Another day, another Paul Krugman post blaming China for US unemployment. And maybe he's right. But it would be nice to see some numbers.
On the same lines as my earlier post about the effect of dollar devaluation on aggregate demand, we can make a rough estimate of trade elasticities to calculate the effect of a Chinese revaluation.
Unfortunately, there aren't many recent estimates of bilateral trade elasticities between the US and China. But common sense can get us quite a ways here. In recent years, US imports from China have run around 2 percent of GDP, and US exports to China a bit under 0.5 percent. So even if we assume that (1) a change in the nominal exchange rate is reflected one for one in the real exchange rate, i.e. that it doesn't affect Chinese prices or wages at all; (2) a change in the real exchange rate is passed one for one into prices of Chinese imports in the US; (3) Chinese goods compete only with American-made goods, and not with those of other exporters; and (4) the price elasticity of US imports from China is a very high 4.0; then a 20 percent appreciation of the Chinese currency still boosts US demand by less than 1 percent of GDP.
And of course, those are all wildly optimistic assumptions. My own simple error-correction model, using 1993-2010 data on US imports from China and the relative CPI-deflated bilateral exchange rate, gives an import elasticity of just 0.17.  If the real figure is in that range, then a Chinese appreciation of 20 percent will reduce our imports from China by just 0.03 percent of GDP -- and of course much of even that tiny demand shift will be to goods from other low-wage exporters.
I don't claim my estimate is correct. But is it too much to ask that Krugman tell us what estimates he is using, that have convinced him that the best way to help US workers is to foment a trade war with China?
 This is a real exchange-rate elasticity, not a price elasticity, so it accounts for incomplete passthrough and offsetting movements in Chinese real wages. It assumes, however, that changes in the nominal exchange rate don't affect inflation in either country; to that extent, it's more likely an overestimate than an underestimate