GDP Growth: Are The Numbers Too Rosy?

Forget faulty jobs data. An overstated GDP may help explain the economic reality gap

The Stat
The economic statistics collected by the U.S. government are probably as honest and accurate as any in the world. Perfect, though, they're not. Collecting data about the American economy is a boundless challenge on a limited budget. And it's even harder in periods of dramatic change -- like today.

Take the puzzling case of stagnant hiring. Economists have been scrubbing the Labor Dept.'s employment surveys to account for the gap between strong reported growth in gross domestic product and the anemic 61,000-a-month pace of payroll job creation since August. But they've been looking in the wrong place: There's fresh evidence that GDP, the most closely watched economic statistic, may be overstated -- and that lackluster job prospects may present a truer picture of what's happening now than the strong growth number.

That evidence is popping up in several measures, including income and manufactured imports. But the most intriguing is summed up in one figure. From 2002 to 2003, U.S. imports of services from abroad, adjusted for inflation, didn't grow at all -- a nice, round 0% change. That comes after only 1.4% reported growth in 2002. That seems implausible. Take one key component of service imports: U.S. companies' transfers abroad of a range of service work, including computer programming, customer support, tax-return preparation, and research and development.

While still small, such "offshoring" is rapidly growing. Yet it's not fully accounted for in government statistics -- an overlooked case of the government getting the numbers wrong. Just one example: The Commerce Dept. reports that Americans paid just $209 million in 2002 to unaffiliated companies in India for business, professional, and technical services. But five big Indian tech-service companies -- Tata Consultancy Services, Infosys Technologies (INFY ), Wipro Technologies, Satyam Computer Services (SAY ), and HCL Technologies -- report that their sales to North America that year were around $2.4 billion. Presumably not much of that came from Canada or Mexico.

In national accounts, imports reduce GDP. So if service-sector imports are undercounted, GDP growth is probably overstated. That, in turn, could explain part of the gap between reported GDP growth and the much weaker gains in payrolls.

By themselves, service imports are a small portion of the economy. At most, any error on service imports would only add a tenth of a percentage point or so to GDP growth. But signs of a mismeasurement in that category give reason to think that broader miscalculations could have overstated recent GDP growth by as much as a full percentage point a year. That's meaningful: A difference of one percentage point in GDP growth equals about 60,000 jobs a month.

Where would the bigger gap arise? Goldman, Sachs & Co. (GS ) economist Jan Hatzius worries that goods imports -- which outweigh service imports by a factor of five -- are also underestimated. The official numbers say that strong U.S. demand is being satisfied by a surge in domestic goods production -- up 8% after inflation over the past year. But that's far higher than the 1.4% increase in the Federal Reserve's measure of industrial production. Normally, Hatzius notes, the two numbers track each other more closely.

Rapid structural shifts in the domestic economy may be making growth estimates seriously unreliable, just as they were in the mid-1990s. Then, the official government figures failed to pick up the information revolution and accelerating productivity: First reports said productivity grew 0.7% in 1996, a figure that was later revised up to 2.5%. Back then, the most accurate growth gauge came from wage and profit data. By noting a big profits jump, Fed Chairman Alan Greenspan and others spotted the productivity boom before it showed up in the GDP data.

PALTRY-TO-ZERO GROWTH. Now the reverse may be happening. Gross domestic income (GDI) -- the sum of wage, profits, and other income -- has been growing more slowly than GDP, even though the two in theory should be equal. Adjusted for inflation, the income measure rose just 2.6% from the third quarter of 2002 through last year's third quarter, vs. 3.6% for GDP, according to figures released Feb. 27.

The statisticians insist that GDP is usually more accurate than GDI. J. Steven Landefeld, director of the Commerce Dept.'s Bureau of Economic Analysis, says he sees no evidence of a big overcount of GDP or undercount of service imports. BEA economists contend that the falloff in Americans' overseas travel -- which also gets counted as a service import -- accounts for the paltry-to-zero growth in the category since 2001. "We have the best survey in the world" on services trade, Landefeld says.

But it could be better. Take published figures for things like overseas call centers and programming work. They're mostly based on a survey done in 2002, adjusted upward for 2003 based on trends of the past three to five years. This approach falls short when economic patterns are shifting rapidly. Also, to save money and limit the paperwork burden on companies, Commerce collects detailed information only from companies that buy at least $1 million worth of a particular service, such as programming. But that misses lots of contracts, especially now that small business is climbing on the outsourcing bandwagon.

Lags between evolving business trends and the data aren't uncommon. "Anytime you have rapid change in the economy, you're likely to miss a bit more," Landefeld says. Acknowledging that there's room for improvement, BEA this year began collecting more data on services trade in quarterly, rather than annual, surveys.

Later revisions may bring the jobs and GDP numbers closer together. But for now, slow-growing income figures may paint a more accurate picture of the economy than the zooming GDP numbers. The world starts to make more sense if you allow for the possibility that the economy isn't growing quite so fast after all.

By Peter Coy in New York

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