Why the theory of comparative advantage is overrated

Paul Samuelson once produced comparative advantage as an example of an economic theorem which was both true and non-trivial.  Obviously the theorem holds as proven, but how on the mark is the theory as an actual explanatory device?  (For background on the theory, you can start with Alex’s video here).  I hold a few reservations:

1. In virtually every other context, we insist that the dynamic perspective is critically important, yet we are remarkably content with the static take of comparative advantage.  If comparative advantage puts you into a job, or puts your economy into sectors, with no good learning curves, woe unto you.

2. The theory sits uneasily with the observation that long-term unemployment is indeed possible.

3. They do indeed send horses to the glue factory, so to speak.

4. In the real world, the “lawyer” and the “typist” are often not fully segregated workers, but they interact, whether in the workplace, in society, or politically.  Complementarity and O-Ring effects may be more important than comparative advantage effects, paging Gilbert Arenas.

5. It is easy enough to see how the theory works when you move from one person to two, or from five persons to ten.  But how about when two large nations trade with each other?  The macro embodiment of comparative advantage, namely the Heckscher-Ohlin theorem, has at best a mixed performance empirically and capital and labor endowments do not appear to predict the content or amount of trade very well.  Furthermore even defining comparative advantage (how do you measure the quality of labor or capital?  Is the U.S. in fact a capital-intensive or labor-intensive nation, taking qualities into account?  How about human capital?  etc.) is deeply problematic once you move past the basic Heckscher-Ohlin assumption that different nations have access to the same production function (and of course that assumption is obviously false).

6. In the simplest explanation of comparative advantage, we measure the productivities of the lawyer and the typist in terms of hours, a physical unit.  Usually, beyond a single input world, we need market prices to measure comparative advantage.  But then the original comparative advantages are endogenous to specialization decisions and other economic factors.  The true theory of comparative advantage, for a multi-factor world, isn’t nearly so simple.  It also can be said that the initial comparative advantages are in fact endogenous to trade.

7. Many of the most important gains from trade come from other mechanisms, including specialization, increasing returns, or the generation of commercial networks which lead to a later transmission of ideas and technologies.  That is not exactly an argument against comparative advantage, but it does suggest the idea is only one part of the case for trade.


The model fails if it assumes capital remains once the jobs leave; and if capital leaves then it is a complete flight from the economy. The British don't export textiles to Portugal for wine, the British close their textile factories and move them to Portugal, where they make both textiles and wine.

And then what does Britain do? Just import everything?

It would be short-circuited earlier on because the British wouldn't be able to buy anything abroad without having exports. In the short run, they could borrow money, but investors aren't going to pump money into a country if it's not going to pay them any returns.

They provide financing and management for the new British owned Portuguese textile mills.

It is a good deal if you are in management or finance.

If nearly all the capital moves to Portugal then Britain could specialise in goods that don't require much capital and still trade with Portugal.

(I say "nearly all the capital" because land is capital and if all the British land moves to Portugal then there's no more Britain left and the question of comparative advantage is meaningless).

Well, the don't physically move their textile factories.
They build new ones in Portugal, and shut the old ones down.
There is a capital loss on one end and a capital input on the other.

Two other advantages of trade: 1. Spreads language. I believe human language (many words) began and grew because it facilitated trade. Pidgin languages illustrate the language-generating property of trade. 2. Reduces hostility. Face to face contact reduces hostility. A friend of mine was in the French navy. Every lunch, everyone on the ship ate together, both officers and privates. He said this made it so much easier to resolve problems that came up. Your commanding officer wasn't suspicious of you, you weren't suspicious of him.

Your #2 using the military as an example was surprising. I thought standard military theory holds that forbidding fraternization between the ranks is a good thing for morale & discipline.

Only if subordinate officers holding the personal loyalty of their men is still a threat to unity of the armed forces. In the case of modern France, this is doubtful. The Algiers putsch failed to sway the army, and that was in the 60s.

As an aside, doesn't the US military frown upon fraternization too? Or is that history?

Yes, but fraternization doesn't include everyone having lunch together.


Where's the line drawn? Dinner together? Drinks?

Genuinely curious as to what would be considered fraternization.

