Well, as I think I've advertised already, I doubt this is so. Long ago, before I took any economics courses, I was struck by the fact that the USA had a ballooning trade deficit during recessions.


Note the trade deficit is increasing during periods of rising unemployment. Mr. Baker would no doubt point sternly at the data points for '98, '99 and '00: these are years for which the trade-weighted exchange rate of the US dollar jumped, while the Korean won, Brazilian real, Mexican & Philippine peso, and Thai baht lost large chunks of their value against the US dollar. QED, the plunge in currencies caused the predictable plunge in US exports, while imports from those countries surged.

For those wondering what the historical record is, here is a set of charts illustrating these international transactions over time. They're not the prettiest or most detailed of charts, but this is supposed to be a quickie post and they were updated yesterday. Notice the trends in American imports over the period 1997-2000: steady upward curve. Exports took a fierce hit, and both are still consistent with what Dean Baker was trying to say: if the currencies of some major trading partners sagged, there might not be that much of an impact on USD-denominated purchases, but obviously USD-denominated exports to that area would plummet. The reason is simple: a 50% drop in the prices of, say, items manufactured in Thailand might cause sales of those items to double, but probably not triple. In the case of Thailand, as much as 60% of the factor inputs involved in production may have come from foreign sources whose currency didn't drop, so a 50% drop in the baht:USD ratio would only cause Thai products to fall in price by 20%. Of course, the ability of Thais to buy American goods with the same number of baht will plummet by 50%.

But the trend overall has been very different. The Asian crisis was a big anomaly, and Baker would probably never advise the Treasury to unilaterally devalue the USD 50% in order to maintain the prior balance in current accounts. Moreover, Baker would probably explain that his principle applies ceteris paribus: in point of fact, a rise in the dollar could conceivably be caused by a stampede towards US equities, depending on the situation in other equities markets. Such an event might accompany large purchases of American capital goods (far and away, America's biggest export), as occurred in mid-'95-96. Naturally, demand for the dollar surged. But I am saying something different.

I have here a copy of International Finance & Open-Economy Macroeconomics (Giancarlo Gandolfo, 2001). In this book, he explains the Marshall-Lerner conditions, viz., that a shift in the relative exchange rate can do different things to the trade balance. It all depends on the elasticity of imports and exports! Well, of course it does! Thus, there is a rate of change of imports per change in the price of the imports; if the cost (in USD) of oil goes up 10%, imports may decline 5%. In order to establish whether imports increased, we multiply 0.95*1.10 to get 1.045, or an increase in USD-denominated oil imports of 4.5%. these are all made-up statistics, but I think you get the idea: American habits of oil or gas consumption are determined by the existing stock of motor vehicles and machinery, or existing stock of housing.

BILL: So what're the Marshall-Lerner conditions?

JRM: Wow, you came in so quietly...

BILL: I like the music you're playing. It's not the Messiah...

JRM: Saint Matthew Passion. Peter's Denial of Christ.

BILL: That explains why your eyes are wet. Cheer up, the Marshall-Lerner Conditions.

JRM: Well, suppose you raise the price of a thing 1% and sales go down 1%. That's an elasticity of one, or "unity." And let's suppose, since it's closer to the truth, that elasticities are continuous—so demand curves aren't straight lines, but convex with respect to the origin. So in this picture, we have the unusual case of constant elasticity of one and no matter what the price is, the amount of money spent on it is the same. In real life, there's going to be some falling off at the edges, and time has an effect too, but it's an approximation.

BILL: Got it. And I suppose the people we export to have the same elasticity?

JRM: The same principle, anyway. Now, if the elasticities of price are both one, and the dollar falls 1%, then exports will go up 1% and imports will go down 1%—which will cause our trade deficit to shrink. From our point of view, if the sum of the elasticities is one, then there will be no net change in the trade balance. If the sum is greater than one, then the change in prices will be more than offset by the change in quantities demanded. But if it's less than 1%, then, in US dollars, the trade balance will worsen. The Marshall-Lerner conditions apply if and only if the sum of the import and export elasticities is greater than one—which is usually the case. But for the USA, I strongly suspect that a time-series regression would show that the sum of the elasticities is usually less than one, and the effect of a decline in the USD is to raise the price of imports by a higher percentage than the reduction of the quantity of imports. The converse is probably true for our exports.

BILL: So you think a strong dollar will improve our trade balance? That's nutty.

JRM:That's what Alfred Marshall thought when he first suggested the idea. I guess it depends on which currencies the US dollar performs best "against." My suspicion is that the Marshall-Lerner conditions don't apply for two reasons: one, the large role of petroleum imports, for which price elasticity is small; and two, the extreme duration of America's Balance of Payments (BoP) deficit, which has its own secular effect on the trade balance.

BILL: Actually, this chart shows that they do sum to 1.1 for the USA. For France and Germany they don't, not even for the long run. 'Course, the import elasticity is statistically insignificant. (p.4).

JRM: Okay, that's interesting. The Marshall-Lerner conditions were a false lead. Over the long run, there's actually a price-expenditure elasticity of .9. That's big—bigger than most countries.

BILL: "Price-expenditure" elasticity?

JRM: Usually "elasticity" is used to refer to the change in quantity for a given change in price. But suppose the price-quantity elasticity is one. Then your budget for that thing, what ever it is, is fixed. For economics students, that's what you get with Cobb-Douglas production or utility functions. So if the price goes through the ceiling, you still spend the same amount of money you did before. I call that a P-E elasticity of zero.

BILL: But in real life—ahem, Mr. Liberal—people get less satisfaction for their dollar if they get less of a certain thing for it. So they spend their money on something else. That ever occur to you?

JRM: Absolutely. That's why I was deadset against the Cobb-Douglas functions. They should have second-order conditions so that if you get less satisfaction, you take your business elsewhere. What, did you think I was James Cobb-Douglass? You should take it up with Herbert A. Simon or James A Mirrlees. So anyway, I dreamed up the idea of P-E elasticity just now. You get it by subtracting one from the normal elasticity. So if the elasticity is greater than unity, people respond by spending a smaller amount of money on the thing because they find substitutes. As you can see, for most items on this chart the P-E elasticity is actually less than zero; that means people bite the bullet and spend more with less satisfaction. My conjecture is that it takes a while for people to find substitutes, and in the meantime making changes poses other expenses.

BILL: That must explain your shift in musical tastes from Bach to Najma.