Renminbi Undervaluation: Any Way You Look at It

April 9, 2010 4:15 PM

Intellectually, there are two ways of being in denial about renminbi undervaluation. The first is to succumb to what John Williamson has called the doctrine of immaculate transfer. Paul Krugman, in his recent blog posts, has been taking Stephen Roach and Joe Stiglitz to task for doing so: i.e., for their claiming that China's current account surpluses have little to do with exchange rates and that surpluses are all about savings and investment in each of the trading countries. Specifically, raising savings in China will somehow—magically and automatically—translate into current account adjustments irrespective of changes in exchange rates (hence Williamson's immaculate transfer metaphor).

But there is a second way of being in denial. And that involves muddying the waters by asserting the supposed fact that estimates for undervaluation vary widely. The New York Times recently suggested that while macroeconomics-based estimates of renminbi undervaluation were high, alternative estimates based on the so-called purchasing power parity (PPP) approach to measuring currency misalignment pointed to a relatively small magnitude of undervaluation of “only” 12 percent. This comes statistically close enough to doubting that the renminbi is actually undervalued.

In a recent policy brief, I undertake new estimates for the renminbi based on the PPP approach. These new estimates draw upon new insights (from a National Bureau of Economic Research [NBER] working paper that I coauthored) about how PPP is measured in the Penn World Tables and their implications for how undervaluation should be estimated. They also correct for the biases in the data that led to the 12 percent estimate (of Reisen, 2009) cited in the New York Times. These corrections, based on the work of Deaton and Heston (2009), are necessary because biases in the data afflict large developing countries and China in particular. Finally, one of the estimates also uses new data from the latest version of the Penn World Tables (version 6.3).

This exercise leads to four estimates (each with its limitations) and when I average them, I find that the renminbi is undervalued today by about 30 percent. This is actually broadly similar to estimates based on macroeconomic models. The bottom line is that there is less variability in estimates of renminbi undervaluation than is being suggested. And these estimates suggest substantial undervaluation, which call for remedial—and preferably multilateral—action.

More From

Arvind Subramanian Senior Research Staff