If everyone is having dinner together on the ship in public, that's not fraternization. If you're inviting someone to your house for dinner, that might be. If alcohol is being served at an official social function, it's not fraternization to talk to each other. If you go out drinking at a bar with someone, that could be. The purpose is to prohibit certain types of social and business relationships. You don't want anything that looks like favoritism, and you don't want relationships that put pressure on people in certain ways. Everyone having a meal together in public doesn't really do that.

"The Algiers Putsch"? There were two. The one in 1958 succeeded, while the one in 1961 failed. Did they change fraternization in the three years in between?

But the 1958 coup did not pit army officers against army officers. It pit the army against the civilian government. That was why it ended the Fourth Republic.

What do you mean? Every coup starts small. Some soldiers in Algeria must convince the rest of Algerian army, then the mainland army. At some stage it is military against military. Perhaps you mean that it is harder for a military to unify behind removing a former general than installing one? That is probably true. De Gaulle certainly thought it worth reminding them that he was a general.

But what does this tell us about fraternization?

"But how about when two large nations trade with each other?" Surely it's largely individuals and companies who trade with each other, rather than countries?

Important point. I am not sure that comparative advantage applies in a clear way to a global economy with long and dispersed supply chains. Globalization has eroded the nation state through trade. The notion that "nations trade with one another" is in fact antiquated.

Isn't specialization a form of comparative advantage?

It's the very definition of comparative advantage in many ways: countries focusing on particular tasks, or sub-tasks of tasks, because it's cheaper to do it there compared to elsewhere.

Maybe the whole idea of talking about "countries" specializing in various things is wrong-headed. At least it will probably lead to bad policy if it leads local government to attempt to direct what industry they will specialize in.

People specialize in various industries. How they are distributed across countries varies. In order for a perfect equilibrium to be achieved, you must allow the the free movement of both people and capital. Right now, we have a situation where as the geographic location of textile factories moves, the textile-worker labor force cannot follow it.

Yes we know this. I believe Krugman won a nobel prize over this issue. What else is new?

This sort of complicated post deserves a more elaborate video at MRU. Let's get some meat into those videos! I don't want to see another simple two country - two product example explaining comparative advantage.

Also if not already mentioned in TC's cryptic and dense prose post, the Dutch Disease that comes with resource comparative advantage should be mentioned.

Finally, I am reading some books that make the case that there are knock-on effects (externalities) to having manufacturing, perhaps akin to Krugman's network effects to specializing in one thing. If true, then maybe there's a case for an infant industry argument, though I think it's a weak case and more likely protectionism will follow.

In spite of multiple attempts I've no clue what #2 or #3 mean.

Can anyone expound? Where does horse glue & unemployment fit in all this?

I believe it's an allusion to technological unemployment. There's a common rebuttal to those who deny its possibility that the advent of the automobile lead to the technological unemployment of the horse, which rather than somehow finding a new occupation mostly got put out to pasture. If your country chases its comparative advantage too heavily, and then your trade becomes obsolete, will you easily restructure or will you so to speak be turned into glue?

Thanks. But that argument (if indeed that was TC's intention) sounds a bet underwhelming.

Of course, specialization has its downsides; it's hardly as if Samuelson was unaware of that. Monocultures always are susceptible to external shock. Deep specialization comes with the cost of reduced robustness & less flexibility.

But those are hardly novel insights.

Mind sharing the books you're reading regarding manufacturing externalities?

Nations don't trade, but individuals and firms in different nations do trade. In a free market political boundary lines are economically irrelevant. Indeed in a real free market they don't exist. The "NP" dished out to PK was useless. It was bestowed more for his anti-Bush statements than for his economics.

This. This. This. THIS. is what should be in your textbook!!

I think it is more complex.

Industries form networks (supply, distribution, knowledgeable employees) and while one factor input may be cheaper in another country, the network in the other country is not efficient or desireable.

I think you should look more at networks of comparative advantage. If you can maintain the network, or improve parts of it, you still win, even if there is a comparative advantage for one input in the other country, such as cheaper labor. All components of the network are more important together than one alone.

Including that word that you didn't want to hear: infrastructure.

I think this may apply to conditions like Autism in that autistic people may have a comparative advantage in something but little in the formation of a natural network for employers to use this labor efficiently which would cause long term unemployment even though they have the comparative advantage.

Wouldn't that just mean that the other country would have a a comparative advantage in things for which infrastructure and other network effects aren't that valuable, comparatively?

