A Deficit of Reasoning on Jobs

By Derek Scissors  

The link between the balance of payments, GDP, and jobs is not as Paul Krugman and others assume.

In the game of telephone, a word or sentence becomes unrecognizable if enough people repeat it, especially when people are not trying hard to get it right.
In the New York Times last week, Jared Bernstein called for an end to the U.S. dollar serving as the world’s reserve currency. There are serious practical matters to consider, but Bernstein’s ultimate reason for supporting such a change is jobs. Bernstein claims a huge number of jobs lost due to the dollar’s role. He cites a recent academic article by Kenneth Austin, who “argues convincingly that the correct metric for estimating the cost in jobs [of the dollar’s role] is the dollar value of reserve sales to foreign buyers.”



The connection between jobs and selling dollar assets to foreigners is not obvious, and neither Bernstein nor Austin make the case for it. In his article, Austin argues:

Krugman has denounced the People’s Republic of China’s “mercantilist” policies (undervaluation of their currency, the renminbi) in his New York Times column and blog. Krugman (December 2009) attributed a proportional share of the rest of the world’s (non-PRC) job losses resulting from the PRC’s exchange rate policies to the United States: 1.4 million jobs.

The Systemic Model shows that when reserve accumulators, like the PRC, finance current account surpluses by reserve purchases, the first burden of adjustment is born by the reserve issuer. It is not spread evenly around the world as Krugman assumed. The Systemic Model demonstrates that even if the PRC’s bilateral surplus is with a third country, such as a member of the Brazil group, the PRC’s purchase of reserves from the United States both enables its own global BOP [Balance of Payments] surplus and causes the U.S. global BOP deficit. Thus, when third countries run bilateral current account deficits with the PRC, the reserve issuer (the United States) will tend to have larger current account deficits as a counterpart.

The Systemic Model supports Krugman’s basic argument but implies that he underestimated U.S. job losses. The Systemic Model shows U.S. job loss should be estimated by U.S. reserve sales.

The middle paragraph is Austin’s  contribution. He models the impact of a country, like China, that buys U.S. dollar-denominated assets in part to keep its currency from appreciating. Austin shows that bilateral transactions between China and the United States are not the only place such a policy can have an impact on the United States. If China alters its trade with Brazil by buying dollar-denominated assets, this can also affect the American balance of payments.
It’s the first and third paragraphs that are the problem. They both cite jobs, but jobs are not part of Austin’s model. The words “jobs” and “employment” only appear in the paper when Austin is referring to someone else. On its own, Austin’s model tells us nothing about jobs. After 22 pages of his own academic, mathematical work concerning the balance of payments, Austin merely cites a newspaper column and blog by Krugman concerning jobs. This looks a bit like the “then a miracle occurs” method.
It looks even more like that method in light of what Krugman actually wrote:

Start with the Chinese surplus. It has been temporarily depressed by the world trade collapse, but seems to be on the rise again. Blanchard and Milesi-Ferretti, at the IMF but speaking for themselves, project a Chinese current account surplus for 2010-2014 of 0.9 percent of gross world product.       

You can think of this as a negative shock to rest-of-world net exports. (Technically, that’s not quite correct — because the shock depresses res-of-world [sic] GDP and hence rest-of-world imports from China, the realized trade surplus is smaller than the shock. But that’s a small correction.)

In turn, this negative shock is like a negative shock to government purchases of goods and services. So it should have a similar multiplier. Multiplier estimates are all over the place, but tend to cluster around 1.5. So we’re looking at a negative impact on gross world product of around 1.4 percent. Not huge — China isn’t the principal obstacle to recovery — but significant.

And, if we think of the United States as bearing a proportionate share, and also use the rule of thumb that one point of GDP = 1 million jobs, we’re looking at 1.4 million U.S. jobs lost due to Chinese mercantilism.

And we have a serious problem. There’s no model, there’s no evidence, there’s not really even an argument. According to Krugman, the only thing a Chinese current account surplus does is reduce the rest of the world’s surplus. Putting aside how wrong that is, even in a blog post, it begs the question. Austin turns an extensive modeling exercise concerning the balance of payments into a simple, striking claim about jobs, purely on the basis of Krugman’s assertion. Then Bernstein claims the result is convincing. No, it’s still an assertion.
If you believe what Krugman wrote in 2009, you can follow the chain from there. But your belief is based on faith, not on facts or even published academic theorizing. Krugman comes close to rendering trade irrational. Why does the rest of the world trade with China at all if the only thing that happens is China’s surpluses displace their own?
Krugman is wrong; that’s far from what surpluses actually represent. Chinese assembly of phones and tablets and production of clothes and toys benefit the world. It allows people to buy cell phones and computers who otherwise could not. It frees up money that would otherwise go for more expensive clothing, allowing it to be devoted to other things. It boosts our economy in many ways, thus making it possible for more people to be employed. It can also support employment directly.
Krugman makes an intertwined error, also implicitly embraced by Austin and Bernstein: “one point of GDP = 1 million jobs.” GDP can be defined using the trade balance. Holding everything else constant — a gigantic assumption — a larger trade deficit means lower GDP. Then Krugman slips in another assertion: that lower GDP means lost jobs. In sum, other economies’ current account surpluses cause our current account deficits, which lowers GDP by definition, which causes job loss.
This is dubious for multiple reasons but most important is: GDP does not cause anything. It’s just an accounting device.
Job increases or losses are changes in the real world, with real impacts on consumption and production that can be measured in GDP. But it does not work the other way — GDP changes are not changes in the real world which cause jobs shifts. Saying GDP causes job losses is like saying your scale causes weight loss. The link between the balance of payments, GDP, and jobs is not as Krugman assumes, and Austin’s (and thus Bernstein’s) job numbers are based on Krugman’s false assumption.
After all this, it is in fact true that China is a trade predator. The problem is not criticizing China, the problem is doing so incorrectly.

Derek Scissors is a resident scholar at the American Enterprise Institute.