China and Comparative Advantage

In response to my rant about China the other day, commenter James (along with a few others) suggested that the problem was that I didn't understand enough economics, and helpfully assigned some remedial reading on the benefits of free trade, and in particular on Ricardo's theory of comparative advantage.

For example, he suggested the essay by Paul Krugman from 1996 entitled Ricardo's Difficult Idea, which I duly read, along with various other suggested references.  Other references will have to wait, but I'd like to take the opportunity of this morning to outline why the conventional theory of international trade, as exemplified in Krugman's essay (but which I originally learned in high school), doesn't reassure me much at all.

I'll begin with some caveats: I've read Paul Krugman for many years, he's obviously a genius, as well as a brave and decent human being.  This essay is from 1996, and doesn't represent his recent thinking on China (which in my opinion still has blind spots but is more nuanced).  So I'm more using the essay above because it's a very accessibly written meditation on the case for the conventional economic wisdom of the benefits of international trade.  Further, I'm not attacking international trade in general, or globalization in general, or suggesting protectionist policies as a specific response.  At this point, I'm simply trying to make a specific case about some dangers to the United States of what is happening in China (probably to other western countries too, but I haven't had the chance to think about that much in depth yet).  So in this post, I want to highlight what the conventional assumptions of the Ricardian model are, and point out that, empirically, they are materially violated here.  End Caveats.

Let's begin with this numerical example to illustrate Ricardo's original point, which concerned the advantages of trade between Britain and Portugal in the early 1800s.  We suppose that Britain can produce a unit of cloth with 100 units of labor, and a unit of grain with 110 units of labor, while Portugal, with a warmer climate, can produce a unit of cloth with 90 units of labor, and a unit of grain with 80 units of labor.

Now the point is that, while Portugal is absolutely more efficient than Britain at producing both grain and cloth, it has a comparative advantage at producing grain (because 100/90 = 1.111 < 1.375 = 110/80).  So while it might intuitively seem like Portugal should produce it's own grain and cloth (because it's better at both than Britain) in fact, both countries benefit by having Portugal specialize in producing the good where it has the comparative advantage: grain, and trading with Britain for cloth.

Let's see how this works.  Suppose initially both countries are producing some mixture of the two products, and, for simplicity, let's consider the relative benefits of moving 90 units of labor in Portugal from cloth to grain, and then making a compensating move of 100 units of labor in Britain from grain to cloth.  The 90  units of labor transfer in Portugal imply 1.0 less units of cloth in Portugal, and the 100 units of labor transfer in Britain imply 1.0 more units of cloth produced in Britain.  So the amount of cloth produced is the same as before in total.  However, in Portugal, the 90 units of labor produce 90/80 = 1.125 units of grain.  Meanwhile, in Britain, the loss of the 100 units of labor in grain production gives us only 100/110 = 0.909 less grain production.  So, in total, we are ahead by 1.125 - 0.909 = 0.216 units of grain.  Clearly, there is an overall benefit to trade!

And, if one moving one unit of Portuguese cloth production produces gains, the next unit has the same effect, and clearly we should keep going until as much as possible of the grain production is being done in Portugal, and as much as possible of the cloth production is being done in Britain (up to the constraint that demand in both countries be satisfied overall).

Now, I've artificially made the example above take all the advantage in grain - clearly the advantage would really likely be split between grain and cloth.  Further, transport costs impose some friction on this and trade only makes sense if that friction is smaller than the gains from trade.  The real world contains many more than two products, etc, etc.  Still, this simple example illustrates  the heart of why trading generally increases economic output, and I don't disagree with that idea.

But now let's explore some of the assumptions that are implicit in the above example.

The first is that this is a theory of the benefits of trade, in the aggregate, and after the switch has occurred.  Clearly, the real world process by which the example above occurs is likely to involve some pain.  In particular, importers are going to begin bringing cheap Portuguese grain into Britain, and marginal British grain farmers are going to start going out of business.  Meanwhile, the reduction in production of cloth in Portugal is going to raise the price of cloth there and start to make it feasible to import cloth from Britain.  Thus, cloth producers in Britain will have an incentive to expand and take on additional labor.

In the short term, this process is going to involve quite a bit of pain for quite a lot of folks - British grain farmers that lose their shirt, farm laborers that likely have to endure a period of unemployment and then learn new skills and perhaps move to a different region of the country.  Meanwhile, in Portugal, cloth distributors will go out of business as their source of supply dries up, and other cloth merchants in different parts of the country (ie near ports) will benefit instead.  When all is said and done, and the dust has settled, everyone is better off, but in the short term, there are serious losers as well as serious winners.

So comparative advantage is really a theory of the benefits of trade in equilibrium - it's saying that the equilibrium state of trade between Portugal and Britain is better than a state with no trade.  As Professor Krugman puts it:

Finally, and most importantly, it is not obvious to non-economists that wages are endogenous. Someone like Goldsmith looks at Vietnam and asks, "what would happen if people who work for such low wages manage to achieve Western productivity?" The economist's answer is, "if they achieve Western productivity, they will be paid Western wages" -- as has in fact happened in Japan. But to the non-economist this conclusion is neither natural nor plausible. (And he is likely to offer those Bangladeshi factories as a counterexample, missing the distinction between factory-level and national-level productivity).

