There are a lot of good things you can say about international trade. But it does not, repeat not, do anything to alleviate a shortage of overall demand. Yes, if you liberalize trade countries will export more. But they will also import more. If you're worried about C+I+G+X-M, it's a wash, because X and M rise equally.
He then went on to criticize expansion of free trade, on the grounds that it does nothing to increase jobs. He might be right: job creation was never the argument for free trade. The economic argument for free trade, rather, was income and wealth creation. But let me stop my own exposition and quote from Bob Murphy. I loved his piece, not just for the content, which was full of nice distinctions and strong analysis, but also for his tone.
I have read Krugman's post a few times to make sure I'm not missing something, but I must confess I think he is committing a very basic error. Specifically, he is confusing the Keynesian accounting identity with a causal theory of how changes in one of the variables lead to changes in the other variables.
Later in the piece, Murphy writes:
To make the fallacy crystal clear, we can reverse Krugman's argument. Suppose governments around the world proposed to completely seal their borders and eliminate trade altogether. If Krugman is right, that should have no effect whatsoever on world output, and hence on the amount of workers necessary to produce all those goods and services.
It's true, every country's export sector would be devastated, but this fall in X would be exactly counterbalanced by a fall in M. Or more accurately, the countries with a trade deficit would see their output rise, but the countries with a trade surplus would see their output fall. There is no trade deficit for the world as a whole, so the two effects would cancel across all countries.
Murphy then closes with a nice numerical illustration.
I particularly like the tone Murphy took with a critic who called him a "damn fool." The threaded discussion between them is interesting and, by sticking to the issue and not responding in kind, Murphy got his critic to say he was sorry and to make a major concession. Well done, Bob.
The Keynesian model suggests that mercantilism can help an economy mired in a Depression. Smoot-Hawley was an almost perfect test. You may not care about the stock market, you may (wrongly) think the stock market only affects fat cats. But the deflationary impact of Smoot-Hawley on goods prices was very ominous. It was a very bad sign for the real economy, for working Americans. It meant lower production was on the way, along with fewer jobs. And the markets were right. And Hoover and Keynes were wrong.
If Kim Jong Il conquered China tomorrow, and closed it off from the world economy, the US trade deficit would shrink. But so would our economy.
I have just one correction to Scott's point above. I think he makes an error that Bob Murphy avoids. As Murphy makes clear, the Keynesian model per se does not suggest that mercantilism can help any economy in any situation, depressed or not, unless you take the word "suggest" to mean that one can be misled by an accounting identity into thinking that way. In other words, there is nothing in the Keynesian model per se that would lead to a belief in mercantilism. But it is true that even very smart Keynesians such as Krugman can be misled when they forget to distinguish between models and accounting identities.