation in recent years has moderated debate over such “indexing” of the economy, economists remain concerned that such links build inflation into the economy.

Employment. It’s not generally known outside of economic circles, but the government actually takes two monthly soundings of employment, providing two looks at the job picture.

Each month, the Census Bureau, on behalf of the Bureau of Labor Statistics, surveys 55,800 randomly selected U.S. households to check on the employment situation of each household’s members. Separately, the bureau, working with state labor agencies, checks the payroll records of about 300,000 establishments.

According to Samuel Ehrenhalt, the New York regional commissioner of labor statistics, about 3,400 households are surveyed in New York state each month (including 1,300 in New York City) and 20,000 employers (10,000 in the city). Data on New York and the 10 other largest states is reportedly monthly along with the national figures.

The nation’s best-known employment statistic is the unemployment rate, which is generated by the household survey. However, most economic forecasters prefer to examine the increase in the nonfarm work force reported by the payroll study.

Because the samples aren’t the same, and the timing of the data differs slightly, the two surveys can produce contradictory results. For example, in May, civilian employment fell by the household survey’s findings, but the number of jobs counted in the payroll survey rose.

The major qualm among economic forecasters is that the unemployment rate was set up to meet social, and to some extent political, needs. It is designed to help policymakers assess whether every American who wants a job has one. Although the Labor Department has sought to base this figure on tangible, measurable activity, it is still, Norwood concedes, “somewhat subjective.”

It is relatively easy for the government to establish who is working, but finding out who is in the work force is another matter. To be in the work force, one must either be employed, or available and actively looking for work. Thus, someone who is able to work but chooses not to look for a job is not counted as part of the labor market.

This can create many anomalies. For example, someone who doesn’t really want to be employed and is only going through the motions of job-hunting gets counted in the labor force, whereas a person who truly desires a job but isn’t looking because of, say, a child-care problem, goes uncounted.

And, the most-debated issue, people who drop out of the labor force because they consider the job market hopeless aren’t counted in the unemployment rate (although the agency does try to count the number of “discouraged workers”).

Moreover, the unemployment rate tends to be influenced by social and demographic changes as much as economic ones. The passing of the baby boom has reduced the rate of new entrants into the labor force, thereby easing the unemployment rate. On the other hand, the percentage of the population in the work force has jumped to 65.6 percent last year from 59.6 percent in 1967, reflecting a rise in two-earner families and greater participation in the labor market by minorities. Both factors tend to increase the unemployment rate.

In contrast, the nonfarm payroll figure has no subjective complications. It tells you exactly how many workers are drawing paychecks. And the month-to-month change tells analysts whether businesses are expanding and hiring workers.

When the number of jobs is growing, the magnitude of the increase is viewed as an indicator of possible inflationary pressures. Rapidly increasing demand for workers often leads to an increase in wages, which, in turn, puts pressure on prices.

To business economists, the nonfarm payroll directly measures companies’ actions, Sinai said, and so is an important “signal of what growth will be.”

Ratajczak warns against making too much of any employment figure as a forecasting tool, however, because “employment is a lagging indicator.

Economic policymakers and analysts, he said, “get the most concerned about it after the damage is already done.”

Copyright 1988, Newsday Inc.