The Globalization Paradox and the perils of simple models

Reading recently: The Globalization Paradox by Dani Rodrik.

Should countries impose taxes on imports?

The bog standard economist answer is: no! The theory of comparative advantage says that trade is positive sum. Tariffs are a prisoner’s dilemma and we should fix the game theory with multilateral treaties. It’s too bad that lobbyists and entrenched interests fight against it.

There are (at least) three major things this answer leaves out.

It’s perfectly possible to incorporate these caveats into the naive model of inter-country trade. But most of the time you ask an economist about tariffs they don’t try to do that; they just tell you that the optimal level of tariffs is zero. Rodrik thinks that most economists justify these elisions by saying that you can make trade liberalization a Pareto improvement by implementing redistribution from the beneficiaries of free trade (i.e., net importers) to the harmed parties (domestic producers, unemployed people, and sectors with positive growth externalities).

But even if these redistributions are possible in principle, it seems very hard to size them properly. No one knows exactly how much it will harm domestic producers if you reduce import tariffs on their goods—let alone how much it will affect unemployment (which depends on skill transferability, union negotiations, future economic conditions, etc.) or growth (which nobody understands at all).

Given that, it’s actually hard to see how you could broker any deal involving such large transfers. If your grain tariff reduction takes away $50 from Pat and gives $51 to Dana, you could try to make it palatable by legislating a separate direct transfer of $50 from Dana to Pat. But if Dana underestimates how much they’ll benefit from the tariff reduction by even 2%, they’ll object that the second compensatory transfer is too big! In order to fix the redistribution, the parties need to agree really precisely on the magnitudes of the costs and benefits.

The pattern here is that the naive model assumes a way a bunch of stuff, and then treats it as a second-order problem when those assumptions are violated—but, because the assumptions are actually about very large effects, they only need to be violated a little bit in order to dominate the calculation.

What happens if you generalize this to other things that economists say?

I think this suggests you should freak out a lot more about technological unemployment. It’s true that the historical track record suggests that technological improvements are usually worth it in the long run. But, the historical track record also suggests that it’s often accompanied by very painful short-term dislocations, people running around the countryside smashing looms, and so on. It seems like it should be possible to do better—but it’s unlikely that we will do better if economists limit their policy suggestions to “yeah, it should be possible to fix that with some redistribution.”

More speculatively, I think it the same reasoning suggests that we should be a lot more confused about wage controls. This is another example where the naive economic analysis strongly suggests liberalization, but also makes a lot of spherical-cow assumptions. In this case, those assumptions are mostly about the efficiency of the relevant markets. It seems like it’s not too hard to imagine scenarios where a minimum wage improves efficiency—for instance, if local employers have much more market power than employees, or if the transaction cost of changing jobs is high enough that the job market isn’t very efficient, or if people are averse to changing jobs for behavioral reasons. An analogous case also applies for rent controls.

  1. The Globalization Paradox (p. 145), which cites this calculation as follows:

    Dani Rodrik, “The Rush to Free Trade in the Developing World: Why So Late? Why Now? Will It Last?” in S. Haggard and S. Webb, eds., Voting for Reform: Democracy, Political Liberalization, and Economic Adjustment (New York: Oxford University Press, 1994).

    As is shown in Rodrik, ibid., we need two other pieces of information besides tariffs to compute this ratio: the import demand elasticity and the share of imports in GDP. For the purposes of this exercise, I have assumed (generously) values of –2 and 0.2, respectively, for these two parameters.

  2. The Globalization Paradox, p. 322. ↩︎


email me replies

format comments in markdown.

Your comment has been submitted! It should appear here within 30 minutes.

I’d like to push back on the third point, since it’s an argument that often bothers me. The idea countries in East and Southeast Asia ought to be emulated for their aggressive tariffs and export subsidies seems to be a striking example of selection bias.

Many other countries in the world pursued Import Substitution Industrialization, and indeed often used the same Harvard-trained advisors as the east asian countries who grew, and failed to grow: this includes many countries in Latin America and Africa. Moreover, even looking within the countries it is unclear that the policies intended to promote export-led growth were responsible. The Taiwanese government, for instance, invested heavily in subsidising and protecting a car industry. Other industries took off in Taiwan, but not car manufacturing.

Massive investment in human capital is something that those countries did have in common. But it’s not clear that this was a result of their growth as much as the cause of it.


I’m actually curious because Rodrik does address this point in the book, but obviously has only one side of the story. So I always love to hear counterpoints like yours!

Here’s his quote addressing this, from p. 371, arguing that (a) ISI countries didn’t try hard enough to export, and (b) their protectionism actually worked out reasonably well anyway:

The misdiagnosis of the experience of countries such as Brazil, Mexico, and Turkey, which had followed more inward-looking strategies, was equally problematic. Unlike East Asian countries or Mauritius, these countries had made little effort to push their firms to export, relying mostly on the domestic market to fuel growth. They had maintained highly restrictive trade regimes well into the 1980s. This was the strategy of “import-substituting industrialization” (ISI), and it had become the dominant model in Latin America, the Middle East, Africa, and parts of Asia (especially India) since the 1930s and following independence. … The strategy placed little emphasis or confidence in the ability of domestic firms to export and compete on world markets.

The revisionists painted a grim picture of ISI’s record. By failing to take advantage of world markets and giving the state too large a role, they argued, these countries had severely handicapped their development. Once again, this depiction overshot the mark. … [T]he overall record of ISI was in fact rather impressive. Brazil, Mexico, Turkey, and scores of other developing nations in Latin America, the Middle East, and Africa experienced faster rates of economic growth under ISI than at any other time in their economic history. Latin America grew at an annual average rate exceeding 2.5 percent per capita between 1945 and the early 1980s—a pace that far exceeds what the region has registered since 1990 (1.9 percent).17 Two dozen countries in post-independence sub-Saharan Africa also grew quite rapidly until the mid-to late 1970s.

Industrialization drove this performance. ISI countries experienced rapid productivity growth as their economies diversified away from traditional agriculture into manufacturing activities. As surprising as it may seem, our best studies indicate that during the sixties and seventies economywide productivity grew more rapidly in import-substituting Latin America than it did in export-oriented East Asia. Latin America’s economies expanded at a slower clip than East Asia’s not because they experienced slower technological progress but because they invested a lower share of their national income.

If you’re curious I can give you the citations for this section as well. I’d love to hear your take on this point!

email me replies

format comments in markdown.

Your comment has been submitted! It should appear here within 30 minutes.