Via Brad Delong, an interesting article (that is, however, likely to be somewhat obscure to a general audience) to consider when reading, say, Paul Krugman on  the Chinese renminbi / US dollar exchange rate.  (Note that in this article, Krugman doesn't actually say the Chinese currency is undervalued; he just says they'd be doing us a favor by appreciating it relative to the dollar (which they would do by lending us less money, notably by buying fewer US treasury bills and such)).  Helmut Reisen thinks the exchange rate is not as undervalued as some claim based on purchasing power parity; low-income countries tend to have a lower exchange rate due to the fact that nontraded goods in such countries are cheap, relative to traded ones whose prices are equalised internationally through trade, and even more, relative to similar nontraded goods in high-income countries.  Based on a linear regression, he estimates the RMB is only 12% undervalued.  The scatter plot of deviation of exchange rate from purchasing power parity versus log GDP per capita doesn't look linear to me (it looks convex downward), and if the apparent nonlinearity reflects some some relevant functional form in the dependence of one on the other, it might suggest the Chinese currency is even less undervalued.  Of course, comparison with such a cross-country regression shouldn't necessarily be decisive if more detailed China-specific analysis is available.  But I wonder what such analysis suggests.