I was recently in Sri Lanka, where, thanks to our friend Ahilan, Arjun and I were able to meet with a number of people in the new left-wing government there. The challenges and opportunities there are something I hope to write more about more in the future. But in this post they’re a jumping-off point for economic theory.
Like many governments coming into office in the wake of an economic crisis, the government of Sri Lanka is currently involved in negotiations with the IMF. Discussing this got me thinking about the theory of comparative advantage, which so often comes up in debates over trade policy. It seems to me that Ricardo’s old model is indeed relevant to the choices facing a country like Sri Lanka – but not in the way that people sometimes think.
If you’ve studied economics at all, you’ve certainly encountered the idea of comparative advantage. But let’s spell the argument out, just to be clear.
Imagine two countries that produce various goods. For simplicity we’ll say the there is only one input, labor. A certain amount of labor is available in each country, with various amounts of it required to produce the various goods — so many hours for this one, so many hours for that one, and so on. Unless the labor required per good is in the exact same proportion in both countries, both of them can benefit from trade, in the sense that each can get a strictly larger basket of goods from the same amount of labor than the best basket each could make on their own.
The canonical version of this argument comes from David Ricardo, showing why Portugal and England are both better off when the former specializes in wine and the latter in cloth.1 If your economics education was like mine, a proof of it was literally the first thing you learned in your first class.
Obviously there are various assumptions behind the proof, which one can criticize on various grounds. I’ve done my share of criticizing them myself, in the past. But there are questions for which the framework of comparative advantage is genuinely useful, and which are more interesting than the textbook trade-versus-autarchy framing.
Here is another result that is less familiar, but that follows just as directly from Ricardo’s model.
Suppose that one country’s productivity improves, in the sense that it now takes fewer hours to produce the same good. Who benefits from the improvement? The answer is: If the productivity improvement is in a good that the country currently exports, then its trading partners benefit relatively more. If it is in a good that the country currently imports but also produces at home (or could produce at home after the improvement), then the country itself benefits relatively more.
It is easy to prove this algebraically or geometrically, but we don’t need to; the logic is straightforward. A productivity increase for good x means an increase in the supply of good x. This leads to a fall in the price of x. If we stick with Ricardo’s model of two goods for simplicity, this means, precisely, that one unit of good x trades for fewer units of good y.
If x is a good that the country imports, then it benefits twice: it produces more x at home, and it gets more x for each good that it sends abroad. If x is a good that the country exports, on the other hand, then it still benefits by being able to produce more at home but that is offset by the fact it now gets less other stuff for each x that it gives.
In fact, productivity improvement in a country’s export sector can leave it strictly worse off, if demand for its exports is sufficiently price-inelastic.
Here’s a simple example. Let’s say there are two countries producing two goods, apples and bananas. Initially, one apple trades for one banana. As it happens, our country is more productive in bananas — one hour of labor produces two bananas, and only one apple. The other country is relatively more productive in producing apples.2 In this case it makes sense for our country to specialize in banana production. After all, one hour of labor can produce one apple directly; or the same hour can produce two apples indirectly through trade, by producing two bananas and trading them for an apple apiece. Obviously the second option wins.
This is your textbook story.
But now let’s suppose that our country’s productivity in bananas improves to three per hour, while the other productivities stay the same. This increase in the supply of bananas means that the price of bananas (in terms of apples) must fall. How much it falls depends how elastic the demand for bananas is — when bananas get cheaper, do you want a lot more of them, or only a few? The more inelastic demand is, the more the price of bananas will fall. Suppose demand for bananas is rather inelastic, so that at the new equilibrium the relative price of apples has fallen by half, meaning it now takes two bananas to buy one apple. It is still worth it for our country to specialize in banana production — trading yields one and a half apples per hour of labor, versus one apple if it is produced at home. But now it is no longer possible to get two apples for an hour of labor, we only get one and half. Consumption baskets that were achievable before (say, one banana and one apple for each hour of labor) are no longer available as a result of our productivity increase.
You might say that this is as artificial and abstract as the original story. And that’s true. It’s the same model, after all, with the same assumptions. But I think it’s interesting.
