The Problem with Product Replacement

Analysts and policy makers rely on data to tell them how world economies are faring. But what if that data is flawed? Chazen Senior Scholars Emi Nakamura and Jon Steinsson uncover compelling evidence that U.S. import and export indexes are faulty. The answer: something called "product replacement bias."
November 12, 2010
Print this page

Analysts and policy makers rely on data — lots of it — to tell them how world economies are faring. But what if that data is flawed?

Emi Nakamura, a Chazen Senior Scholar and assistant professor at Columbia Business School, suggests in recent research with Jon Steinsson (also a Chazen Senior Scholar and assistant professor in the Columbia economics department) that standard measures of import and export prices to and from the United States may understate the effects of global factors on the U.S. economy. That’s cause for alarm, she says, because those measures are used to gauge the health of the U.S. economy and affect government policy on issues including interest rates, trade tariffs, and money supply.

Dr. Nakamura began to suspect something might be amiss with the data while thinking about the well-known observation in international economics that import and export price indexes move relatively little with fluctuations in currency exchange rates. When the U.S. dollar depreciates, one might expect a foreign firm exporting to the U.S. to raise its U.S. dollar prices.

“But if you look at the raw data, this happens to only a small extent,” says Dr. Nakamura. In the data she examined, prices of imported goods moved very little in response to big swings in exchange rates, implying that foreign manufacturers were adjusting their prices so that they remained stable in U.S. dollars — sometimes, seemingly, to the point of selling at a loss.

“This has been a major puzzle in international economics,” Dr. Nakamura says. “Our thinking was that we needed to take a second look at the data and think whether there might be measurement issues that are affecting the numbers. Things might not be quite as they appear.”

The Evidence Mounts

By analyzing the data, Dr. Nakamura found that a huge amount of product churn in the marketplace was being missed when import and export price indexes were calculated. Two such indexes are the Import/Export Price Indexes (MXP), issued by the U.S. Bureau of Labor Statistics, which track the price of a market basket of imported goods over time. These products include food, electronics such as computers, industrial supplies such as chemicals, and transportation services such as air freight. If the Import Price Index rises, that means that the aggregate price for imported goods has gone up.

Indexes such as the MXP are easy to calculate in situations where the group of products stays constant over time, but that’s not what happens in the real world, as anyone who’s been to an electronics store lately knows. New, lower-price models are being introduced constantly as old models are phased out. That product switch — from, say, an older, more expensive 48-inch flat-screen TV to a newer, cheaper 48-inch flat-screen — might incorporate a price adjustment that is not accounted for in the MXP, a phenomenon Dr. Nakamura calls “product replacement bias. ” If this happens at the same time currency rates fluctuate (which often happens, she says), it’s difficult to tell what's responsible for the price change: movement in currency exchange rates? Product replacement? Or both (or neither)?

That’s important, she says. If exchange rates have scant effect on prices, then fluctuations can play little role in global economic rebalancing — the traditional argument for flexible exchange rates. But if there’s a greater effect, that might support the traditional role for exchange rates, as well as suggest that the Fed should pay greater attention to exchange rates in fighting inflation. Import prices also play into calculations of U.S. productivity, which can affect tax policies, funding for unemployment programs, and other government initiatives.

Next Steps

So what should be done? Dr. Nakamura suggests two fixes. One, devote more funding to government agencies that produce these indexes. “It’s amazing how much we rely on national statistics but how little we spend on them,” she says. “In fact, they’ve undergone a lot of funding cuts in recent years. In the case of the import price index, one of the really scary things is that it doesn’t measure the services trade at all. There was a small program to try to do that, but it was cut in the later years of the last [U.S. presidential] administration.”

A second step is for researchers to tease out other variables that distort import and export prices. This will help the Federal Reserve Board and other government entities make more precise projections of U.S. economic growth. One example: Researchers from the W.E. Upjohn Institute for Employment Research are studying whether price changes due to a switch in sourcing from, say, a French supplier to a Chinese supplier is captured in these indexes. Says Dr. Nakamura, “That’s skewing the data in ways people don’t realize.”

Add new comment

Plain text

  • No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.


* Required fields

Thank you for your subscription.