There's a lot of heat, but surprisingly little light. Is the objection to the renminbi peg as such, or is it just he thinks a different peg would be better? Better for who -- the US, China, the world? Why is he so sure that the current account imbalance between the US and China is due to the peg? And what general principles for currency value and trade flows, if any, underwrite the argument in this specific case?
One thing he's got right, at least -- the US current account deficit with China is big. The US c.a. deficit in January was about $40 billion, or 3 percent of GDP. (That's down from 5 percent of GDP in 2008, and 7 percent in 2005-06.) China accounted for $18 billion of that ($17 billion if you include Hong Kong, with which the US runs a surplus), or about 40 percent of the total. At the height of the c.a. deficit, in mid-2006, China accounted for about a third of it. The rise in China's share is largely accounted for by the fall in oil prices; in recent years China has consistently accounted for about half the non-OPEC deficit. So if you are worried about the current account deficit, you should worry about the deficit with China.
But should you worry about the current account deficit? And does worrying about the deficit with China mean worrying about Chinese "manipulation" of the renminbi?
To take the second question first, it's far from clear that the currency peg is responsible for the deficit. One reason the US deficit with China is so big is that US trade with China is so big -- China is by far the largest recipient of US exports outside of North America. Yes, China's exports to the US are four times greater than its imports from the US, an outlier among major trade partners, but countries like Germany, Japan and India regularly see ratios in excess of two to one, and their currencies float freely. So at the least, it's clearly not true that laissez-faire in the foreign exchange markets guarantees balanced trade -- which would be a strange thing for Krugman to believe in any case.
But of course, it could be true that the vagaries of the foreign-exchange markets, demand for US assets, or government policy can all keep a currency away from its trade-balancing level. So set aside the loaded language of manipulation, which suggests there is something inherently immoral or dishonest about currency pegs -- which have, after all, been the norm in international trade for much longer than floating currencies, and are used by lots of other countries beside China today. Is it the case that a higher renminbi would narrow the US deficit with China?
There is a long tradition, going back at least to Keynes, that doubts whether trade flows are sufficiently responsive to exchange rates to make them an effective way of achieving balanced trade. In the short run, the elasticity pessimists are clearly right that income changes, not exchange rates, are the decisive influence on trade -- casual examination of the trade statistics shows that exports to and imports from any given country tend strongly to move together, whereas they should move in opposite directions if exchange rates were dominant. Nor is this surprising -- in the short run most trade is contractually committed; over the medium run market share is expensive enough that sellers absorb some part of exchange rate movements in profit margins rather than adjusting prices, and even when prices do adjust few close substitutes are available for many traded goods, especially intermediate goods. Thus the familiar J-curve, where the short- or medium-run effect of an exchange rate change is in the "wrong" direction.
But in the long run, surely, prices are decisive? Maybe, maybe not. Krugman brings up "the smaller East Asian nations in the aftermath of the 1997-1998 crisis" as if they were an obvious case of exchange rate effects. It is true that those countries did devalue their currencies during the crisis, and did see sharp improvements in their current account in the following years. But they didn't just devalue; they also saw dramatic falls in domestic income and consumption. How do we know if it was the devaluations or the contractions that led to the improved trade balances? Well, take Indonesia and Korea as examples (I pick them because they are in the OECD international trade database.) In both, the improvement in the current account balance (from a $250 million to a $2.7 billion surplus 1996-2000, and from a $2.4 billion deficit to a $3 billion surplus 1996-98, respectively) came entirely from declining imports. Indonesian exports were no higher at the end of 1999 than at the beginning of 1997 -- but imports had fallen by half. That looks a lot more like an income effect than a price effect to me.
Still, in the very long run exchange rates presumably do have a major effect on trade flows. But it matters how long the long run is. If the problem with the current account deficit is its anti-stimulus effect, and not the broader "global imbalances," then a solution that only helps five years from now, and makes things worse in the next year or two, is no solution at all. (Someone should ask Prof. Krugman how long he thinks it would take for a renminbi revaluation to have a net positive effect on the US current account.) But set that aside. Let's say that a change in Chinese policy would lead to a higher renminbi, and that that would narrow the US deficit with China, and fast enough to work as fiscal policy. Should we care? And does the answer depend who "we" are?