One thing that’s interesting is that this very familiar model has this other less familiar implication. But more importantly, it is a useful framework for thinking about some important real world questions.
The possibility that a technological improvement in one country can primarily or even exclusively benefit that country’s trade partners, depends on the assumptions that demand for the good in question is more or less inelastic, and that the exporting country is large enough to influence world prices. The first assumption is reasonable enough for many primary products like, well, bananas. The second is not so reasonable for an individual country, but certainly applies to the exporters of some particular commodity in the aggregate. An individual banana producer that increases productivity will probably reap the benefits without inducing adverse movements in its terms of trade. But if productivity rises for banana production in general, it’s much more likely that the gains will be canceled out by a fall in the price of bananas. This is why development economists sometimes talk about a “fallacy of composition” — for an individual country, lowering costs for current export industries may seem reasonable, but if all the countries exporting similar goods do so, they may all end up worse off.
Another implicit assumption in this story is that while countries may be large relative to the markets they participate in, individual producers are small. This is why we say that prices are set by supply and demand. If this isn’t true, and producers have significant market power, then productivity improvements are more likely to show up as higher producer incomes rather than lower prices for consumers.
This was a core insight of the theories of unequal exchange developed by people like the great Argentinian economist Raúl Prebisch. Because primary products and lower-end manufactures have relatively inelastic demand and are sold in competitive markets, the benefits of productivity increases go mainly to consumers, while more sophisticated manufactured goods have elastic demand and a high degree of market power, so the the benefits of productivity improvements go mainly to producers. Which is why, from the point of view of a primary product producing country, you want to get out of your comparative advantage rather than leaning into it.
The story I am telling here is very familiar in development economics. Economists from the South (like Prebisch, or Arthur Lewis) long ago understood the importance of changes in the terms of trade. But the same dynamics can apply in domestic contexts too, with changing terms of trade between industry and agriculture, which has often constituted a sort of domestic periphery to industrial countries’ urban cores.
Historically, productivity growth in US agriculture has been considerably faster than in manufacturing. Over the postwar period, labor productivity in farming has increased by about 3.3 percent per year on average, compared with 2.2 percent in the nonfarm economy. Does that mean that a rising share of national income has gone to farmers? Not at all. In fact, essentially all of the gains from rising farm productivity have been captured by the consumers of farm products — the 1.1 point difference in average productivity growth between farm and confirm productivity growth is almost exactly equal to the difference in the growth in the prices of farm output versus the price of output as a whole. All of the gains from rising farm productivity, in other words, have come in the form of cheaper food (and higher incomes for the processors and distributors between farmers and consumers). The reason that this has not led to mass poverty in the countryside is that people can and did leave for jobs elsewhere. This option is less available when core and periphery are on opposite sides of an international border.
(The combination of productivity-boosting investment in agriculture and falling prices that meant the benefits all flowed to consumers rather than producers, is not something that just happened. It was the result of a particular model of public investment in agriculture — a point emphasized by the great Marxist biologist R. C. Lewontin.)
The same comparative advantage model has other implications. For example, while two countries are always better off by trading with each other, it is not the case that two countries already trading with each other will always be better off by trading with a third. The more similar the new country’s relative productive are to one of the existing countries, the more likely that country is to be made worse off by trade with the newcomer. A country that imports and exports a similar mix of goods as our country offers us little opportunity for further specialization, while its presence in the trading system will move the terms of trade unfavorably to us.
This was the topic of one of Paul Samuelson’s last published papers. One guess what countries he chose for his examples.
One might wonder, again, if this sort of story is relevant to the situation of a small country. Sri Lanka is responsible for something like 14 percent of global tea exports, which is a lot, given the size of the country. But it’s not so much that one would expect an increase in production there to result in a significant fall in global tea prices. Its share in apparel or tourism, or in any of the things it imports, is much smaller. Still, I wouldn’t dismiss the terms of trade considerations even for a small country; there are any frictions in world trade that can mean that there is something very far from a single frictionless world market for most commodities. There is a reason that one of the most successful empirical models in world trade is the “gravity model,” which predicts that trade between any two countries is inversely proportional to the distance between them. In a world where trade frictions mean that in practice trade happens much more easily between some countries than others, a country might make up a very small part of global trade in some commodity but still have significant market power with respect to its own trade partners.3
I wouldn’t want to push this story too far. A framework that considers only trade flows leaves out finance, exchange rates and the balance of payments, the latter of which is arguably a much more immediate problem for Sri Lanka and for many similarly situated countries than the terms of trade. The worst part of the IMF’s advice isn’t the suggestion to develop existing export industries, but the insistence on removing existing controls on financial flows (which are surprisingly robust, in Sri Lanka’s case).
