Everyone is familiar with this chart–total imports as a share of GDP. (In fact, I can’t count the number of times I ran this chart, or something similar, in the pages of BusinessWeek). The chart appears to show that the import share of the economy in 2009 was basically the same as it was in 1999 (13.7% vs 13.4%).
The only problem–this chart is nonsense, and tells us almost nothing about the size of imports relative to the U.S. economy. Here are the problems*:
1) Purchaser versus producer values: GDP values a shirt in the store; the import figures value that same shirt when it enters the country, not including the cost of getting it to the store and selling it. That difference between the store ( ‘purchaser’ ) value and import (‘producer’) value can be enormous, depending on the particular item. In the case of men’s clothes, for example, the BEA’s data shows that the store value of men and boys apparel is almost triple that of the producer value, with the bulk of that difference coming in the margin of retailers. More typically, the cost of goods to the final purchaser is roughly about double the producer value. Adjusting for domestic transportation, wholesale and retail margins gives us a higher share of imports in the economy.
2) Final vs intermediate : GDP values a car when it is bought by you, the ‘final purchaser’, in the showroom. It does not count separately all the ‘intermediate’ goods and services that go into that car. So the value of the car radio, say, does not show up separately in GDP, even if it is made by a different company. By contrast, imports include both final goods and services (sold to the ultimate purchaser) and intermediate goods and services (used by businesses, nonprofits, and government to produce final goods and services). So ideally we would divide imports by a measure of U.S. economic activity that includes both final and intermediate goods and services. That’s not an easy concept, but the closest that we have is the “gross output” measure constructed by the BEA as part of its industry accounts, which counts purchased intermediates as well as final goods and services. In 2008, gross output was $26.6 trillion, roughly double the $14.4 trillion in GDP. Adding in the production of intermediates gives us a lower share for imports, though it has an ambiguous impact on the time trend.
3) The import-domestic price differential: In many cases, the producer price of importing an item from China or another country is substantially less than the price of the same or similar item produced domestically. It has to be–that’s what drives the whole offshoring movement. No one would source from overseas unless it was cheaper.
However, when we divide the dollar value imports by the dollar value of GDP (as in the chart) and use that as a measure of import penetration, we are implicitly assuming that the price of imports is the same as the price of the comparable domestic goods. Let me give you an example. In 2007, Americans bought about $100 billion in furniture (valued at producer prices). We imported about $30 billion in furniture. How much has imported furniture penetrated the American market? If we divide $30 billion into $100 billion (30%), we are assuming that $1000 buys the same physical amount of furniture, whether it is spent on imported or domestic furniture. But what if imported furniture is roughly one-third cheaper (producer prices) than the comparable piece of domestic furniture? Then $1000 actually buys much more imported furniture (perhaps a table plus chairs) than the same $1000 spent on domestic furniture. Then the ‘real’ import penetration–measured in physical pieces of furniture–might be closer to 40% or 45%. Adjusting for the import-domestic price differential would tend to raise the measured share of imports.
Unfortunately, there is no statistical agency that collects data on the import-domestic price differential. None. The BLS does an underfunded job of collecting data on changes in import prices, but there is as yet absolutely no official data on the import-domestic price differential (see here for an outline of how the BLS might do it).
However, anecdotal evidence suggests that the import-domestic price differential for goods from China is on the order of one-third. That is, the price of goods sourced from China is roughly two-thirds of the same goods sourced domestically. (see, for example, BusinessWeek’s 2004 story “The China Price”). New research by Susan Houseman of the Upjohn Institute and Christopher Kurz, Paul Lengermann, and Benjamin Mandel of the Fed finds “systematic import price differentials between products from advanced versus developing and intermediate countries,” which suggests ranges for the import-domestic price differential for the U.S.
Using one-third as a starting point for the China-U.S. import-domestic price differential gives us a chance to calculate China’s ‘real’ share of the U.S. market for manufactured goods.
This chart shows China’s goods imports, as a share of U.S. domestic market for non-oil manufactured goods, under the assumption that goods imported from China are one-third cheaper than the comparable goods manufactured in the U.S. The denominator is the gross output of non-oil manufacturing at producer prices (from the BEA’s industry accounts) minus exports plus the adjusted value of imports. Imports are adjusted under the assumption that non-oil imports from developing countries are one-third cheaper than comparable U.S. products, and non-oil imports from intermediate countries are one-quarter cheaper.
Using this procedure, I estimate that the ‘real’ import penetration of Chinese-made goods was roughly 9.7% of non-oil manufacturing in 2008, up from 2.9% in 1999.
China also has a much bigger real import penetration than other major import partners of the U.S. Japan, Germany, and Canada are advanced countries, and we would expect a relatively small import price differential. Mexico is somewhere in the middle–a smaller price differential than China, but bigger than industrialized countries. So assuming a 33% import-domestic price differential for China and a 25% import-domestic price differential for Mexico gives us the following chart.
China’s penetration into the U.S. manufacturing market is more than twice as big as Mexico and Canada (omitting oil). That explains why China has become so much the focus of political debate, rather than Canada and Mexico, or Japan and Germany.
IMPORTANT CAVEAT: These are preliminary calculations. Have fun, shoot holes in them, tell me that I’m an idiot. But in exchange, I reserve the right to change them and put out new calculations without an ounce of apology.
*Yes, I know the correct denominator is gross domestic purchases rather than gross domestic product. When I do the calculations for China and the other countries at the end of the post, I adjust appropriately for exports and imports