How Those Deceptive Numbers Creep In

You might think the most important fact about the price of an imported chair from China, say, is how much less it costs than the comparable chair made in the U.S. After all, that cost difference has driven the offshoring boom.

But that's not a number the government tracks. Surprisingly, import price data are "not designed to measure goods as they shift from domestic to foreign production," explains William Alterman, who supervises the international price program for the Bureau of Labor Statistics. If a chairmaker in North Carolina now manufactures its wares in China, the import price data wouldn't capture its falling production costs.

The import price figures also don't record the price cuts, productivity gains, or quality improvements that enabled the Chinese factory to undercut the U.S. producer in the first place. The authority of the BLS stops at the U.S. border. So when a particular new product shows up on the docks, that's the first time the import price trackers know anything about it.

Why does this matter? If import prices don't reflect the big cost savings from offshoring, then measured import-price inflation is higher than it should be. As a result, the real, inflation-adjusted growth rate of imports is reported as lower than it really is. And that's bad because all of the statistics calculated using real imports--including real GDP growth, productivity, and the real value-added of manufacturing industries--are thrown off. In particular, any underestimate of real import growth will lead to an overestimate of domestic production growth. This is how phantom GDP is created.

Let's walk step by step through an example. Suppose that in 2000 a North Carolina factory made a dining room table for $1,500. At the time, a factory in China could make the same table, but not cheaply enough or quickly enough to take business away from the American factory.

Fast-forward to today. The owners of the Chinese factory have invested in new equipment, boosting productivity. They've learned how to apply better finishes, and they've bought new software for taking orders from abroad. Now the Chinese factory can sell the table for $1,000 in the U.S., including shipping. The American company closes its factory and starts buying from China. U.S. consumers are still buying the same tables as before, but domestic production has ceased.

How does this look to the government statisticians? They attempt to calculate the "real" value of consumption by measuring how much consumers spend on tables at different times and adjusting for the changes in price. Then they do the same for "real" imports. In this example, real domestic production is more or less the difference between the real value of consumption and the real value of imports.

Let's stop here to explain some economic jargon. To tally the real value of things like consumption, economists sometimes use what they call "inflation-adjusted" dollars, measured against a base year. So if a gallon of gas costs $1 in 2000 and $3 in 2007, economists would say the real value of the gallon of gas was still $1 in "2000 inflation-adjusted dollars." It still propels your car the same distance.

Now, back to the dining room table. Between 2000 and 2007, the statisticians see that consumer spending on tables has dropped by one-third and that the price has dropped by one-third (assuming all savings are passed along to consumers with no markup for profits). "Aha," they say to themselves, "The price drop accounts for all the change in spending, so the real value of consumer spending hasn't changed." They type in their spreadsheets that the table is worth $1,500 to consumers in 2007, measured in inflation-adjusted 2000 dollars. After all, it's the same table as the ones made in 2000.

Here's the rub. The statisticians need to apply their inflation-adjustment magic to the imported table as well. But the folks at the BLS import price office are stuck: They have never seen that table before and don't know about the domestic table it's replacing, or the previous table from the Chinese factory not cheap enough or high-quality enough to be imported to the U.S. So instead of figuring in a price drop of one-third for the imported table, they assume a much smaller price decline, or perhaps no price drop at all. When they tally it up, they give the imported table a value of about $1,000, measured in inflation-adjusted 2000 dollars.

Their method leads to a serious mistake. The real value of imports looks less than the real value of consumption, by $500. After applying an arcane statistical adjustment, the government would report domestic production of roughly $250, even though no furniture production has taken place in the U.S.

This happens over and over as production of new types of goods and services moves offshore. It's not just the shift from domestic to foreign production that creates problems. Anything that changes the nature of the import--a shift in the country of origin or a change to a more advanced model--can lead to an undercount of real imports.

Consider the fast-changing television market, where consumer prices have plummeted by 42% in the past three years and imports of liquid-crystal display and plasma sets have soared. The BLS says the import prices of TVs fell only 15% in the same period. That's odd, given improving technology and the intense price wars in the industry. Indeed, the "landed cost" or import price of a 32-inch LCD television has fallen 69% during the past three years, reports DisplaySearch, a flat-panel market research firm. This apparent mismeasurement of television import prices ends up creating phantom GDP, perhaps as much as $10 billion over the past three years.

Government statisticians will have trouble catching up with the problem. As the global economy changes, "capturing the shift from domestic to foreign production [or vice versa] and its associated impact on prices is at the forefront of methodological challenges we face," says Michael Horrigan, associate commissioner at the BLS. One impractical solution would be to have the BLS track prices directly at Chinese factories as well as U.S. factories. That's good in theory, but it won't happen anytime soon.

By Michael Mandel

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