Still, the comparative advantage frameworks focuses attention on one key point: Any change in countries’ productive capacities, or in the mix of countries trading with each other, will affect relative prices. And in any such price change there are opposing interests between exporters and importers. Ricardo’s model is really a model of conflict — between town and country, between industrial and countries and producers of primary products, between North and South. The abstraction away from finance and exchange rates, in this context, is helpful to focus attention on the terms of trade.
Sri Lanka’s own policies may or may not have any impact on the relative prices faced by it or its trade partners. But the policies adopted by countries like Sri Lanka in the aggregate certainly will. Which is an important thing to consider when thinking about what kind of advice IMF economists are likely to give. Which countries are they mostly from, where do they get their economics training, which governments pay their salaries? Incentives matter, after all.4
Back when I taught this stuff, I used a version of this as an exam question. If a country is considering policies that could shift its comparative advantage in various directions, why advice will an economist working for that country’s government give? What advice will an economist working for one of its trade partners give?
As a matter of fact, something like this question is the original context for Ricardo’s model.
As described by the great Marxist economist Harry Magdoff, and more recently by Matias Vernengo and Jesus Felipe, the relative advantages that Portugal enjoyed in wine making and England in cloth making was not due to longstanding differences in climate or culture, but to politics and (relatively recent) history — in particular, the 1703 Treaty of Methuen, which Portugal agreed to under duress after it became clear that they were unable to protect their colonies without the cooperation of the British navy. As Felipe and Vernengo write:
During the last years of the 17th century, Portugal had developed a protectionist policy… Simultaneously, Portugal had made successful efforts to stimulate the domestic manufacture of cloth. This negatively affected British manufacturers and merchants. … The Methuen Treaty stipulated that Portuguese restrictions on English cloth and woolen manufacture be lifted. In return, Britain guaranteed a lower tax on Portuguese wines than on French wine. The result was obvious: Portuguese cloth manufacture was strangled in its infancy. … instead of developing a dynamic garment and textile industry, Portuguese capital flowed massively into wine making…. On the other hand, England expanded its garment and textile industry, and the achievement of larger-scale production meant a reduction in costs…
This was the origins of the division of labor that Ricardo would go on to present as a mutually beneficial adaptation to inherent capabilities.
To be clear, creating a more favorable international division of labor was not the sole or probably even primary goal of the Methuen Treaty. Presumably what was uppermost in the minds of the British negotiators was securing a strategic advantage in their long conflict with France. For which purpose locking in Portugal as an alternative source of wine might have been as important as gaining a captive market for British textiles. It’s funny to think that the development of the theory of comparative advantage at the dawn of the 19th century, like much of development economics in the 20th, was a side effect of imperial struggle.
Be sometimes to your country true,
Have once the public good in view;
Bravely despise Champagne at court
And choose to dine at home with Port.
- The original version is in Chapter 7 of his Principles of Economics. The gains from trade that modern economists focus on are not Ricardo’s main interest there; the question he wants to answer is the effect of trade on the rate of profit. (Classical economists in general were more interested in the distribution of income between classes than in the problem of efficient allocation that is the main concern of mainstream economics today.) Ricardo also doesn’t offer any story for how prices will be set except that they will be somewhere between the two sets of prices that would exist under autarchy.
- Meaning it takes less than twice as much labor to produce an apple there as it does to produce a banana.
- This is analogous to the reasoning behind monopsony models of the labor market. In the presence of frictions, a transactor doesn’t have to be large in absolute terms in order to have significant market power.
- It’s far off the subject of this post, but it seems to me that the question of whose incentives are considered and whose are ignored, is a very common way that a political program gets smuggled into an ostensibly positive or objective economic analysis.