The simplest way of looking at the demand effects of the deficit is that every dollar of "anti-stimulus" in the US is balanced by a dollar of stimulus in China. And to be honest, it's hard to find much beyond that simple level in Krugman's stuff on China. He seems to be saying the reniminbi should be revalued so that the US gets more, and China less, of a fixed pool of global demand. Now, I am not as allergic to mercantilist-type arguments as most people. But why on the world should China go along with this? And why, from behind the veil of ignorance, should we-in-general want them to? 
In a recent blog post, Krugman had an interesting answer to this question -- interesting because it's so clearly wrong. He writes, "the global macro aspects of the situation are reminiscent of the late 1920s, when the US was simultaneously insisting that European nations repay their dollar debts and that they not be allowed to export more to earn the dollars. That didn’t end well." There are two key differences, ,though. First, China accepts payment in dollars -- it doesn't insist on its own currency. And second, China has no problem exporting capital to match its current account surplus, unlike the US in the 20s, when our surplus was offset by politically contentious loans related to WWI reparations and highly unstable private flows. These two factors mean that while debtors then found themselves forced to accept domestic deflation and contraction, the US is under no corresponding pressure. As long as China is willing to finance the US deficit with low-interest loans to the US, the deficit need have no negative effects for this country; and as soon as they stop financing it, it will go away -- as Krugman rightly stresses, China's trade surplus and capital exports are two sides of the same coin.
Let's take a step back.
Imagine that China appeared out of the blue one day and began selling stuff to the US. But instead of buying stuff in return, they simply lend their dollar earnings to the US government, i.e. bought Treasury bonds. This reduces demand for producers of US goods. But now suppose the US government increases its total borrowing by the amount of the Chinese bond purchases, and uses the increment for domestic purchases. Supply and demand for bonds among non-China buyers are unchanged, so there is no reason for interest rates to move. The stimulus of the additional government spending exactly offsets the anti-stimulus of the Chinese imports, so there is no inflation. And the offsetting goods and capital flows between the US and China mean there is no pressure on the dollar. At the end of the day, domestic output and employment are unchanged, and domestic consumption is increased by the amount of Chinese imports. In short, to exactly the extent that the imbalance with China reduces private domestic demand, it removes the constraints on expansion of public domestic demand. This is the fundamental difference between the position of the US today and the position of countries on the gold standard or equivalent systems, as in the 1920s. In the latter case there is no mechanism to guarantee offsetting capital flows to trade imbalances, so countries can find themselves facing a foreign exchange constraint on output and growth.
And now we are coming toward the point of this very long post. Krugman calls Chinese capital exports "artificial," by which he means -- well, what, exactly? That they are undertaken by government? that they are not motivated by returns? that they correspond to a big current account surplus? It isn't clear. Nor is it clear what he thinks a world of "natural" flows would look like. More balanced trade overall? or just a more favorable balance for the United States? And finally, why should we support a policy whose benefits to American workers are offset by costs to (much poorer) workers elsewhere?
Let's turn back to that previous era of global imbalances, the 1920s and 30s -- whose lessons, I think, Krugman gets wrong. It's well known that when countries left the gold standard and devalued their currencies, their economic performance dramatically improved. But was this from the stimulus of an improved current account,as Krugman imagines the US would enjoy following a renminbi revaluation? Not at all. As Peter Temin and Barry Eichengreen both emphasize, countries that devalued did not, on balance, experience any improvement in their current account at all! Rather, the devaluations allowed them to achieve the same external balance at a higher level of income. Removing the foreign exchange constraint allowed national governments to take much more aggressive steps to boost domestic demand -- mainly through looser monetary policy once central banks no longer had to defend the peg to gold. In effect, going off gold enabled a movement from a low-employment to a high-employment equilibrium by allowing countries to reflate one at a time, instead of needing a coordinated expansion; but trade flows at the beginning and end of the process were basically the same.
The key point is that what matters is not the balance of trade between any particular countries, but the presence or absence of a foreign exchange constraint. If a country can offset adverse movements in its current account with expansions of public, or private, domestic demand, i.e. if a worsening current account is reliably offset by capital inflows or if it can settle international claims in its own currency (of course both are true of the United States), then deficits need have no effect on output or employment. But some countries must restrict domestic demand to keep any current account deficit at a level they can finance; adverse movements in those countries' current account reduce output and employment without any corresponding gains elsewhere. Thus the deflationary bias of the gold standard. This was the decisive consideration for Keynes in the design of postwar international financial arrangements: No country should be prevented from pursuing full employment by a foreign exchange constraint. 