Of course the second country might be better off investing in infrastructure and other network effects of their own in the long term. All the theory of comparative advantage says is that it's still profitable to trade even if the other guy is more productive than you at everything, it doesn't say there's no gains from improving your own productivity.

If you are a medical device manufacturer, for example, you want a plastic molder specialist, specialists on electronic components and assembly, a local medical school faculty, local investors who are familiar with technical issues, etc.

If you look at trade associations for certain products, say medical devices, you will see clusters of producers. If you believed that it was just cheap labor, they all would be located in Mexico or China, but they are located in high cost urban areas because the expertise for existing firms and start ups are conveniently located.

An interesting project would be to identify various product trade associations and identify the plant locactions of members to detect clusters.

That may be. The theory of comparative advantage merely says that even if high cost urban areas are more productive at producing everything than Mexico or China are, it's still possible for Mexico and China to profitably trade with high cost urban areas by specialising in whatever Mexico and China are least relatively bad at.

"paging Gilbert Arenas"


"Externalities are the last refuge of a scoundrel. " :-)

For any economy to be able to fully utilize comparative advantage for the long run, they need to establish most of their services in terms of endogenous wealth formation (fully validated and built upon at endogenous community levels, instead of highly limiting exogenous institutional levels). Then, and only then, can economies truly gain from the exogenous factor of separate product formation and commodities in their naturally exogenous settings.

For any economy to be able to fully utilize comparative advantage for the long run, they need to be able to establish most of their services in terms of endogenous wealth formation (fully validated and built upon at endogenous community levels, instead of highly limiting exogenous institutional levels). This would allow local economies to fully gain from the comparative advantage of separate product formation (separate from time use) and commodities in their naturally exogenous settings.

I love this new feature of MR where people attempt to re-write their previously posted comments.

What's taking that annoying bug so long to fix. Or will it be rechristened a feature eventually?

I thought maybe it was something they put in deliberately to prevent people from arguing too much.
Only one response per hour, people!

Holy God, it's fixed!

Can you please name those economies which you think have historically achieved what you outline?

I ask because "fully utilise" is an awfully tough standard to achieve. So I'm very curious as to where and when you think this happened.

The comment about Krugman is indirectly relevant. Here's a direct test of the theory: http://economics.mit.edu/files/6714. The punchline: "the disappearance of [Ricardian] forces would only lead, on average, to a 5.3% decrease in the total gains from trade."

Tyler, trade can mean two things -- trade between people, and international trade. As you wrote, in daily interactions we see comparative advantage at work all the time. Think about work on a group project. If I am a bit better at running regressions than my foreign partner, but far better at writing in English, he is going to write the program and I am going to write up the results.

International trade, which is what you are talking about, is totally different. Why might countries matter at all, other than being an arbitrary partition of the world? Hecksher-Ohlin says that capital and labor are stuck inside countries. New trade models say that technologies are stuck, either within countries or within firms which are stuck in countries. A recent statistical paper by Armenter and Koren argues that maybe countries don't matter much at all, as many puzzles in international trade are just a result of random chance and a finite number of firms.

Another way of putting it is that home bias and ridiculously high trade costs are the burning puzzles in international trade. To square the level of international trade in the world and the level of domestic trade, trade models must either posit unbelievably high international costs of trade, or alternatively that people very strongly prefer goods made domestically. Something is missing. Something matters about countries, and no one really knows what it is.

Countries matter because they often form a political unit where merchants can easily lobby the government to try to keep out their competition, based on mercantilist theories.
Not to deny of course the existence of more more local protectionist movements nor to deny that some countries at times are in too much disarray for such lobbying to work.

Comparative advantage theory deals with production by labor and consumption by laborers.

Corporations engage in trade that is often based on pillage and plunder. Nations often base their economies on pillage and plunder.

Simple example would be Saudi Arabia which at one level looks like a rich nation that demonstrates the benefits if comparative advantage. It looks like camel jockeys can labor to produce and sell energy based on the unique skills of the camel jockeys, and in exchange they get to consume vast quantities of gold and luxury cars.

But the reality is Saudi Arabia is just a nation-corporation with citizen shareholders pillage and plundering the land they by luck live on, with the shareholders getting paid to not overthrow the CEO and BOD. While everyone in theory knows the oil and gas will run out and they need to build a real economy that fits the Adam Smith model, they see themselves as too rich to actually labor, but instead all demand to be managers who direct labor. The UAE with a fraction of the population is embarked on building an office park for the world so each citizen can be a manager of real estate. In any case, the citizen shareholders have very little control over the management, just as we see in all the US based global corporations, so a great deal of frustration exists, with the management simply focused on short term returns to the shareholders by any means.

