Imports are increasingly high-tech but the way they are measured is decidedly not.
In a case of economic myth busting, new research from professors Emi Nakamura and Jón Steinsson (Department of Economics) reveals that import prices may be substantially more responsive to exchange rate changes than they appear to be in the government’s data. And that has the potential to cause all sorts of problems for the GDP, according to BusinessWeek’s chief economist Michael Mandel. In a recent cover story, he cited their research and discussed how measurement problems in import price data can affect growth statistics.
Import goods like computers, machine parts, services and other items with prices that change quickly and where units are fundamentally hard-to-count can cause data issues. For example, model upgrades for high-tech goods and services happen remarkably often, potentially changing every few months, but the BLS is only able to log price changes for models that are identical from one month to the next. That means that price changes associated with these granular fluctuations (upgrades, discounts, manufacturing changes) are getting lost in the shuffle.
To make matters worse, Mandel notes that many services that are outsourced offshore, which should be measured as an imported good, are not measured at all.
“The kind of analysis the BLS needs is labor intensive,” says Nakamura. “If it’s true that we want to measure the impact of high-tech industries on the economy or the impact of services, we probably need to have these national statistic entities grow with the economy to be able to take on these additional tasks. This requires more funding.”
Nakamura and Steinsson uncovered the import price discrepancies by looking at the relationship between import prices and the exchange rate and started connecting the dots.
“When the U.S. exchange rate depreciates, imported products become more expensive and American consumers theoretically switch from buying imported to domestic products,” explains Nakamura. “But if you look at the data it looked like the import prices were responding very little to the exchange rate. That was a major puzzle: why did they look so smooth? It occurred to us that one of the reasons it might look so smooth was because a lot of the price adjustments during model changeovers were getting lost.”
For example, in the period between 2002 and 2008 there was a 20% movement in the exchange rate, but Nakamura shows that the foreign firms didn’t raise their U.S. dollar prices by nearly that much — raising prices by less than half of the amount they would have needed to protect their margins.
Mandel argues that the measurement problems uncovered by Nakamura and Steinsson have important implications for productivity growth. If import price inflation is overstated then productivity growth would also be overstated. That would mean we would be wrongly attributing the fact that we observe American companies producing more despite spending the same amount on foreign inputs. In fact, these gains might be arising from unmeasured growth in imported intermediate inputs.
And the issues here — both in measurement and types of imported goods — are only growing in magnitude.
“To the extent that the economy moves in the direction of importing more high-tech goods and services, where products are fundamentally hard to count, the problem will get worse,” Nakamura says.
Photo credit: Evan Leeson