The BEA responds

The BEA has published an FAQ (Frequently Asked Question) in response to my story. The fair summary of their response is that they do not dispute the problem, but they think it doesn’t make a big difference:

analysts at BEA—who are continuously updating the official estimates to reflect the impact of globalization—do not think that there is a significant bias on measured GDP or productivity growth.

As I wrote in the story, I think there is a substantial bias in the 2004 to 2006 period of rapid offshoring, though the exact magnitude is unknown because the data is not available.

I will go through the different parts of their argument, and explain where I differ. First, the BEA agrees with my story in principle:

import price indexes are designed to measure the change over time in the prices paid by U.S. residents for imported goods and services; they do not directly compare the prices of imported goods (and services) to their counterparts produced in the United States.

The implication is, as I stated in the paper, that the import price statistics don’t pick up the big cost cuts that are driving globalization.

Then the BEA gives three reasons why they don’t think this is significant.

That is, if real imports are understated because shifts to foreign suppliers are not being adequately captured in the price data, it is also likely that domestic production is understated because of shifts to more efficient domestic suppliers.

True—but the shift to foreign suppliers has greatly accelerated in this period because of the rise of China and globalization in general. By contrast, there’s no reason to expect that the unmeasured shift to more efficient domestic suppliers has taken a similar jump.

BEA uses chain-type measures that are designed to account for much of this substitution.

Agreed, and I figured that into my calculation.

Any remaining measurement errors should be offsetting because researchers have not demonstrated that errors in measuring imports are larger than offsetting errors in measuring domestic production.

This is the one that really makes me scratch my head. As far as I know, researchers haven’t even asked this question. How do they know the errors are offsetting, especially in this period of rapid globalization? I would love to see some more research—but the anecdotal evidence that I presented in the story suggests big errors in measuring import prices in particular industries.

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