happy new year, everyone!
Krugman notes that Chinese mercantilism (exchange rate manipulation to increase exports) actually does steal jobs from the US and help China when we're not at full employment.
For once, I agree with Krugman (and this is saying something, because I thought his last 4 or 5 major columns were idiotic at best - that said, his specialty is international trade, not the other stuff he writes about, so perhaps its unsurprising that he is insightful on international trade).
I've always interpreted the superiority of free trade over protectionism to be the necessary case if and only if exchange rates can float. When they are artificially fixed, there can be a huge number of problems on both sides, depending on where the peg is and the nature of the individual countries' economies.
What Krugman neglects to mention is the nature of acceptable protectionism. I've noted before a few circumstances under which protectionism makes sense:
Protectionism at full employment has huge, well-documented problems, as Krugman acknowledges. However, the biggest problem with protectionism when NOT at full employment (not mentioned in the article, but he certainly understands it) is the redistribution protectionism creates from workers in industries that are protected to workers everywhere else. Targeted steel tariffs make life way more expensive for everyone else, but create jobs for steelworkers. These protections generally remain in place whether the economy is at full employment or not, and overall have been major net negatives for the country (one study I read months ago claimed that the benefit of killing all tariffs in the US would be a shocking $12,000 per household on average).
Thus, a rational response to mercantilism elsewhere is protectionism that targets the exact same industries with the exact same weight. In this case, an exchange rate policy blankets all Chinese goods on a currency-weighted basis, so any protectionist response cannot come in pieces (tires here, steel there, etc). The appropriate response is an "import certificate" policy, as espoused by Warren Buffett, but towards China only, using yuan-denominated trade imbalances instead of the whole deficit. (see here: http://tfideas.blogspot.com/2009/01/trade-deficit.html). That would allow for an exact currency weighted reversal of Chinese mercantilism - the more mercantilist they are, the bigger the effect. This way, protectionism doesnt continue at a level higher or lower than appropriate when Chinese mercantilism changes (as with a flat, Congress-dictated tariff), and it doesnt hit specific industries harder than others (as with targeted tariffs), except insofar as Chinese industries are competitive with American productivity (this is the point - Chinese industries that would have lots of US exports even if the Yuan were stronger are the ones you want to see continue their exports). It would quickly end Chinese mercantilism also, because the Chinese have no incentive to manipulate the currency because imbalance is just going straight towards American government spending (this neglects a potential black market for Chinese goods, but let's assume, probably fairly, that the cost of smuggling into the US is substantial enough that the cost of continued mercantilism isn't worth it for the few goods that would be profitably smuggled in, especially since those are the ones that would still thrive in an import-certificate world).