As a followup to the last post, Professor Dave Donaldson kindly agreed to speak on his work in trade economics. Professor Donaldson is a Professor of Economics at MIT and the winner of the 2017 John Bates Clark Medal, awarded by the American Economic Association to “the American economist under the age of forty who is judged to have made the most significant contribution to economic thought and knowledge”. He currently serves as a co-editor at Econometrica and as a program director for Trade at the International Growth Centre. It was a privilege and pleasure to speak to him about his work and its policy implications on trade policy. (The interview is edited for clarity and brevity.)
Trevor Chow: Thanks a lot for agreeing to talk about your research. To begin with, I wanted to discuss industrial policy, which has seen a resurgence in the 2020 election cycle, whether that’s in the form of the Buy American plan from Biden or Trump praising the USMCA coming into effect. Based on your research on external economies of scale, how large do you think the gains are from using industrial policy?
In some sense it depends on what you mean by large. You can definitely make an individual sector or individual region grow - and sectors are often quite geographically concentrated. So industrial policy can have non-trivial effects on a given industry, but the research across the board suggests that it would be difficult to get large beneficial effects for the economy as a whole.
Why is this? For one, when people imagine industrial policy, it is typically seen as quite targeted - which inherently means that any gains to that sector are limited by the sector being a small part of the economy. Second, even a hypothetical industrial policy across every sector, which hasn’t been seen before but we could imagine it, would be limited by the scarcity of factors of production. Whenever you try to make one sector larger, that’s going to constrain the growth of another sector. Finally, our simulation suggests that making a sector much larger requires demand for those goods or services, which is usually fairly fixed - that is, the main areas where this might be beneficial is in sectors that export and hence for which domestic demand is not the limiting constraint. The problem with that is a lot of modern economies specialize in services, which may not be tradable as easily. All of these reasons make me skeptical of large overall gains from industrial policy, and our numbers suggest the gains from a truly optimal economy-wide industrial policy for a large and fairly closed economy like the USA would be about 0.5% of GDP. Even for small and open economies like Ireland it would be less than 2%.
TC: Given a lot of these limitations, would you consider optimal trade policy to bring larger gains?
In the research I referenced beforehand, we compared the gains from optimal industrial policy with those from optimal trade policy, where we assumed that the only reason a government would impose any trade restrictions was to provide favourable terms-of-trade. This could provide some gains, although I am a bit skeptical of the idea that, in many settings, the world price can be affected much by the change in demand from a single country. As for the competition-increasing role of trade, we still need more quasi-experimental evidence on that, but we have strong theoretical instincts that it is beneficial, all else equal. Of course that speaks even more in favor of a laissez-faire trade policy.
TC: How might the benefits of trade play out in developing countries that face a higher degree of risk from things like food insecurity?
Free trade is a controversial proposition these days in many countries, and if you look at the political rhetoric you’ll definitely see some opposition. But there is something puzzling there because it is very hard to find any country where many people would think it is a good idea to impose trade barriers within the country - and in the USA, that’s in fact constitutionally forbidden. To me, this basic observation suggests that a lot of the rhetoric about protectionism is just the voice of special interest groups that would stand to lose from any competition and they know they wouldn’t get away with complaining about domestic competition so their complaints are focused on foreign competition.
As a general rule of thumb, I think it would be a mistake for governments to think they can do better for their average citizen by imposing trade barriers instead of free trade. We can come up with arguments why departing from free trade can come up with some small benefits in special cases when done well and with infinite knowledge about the right way to do it, but in the real world, in most settings, my bet would be that you can maximise some notion of societal welfare with free trade.
When it comes to developing countries, these gains can be especially large, because trade can help smooth out the volatility in the availability of necessity goods, which have substantial effects on welfare levels.
TC: With reference to that backlash against trade, the work by Autor and others has been very prominent as flagging up some of the distributional and adjustment costs to trade - how much does that affect our sense of trade providing some sort of Kaldor-Hicks gains?
A lot of that empirical work was about how an economy responds to a shock of more trade - the idea being that the USA was ticking along in a steady state and suddenly China joins the WTO, exporting a lot more goods. Of course, this played out over 20 years or so, but in the grand scheme of things that’s still relatively sudden.
The evidence was hence about how an economy responds to a shock, rather than necessarily saying something about whether the change was a move in a positive or negative direction (as you would be interested in if you wanted to ascertain the correct level at which we should be trading). Changes in economic life always have costs of adjustment, and you probably could also have found that a sudden reduction in trade would incur costs too.
That work, which has also been replicated and extended in many ways and settings, has been wonderful for demonstrating just how sizeable are the costs of adjustment that are borne by a relative minority of unlucky people. The highlight how much better off we would be if we lived in a society that had more social insurance such that unlucky individuals didn’t have to face the brunt of those costs on their own. It’s not a good thing if we live in an economic world where any change, from technology or policy or anything else, creates a setting where say 90% of people benefit a bit in exchange for 10% taking a significant hit and there is no insurance or social safety net there to smooth that out.
Finally, it is worth recognising that it can be difficult to identify those 90% and measure how they benefited from a shock, whereas finding the 10% is perhaps easier. So these studies probably don’t tell us as much about the benefits of those trade shocks, and so even if you wanted to do some kind of Kaldor-Hicks aggregation of costs and benefits of a trade shock it would not be easy.
TC: Insofar as trade is useful, what do you think the magnitude of the gains are from governments helping develop transport infrastructure and lower barriers to trade?
When it comes to things like improving the quality of your ports or streamlining the customs procedures or building roads to take goods into non-coastal areas, there’s some cost from building that infrastructure. In the examples I’ve seen of those projects, and for example in the case of my research into railroads in India, the estimated benefits always exceed the construction - even on the margin for the case of improving already very well-developed infrastructure systems – so they pass a cost-benefit test.
TC: Finally, I am curious about what you think the frontier of economic research is in international trade - is it heading towards a more theoretical or empirical direction?
Economics as a whole has become incredibly empirically-oriented across the last 30 years - and that has happened in the field of international trade among other fields. That’s primarily because we have more data and computing power, all of which makes empirical tasks easier nowadays. There’s also a limit to the amount of stuff you can say if you were prepared to be purely algebraic about the way an economy works - when I say algebraic, that is to say, to proceed with purely arbitrary values for all economic parameters. So it’s natural that once we’ve seen how it works for some arbitrary values, we should go on and measure those numbers, and a lot of recent work has been trying to do that in one way or another. Obviously, however, there’s still plenty of theory left to figure out – not least because new economic phenomena arise all the time. Finally, a developing area is about the methodological intersection between theory and empirics - looking at how we can use what we’ve learnt from data into the theory we know.
TC: Thank you very much for your time.
It was a pleasure.