- Constant employment is a reasonable approximation: The standard textbook version of the Ricardian model assumes full employment in both countries. But in reality unemployment is constantly a concern of economic policy -- so why is this the usual assumption? There are two answers. One -- the answer that Ricardo would have given -- is that international trade is a long-run issue, and that in the long run the economy has a natural self-correcting tendency to return to full employment. The other, more modern answer is that countries have central banks, which try to stabilize employment around the NAIRU; so that it makes sense to think of the Federal Reserve and its counterparts acting in the background to hold employment constant.

Emphasis added - international trade is a long run issue, and in the long run, the economy is likely to return to full employment. Similarly, Chinese wages are likely to, over time, equilibriate with those in the West.

I don't disagree.  In particular, there is no question that this has become true of Japan: the data in question are here:

But let's think for a minute here. Exactly how long is the "long run" in the case of China, and how long is the "short run". Starting from Chinese manufacturing wages being three cents on the dollar, how long till they reach Western standards? Well, the same BLS paper gives estimates of average manufacturing wages in 2002 being $0.57/hour, and rising to $0.81/hour in 2006.  That's a combined annual growth rate of 9.1%.  That doesn't correct for inflation, so the real appreciation is a couple of percentage points lower.  On other other hand, that is across the entire economy - in urban areas, wage inflation was a higher 11.5% (prior to inflation).  Let's take 10% as a nice round number rough approximation for real increases in Chinese wages from 2002-2006.  So, assuming that continues, we can use the old rule of dividing into 70 to estimate that Chinese wages will double about every seven years.  It will take about five doublings for them to reach US levels, so that's 35 years.  Longer if, as seems likely, the rate of increase slows down over time.

So, we can say that it will take several decades for Chinese wages to equilibriate with western wages.  So, it seems to me, the "long run" when the benefits of trade will fully accrue to western workers, is several decades off (and who knows what else will have occurred by then).

So, really, for all practical purposes, we are talking about the short run painful period of adjustment lasting for the entire working lives of anyone in the workforce today, rather than being a relatively brief period.

And not only that, the scale of Chinese manufacturing capacity is already enormous -- steel production as large as the rest of the world put together, recall, and a car industry that has now reached about the size of the US car industry -- despite the fact that their labor costs are still miniscule.  So we are talking about making an adjustment that is very large indeed.

Now, let's take a look at this idea that "constant employment is a reasonable approximation".  Ok, what exactly is constant about this?

Clearly, in the era of the last several decades, when competition with Asia has been a major factor for US manufacturing employees, employment for men has not been constant, indeed it has trended down steadily.  Now, I'm not saying competition with Asia is the only factor - technological trends have been important too, and I'm quite prepared to cede some role for the entry of women into the workforce (most particularly in making it more feasible for some men to not work if their female partner is willing to support them).  I would argue that the technology and Asian competition factors are synergistic - unskilled work differentially moves overseas, and the work that remains here is increasingly automated and requires more intelligence and aptitude with technology.

But what I think we can say with some confidence is that this not a happy trend.  We are not seeing early retirements here, or affluent dot com millionaires leaving the work force for the Caribbean.  Instead, the rate goes down sharply with each recession as men are laid off, and then it doesn't fully recover in between as not all of them are able to return to the workforce.  In particular, if we look at some of the internals, we can see that it's the most disadvantaged groups that are particularly vulnerable to this trend.  Whereas overall employment/population for men is down just over 80%, it's 69.4% in Q4 of 2009 for black men, but 82.7% for white men.

Similarly, for men over 25 with less than a high school diploma the employment population ratio is only 49.5%, for high school graduates with no college, it's 63.7%, whereas for college educated men, it's 77.8%.  Those are enormous differences.  (Because of the vagaries of BLS data reporting, I have to give the figures for all men over 25 to break it out by education level, so this includes retirement age men, but we would expect that effect to lower the relative level of employment in educated men, since the educated generally live longer - so amongst working age men, the disparity is likely to be even greater than reported here.)

And finally, let me repeat this graph on manufacturing employment:

While we await the era of happy Ricardian trade benefits in 40 years time, there is plenty of space for that graph to get pretty damn close to the x-axis on the trend of the last ten years (the period when Chinese exports to the US have really ramped up).  And recovery won't be quick either - does anyone think Michigan has fully recovered from competition with Japan, even now that Japanese wages have more or less equilibriated?

So I continue to maintain: the kinds of non-college-educated guys who would traditionally have worked in factories are in serious trouble and on present trends it looks like it's going to get much worse before it gets better.  That is not a recipe for social stability.  And it is not just a transitory effect of the current financial crisis.