So where does that leave us? A new international financial architecture isn't on the menu (and doesn't seem to interest Krugman, anyway; for him the problem is all China.) But under the current system, it's clear that the highest level of output and employment will be achieved if a simple condition is met: current account deficits are run by countries that can most easily finance them. Countries that can easily attract capital inflows and issue liabilities in their own currency should have exchange rates that result in deficits at full employment; countries without those characteristics should have "undervalued" exchange rates so they run surpluses at full employment. This preserves the flexibility of every country to manage domestic demand to preserve full employment. Expansionary fiscal or monetary policy leads to adverse movement of the current account. It's no problem if this increases the deficit for countries that can finance a deficit, but for others this will rule out expansion unless they start from a position of surplus. And indeed this was one of the lessons of the Asian crisis -- countries that found they could no longer count on private capital flows concluded they need to run large current account surpluses to ensure that their growth was not choked off by foreign exchange constraints.
Given that the US issues the global reserve currency, a world with a large US current account deficit will almost certainly see higher and more stable output than one in which the US current account is balanced. And given that the US does not face a foreign exchange constraint, our unemployment can't be blamed on the Chinese – they're not holding back domestic demand here.
Krugman is a smart guy: why doesn't he recognize this?
The answer, I think, goes back once again to Keynes. In chapter 23 of the General Theory, he writes:
When a country is growing in wealth somewhat rapidly, the further progress of this happy state of affairs is liable to be interrupted, in conditions of laissez-faire, by the insufficiency of the inducements to new investment. Given the social and political environment and the national characteristics which determine the propensity to consume, the well-being of a progressive state essentially depends, for the reasons we have already explained, on the sufficiency of such inducements. They may be found either in home investment or in foreign investment... Thus, in a society where there is no question of direct investment under the aegis of public authority, the economic objects, with which it is reasonable for the government to be preoccupied, are the domestic rate of interest and the balance of foreign trade.
Before the 20th century, there was "no question of direct investment under the aegis of public authority" simply because the public sector -- apart from the military -- was very small. Of course that's not the case today – no technical reason why federal spending couldn't increase by 10 percent of GDP if need be. The obstacles are political – as Krugman acknowledges when he stipulates that the Chinese surplus matters because in its trade partners, "both central banks and governments are unable or unwilling to pursue sufficiently expansionary policies to eliminate mass unemployment" (my emphasis).
The unwillingness of the US to pursue sufficiently expansionary policies is not a fact of nature. Given that unwillingness, Krugman (and Dean Baker, etc.) may be right that an improvement in the US current account is the most practical way to boost US demand – even if they exaggerate how quickly and reliably exchange rate changes will deliver it. But it's neither a necessary nor a particularly good way of achieving this. The real problem is the inability of the US financial system to channel savings into productive investment, and the unwillingness of the state to step in in its stead. Too bad folks like Krugman are trying to shift the blame to China.
 Formally, Keynes and post-Keynesians like Davidson doubt that the Marshall-Lerner-Robinson condition is satisfied, i.e. that the elasticity of imports and exports with respect to exchange-rate changes sum to at least one.
 He actually does answer this question with respect to China, sort of. Might makes right: "Because the United States can get what it wants whatever China does, the odds are that China would soon give in."
 In the real world the problem is that US government borrowing did not increase by as much as the capital inflows from China, with the result that Chinese purchases of Treasurys drove down the yield and forced more return-sensitive investors to look for similar assets elsewhere -- thus the demand for asset-backed securities. This point is made by Perry Mehrling and by Daniel Gros. Neither draws the logical conclusion that the financial crisis might have been averted if the federal government had only borrowed more. Topic for another post.
 Countries other than the US are in more the gold-standard situation -- they face the problem of earning the dollars to cover their deficit with China, or euros for their deficit with Germany.
 Since he doubted the effectiveness of exchange rates to provide the needed flexibility, the key goal for him was an automatic mechanism to provide the offsetting capital flows to deficit countries, without depending on either private investors or the governments of surplus countries.