While China seems to be engaged in similar pillage and plunder like Saudi Arabia, but of its labor, China is running, contrary to claims, a fairly balanced trade, just not with the US. China uses its labor to sell goods to the US and then uses almost all the cash buying resources from Oz and the US and Africa to build productive capital assets in China that will benefit most Chinese citizens. Yes, there are wasted investments, but that is no different than the wasted investment in capital assets in the US - lots of real estate has been built in the US in expectation of demand that never materialized just like in China, and in the US, government redistribution of wealth took place to pay for the losses just like will occur in China. Donald Trump has resorted to government redistribution of wealth several times by declaring bankruptcy whenever he has made bad investments in real estate that were wasteful.

When someone makes the case for free trade in pillage and plunder goods, say for the Vikings to raid the Barbarian Horde villages for pillage and plunder and then the countervailing pillage and plunder raiding by the Barbarians on Viking villages as enriching both the Vikings and Barbarians, then we will have comparative advantage model covering the real global economy. I recall a justification of pillage and plunder by raiders as virtuously redistributing the wealth that was poorly utilized by mercantilists.

Thank you for writing "camel jockey" so early in the post. That's a Stop sign.

"to build productive capital assets in China that will benefit most Chinese citizens"

And what mechanism is there to ensure these capital assets benefit most Chinese citizens?

As a consistent follower of Tyler and someone who has taken graduate studies in trade - I have no idea what he is talking about. Here is a list of phrases that are absolutely confounding to me:
"comparative advantage puts you"
"good learning curves"
"long-term unemployment is indeed possible"
"fully segregated workers"
"initial comparative advantages are in fact endogenous to trade"

Comparative advantage is easily recast as a dynamic theory. Country X will invest in human or physical capital in producing commodity C if the payoff in the future generates a greater return on investment than any other return on investment in commodity D. And that return on investment depends on expected comparative advantage, which in turn depends partially on present factors.. Israel did not have a comparative advantage in producing avocados until they were planted, but it will never have a comparative advantage in producing redwood lumber.The same holds for individuals within the country. It about potential given the realities of what can be changed and what cannot..

If the typist went to law school, then maybe the typist would have the comparative advantage in law after all. Does the theory of comparative advantage take human capital formation into account?

The original hypothesis was that the lawyer was both more productive as a lawyer and a faster typist than the typist.
If the typist goes to law school and improves their lawyering ability to become more productive as a lawyer, but doesn't improve their typing at all, then the ex-typist now has an absolute advantage as a lawyer and the original lawyer now has an absolute advantage as a typist. So anyone should be able to see the advantages of them trading.

The comparative advantage argument is agnostic about how the differences in productivity came about. All it shows is that trade can still be mutually beneficial even if one party is more productive at everything.

Overrated or not depends on the context. I'd say it's not overrated in refuting naive merchantlism, which is still about the only alternative basis on which trade issues are discussed publicly in the US.

The discussion reminds me of "There is much that is good, and much that is new. Only the good is not new, and the new is not good" :-)

assumptions make an ass. out of all of us. Caplan's latest shows his ass-burgers quite clearly.

While I find most of these points, um, strained, in the "good sentence" category I now include: "Complementarity and O-Ring effects may be more important than comparative advantage effects, paging Gilbert Arenas."

I never thought the special status of comparative advantage was as a general theory of everything there is to say about international trade but rather as a refutation of zero-sum economic thinking about trade and markets generally. In most cases it is the first time an economics student is presented with an intellectual challenge to their natural intuition that every trade must have a winner and a loser.

This was a borderline Tyrone post.

Yep, I'm considering this a Tyrone post.

Ricardo's famous numerical example has not been overrated but rather completely misunderstood. For a correct understanding of the numerical example one should read the Principles and if one doesn't get it, then perhaps this paper might help: http://www.academia.edu/1948311/Comparative_advantage_and_the_labor_theory_of_value

I don't understand this post at all. None of it. This is not atypical of my experience with MR, which strangely keeps me coming back for more. Someone should start a parallel blog to translate TC's posts for people not inside his head.

Comments for this post are closed