Scott Sumner, a friend of this blog, has noted that there's something of a difference between Paul Krugman the excellent economist and essayist of the nineties and Paul Krugman the political commentator of the teens.

I suspect that Krugman wants to believe that government regulation can improve the appalling working conditions in Bengali factories. So would I. But wanting to believe something doesn’t make it true.

That's actually something of a forlorn hope. For Bangladesh has many regulations about factories and buildings: it's just that everyone ignores them or bribes their way around them. Introducing more regulation into such an environment does not improve conditions, it increases either the ignoring or the bribery. The more specific complaint is that Krugman today says this:

At this point, however, there really isn’t any competition between apparel production in poor countries and rich countries; the whole industry has moved to the third world. The relevant competition is instead among poor countries — Bangladesh versus China, in particular. And here the differences aren’t as dramatic: McKinsey (pdf) estimates Bangladeshi productivity in apparel at 77 percent of China’s level. Given this reality, can we demand that Bangladesh provide better conditions for its workers? If we do this for Bangladesh, and only for Bangladesh, it could backfire: the business could move to China or Cambodia. But if we demand higher standards for all countries — modestly higher standards, so that we’re not talking about driving the business back to advanced countries — we can achieve an improvement in workers’ lives (and fewer horrible workers’ deaths), without undermining the export industries these countries so desperately need.

As above, everywhere already has regulations: it's whether anyone takes any notice of them or not that is the issue. But beyond that, there's that very odd reference to productivity. For old Krugman wrote this very perceptive essay:

- Wages are determined in a national labor market: The basic Ricardian model envisages a single factor, labor, which can move freely between industries. When one tries to talk about trade with laymen, however, one at least sometimes realizes that they do not think about things that way at all. They think about steelworkers, textile workers, and so on; there is no such thing as a national labor market. It does not occur to them that the wages earned in one industry are largely determined by the wages similar workers are earning in other industries. This has several consequences. First, unless it is carefully explained, the standard demonstration of the gains from trade in a Ricardian model — workers can earn more by moving into the industries in which you have a comparative advantage — simply fails to register with lay intellectuals. Their picture is of aircraft workers gaining and textile workers losing, and the idea that it is useful even for the sake of argument to imagine that workers can move from one industry to the other is foreign to them. Second, the link between productivity and wages is thoroughly misunderstood. Non-economists typically think that wages should reflect productivity at the level of the individual company. So if Xerox manages to increase its productivity 20 percent, it should raise the wages it pays by the same amount; if overall manufacturing productivity has risen 30 percent, the real wages of manufacturing workers should have risen 30 percent, even if service productivity has been stagnant; if this doesn't happen, it is a sign that something has gone wrong. In other words, my criticism of Michael Lind would baffle many non-economists. Associated with this problem is the misunderstanding of what international trade should do to wage rates. It is a fact that some Bangladeshi apparel factories manage to achieve labor productivity close to half those of comparable installations in the United States, although overall Bangladeshi manufacturing productivity is probably only about 5 percent of the US level. Non-economists find it extremely disturbing and puzzling that wages in those productive factories are only 10 percent of US standards.

New Krugman is implicitly equating working conditions with wages: for obviously, the cash is coming out of the same pot whichever it is spent upon. In a static analysis more spent on conditions will mean less for wages and vice versa. But he then goes on to say that we can adjust those working conditions without any blowback because productivity rates are close enough to be similar..

Old Krugman saw through that assertion: wages and working conditions are not set by the productivity in whichever sector it is. They're set by the productivity levels of the entire economy of the nation. That's why Bangladeshi wages are low, why working conditions are crap. Not because of anything about the apparel industry at all and certainly not because of any measurement of the productivity inside it. They're, respectively, low and bad because the rest of the Bangladeshi economy is a basket case.

It's worth recalling that Krugman was and is an excellent economist, that Nobel was truly deserved. Read and enjoy his early essays by all means. The stuff that turns up in the New York Times not so much.