Theory is vital to understanding the world. It both helps us make sense of the world around us and make predictions. Theory is like a pair of glasses: a good pair helps you see better. A bad pair makes things worse. And no pair is just useless.
One of the major problems these days is that the trade policy of the Trump Administration is guided by horrifically bad theory (when guided by theory at all). Even worse, they do not even seem to grasp easy-to-verify facts and figures (e.g., Trump’s claim that the US trade deficit with China is $1 trillion. It’s less than $300 billion. Personally, I’d have relatively more faith in central planners if they could get even the basics right).
This post aims to be a corrective lens on some of the nonsense. Of course, to the True Believers, none of this will matter; this post isn’t for them. This post is for those who want to understand trade theory.
International trade does not differ from domestic trade. Whether one is trading with Bret in Boston, Brad in Baton Rouge, or Bobby in Berlin, the same principles hold. Specifically, people think on the margin when making buying and selling decisions. They will buy a good when the marginal cost of making the good exceeds the marginal benefit of making the good. In other words, they will buy when it is cheaper to buy and make when it is cheaper to make.Likewise, they will sell when the marginal cost of making the good is less than the marginal benefit of making the good. Or, they will sell when they can get more for selling the good than it costs to make it.
Ricardo’s simple model, in which comparative advantage determines the patterns of trade, has been shown time and time again to hold, both at the individual level and at the international level. Technical elaborations have come about (e.g., the Hecksher-Ohlin ‘Standard Trade Model,’ Stolper-Samuelson and Factor Price Equalization, the Specific Factors Model, the Product Life Cycle Theory, etc), but we will stick with the simple model, as these elaborations do not change the underlying logic.
Since people are thinking and acting on the margin, we should expect international trade patterns to reflect these margins. In other words, a nation should generally import what it is relatively bad at making (i.e., low productivity goods/services) and export what it is relatively good at making (i.e., high productivity goods/services). Furthermore, since wages are tied to productivity, we should expect the wages in import-competing industries to be relatively low and the wages in export-competing industries to be relatively high. Indeed, that is what we see.
Writing for the Peterson Institute for International Economics, J. Bradford Jensen and Lori G. Keltzer show that the overwhelming majority of jobs “at risk” to trade are in sectors with low productivity and low wages. Conversely, the sectors with the highest productivity and wages are exporters (see figures 4 and 7). These data are a little old (the report is from 2008), but I have been working on updating the figures. The pattern isn’t changing; just the numbers.
Since trade patterns follow a certain logic and are not random, we should not expect job losses from trade to be random, either. Protectionists like to argue implicitly that trade job losses are either random (and thus they cite average wages) or that trade job losses are non-random, but for some reason, firms look to offshore their most productive, least costly business (thus focusing solely on high productivity industries and wages). But job losses would be in low-wage areas, and job gains would be in high-wage areas.
Consequently, any jobs “saved” by tariffs will also be those low-productivity jobs at the expense of high-productivity jobs. Take textiles, for example. Textile manufacturing in the US faces staunch competition from abroad. According to the BLS, textile manufacturing workers earn an average of $17.78/hr. That’s just 54.4% of the national average ($32.66). Conversely, oil & petroleum extraction workers—one of our major exports—earn an average of $28.39/hr. (Note: these data exclude managers and other supervisory workers. These are just non-supervisory.) Tariffs may “save” some low-productivity jobs, but at the cost of higher-productivity jobs. (To head off any objections, it is true that the two industries chosen are the polar opposites of the scale, but the point remains.)
Now, of course, as trade expands, some textile workers may get laid off. What are their alternatives? Are they doomed to live on the public dime the rest of their lives? After all, their skills are no longer needed in the US economy (what economists call “structural unemployment”). The answer is: probably not. Even relatively low-productivity service jobs pay approximately the same as these low-end manufacturing jobs. Food prep workers earn an average of $17.32/hr. Retail sales workers: $17.05/hr. Are these declines in wages for the textile manufacturing worker? Sure. But they are pretty comparable wages. If the worker wasn’t on welfare with textiles, he likely wouldn’t be working as a retail or food prep worker either. And all this assumes the worker takes no steps to re-train. If they do acquire skills for which there is more demand, they can increase their wages.
Life happens on the margins. Thus, adjustments to trade will happen on the margins as well. Good theory helps us see the potential impacts of tariffs along various margins (and to dismiss what impacts are unlikely to occur).
Let me end with a personal story from grad school:
I was taking Robin Hanson’s Law & Econ grad class (ECON 841). For part of the grade, we had to present an original paper. I had come up with what I thought was a very clever model. The math worked out, and it was quite pretty. I presented the paper, and Dr. Hanson said just one thing: “This is interesting, but where’s the economics?’ With that one simple question, he blew up my model. I couldn’t answer. The mathematics were precise. The model was logically perfect. But it explained nothing. There was no theory. It amounted to little more than “what if?” I learned two important lessons that day: 1) if I never wanted to be embarrassed like that again, I’d have to make sure theory is included, and 2) give someone enough assumptions, and they can prove anything they want.
Good theory prevents bad vision. Bad theory, no matter how mathematically precise, no matter how many fancy Greek letters you throw in, will mislead.
READER COMMENTS
Ahmed Fares
May 15 2025 at 4:53pm
On comparative advantage, Oren Cass writes (emphasis in the original except the bolding in the second paragraph which I added):
Comparative Disadvantage
An example in this Tweet by Ben Norton:
More at the Tweet:
Tweet
Jon Murphy
May 15 2025 at 5:55pm
Cass doesn’t understand the first thing about comparative advantage. I don’t read him and neither should you.
robc
May 16 2025 at 2:25pm
I think you can strike through “about comparative advantage.”
I hate to say this because I know how much I hate when people try to tell me things about my fields, but it seems like with two undergraduate classes and 35 years of reading about economics, I know more about it than Cass. Almost assuredly is false, but it seems that way.
Jon Murphy
May 16 2025 at 2:41pm
Not likely false. I’ve read both you and him. It’s not even close who grasps the concepts better.
Richard W Fulmer
May 18 2025 at 11:07am
I think it’s wrong to dismiss the endogeneity argument out of hand. For example, a culture that values education may develop strengths in knowledge-intensive fields such as law or medicine. Similarly, a poor region may have a comparative advantage in cheap labor, though its poverty could stem from either endogenous or exogenous causes.
The claim that comparative advantage can arise from scale — the more you produce, the better you get — is plausible, but scale itself requires an explanation. Without addressing why production concentrated in one place to begin with, the argument risks circular reasoning.
Jon Murphy
May 18 2025 at 11:46am
I don’t dismiss the endoginicity argument out of hand; I dismiss Cass out of hand. Comparative advantage is inherently endogenous. But that’s one of the key things Cass doesn’t understand. He treats it as exogenous and arbitrary.
Ahmed Fares
May 15 2025 at 5:08pm
Joan Robinson writes (emphasis added):
Jorge Morales Meoqui
May 16 2025 at 3:38am
Perhaps you are interested in reading an interpretation of David Ricardo’s theory that is very different from that of Joan Robinson, Oren Cass, and Jon Murphy but closer to Ricardo’s writings.
Spoiler:
Ricardo did not formulate any theory of comparative advantage
His rule for determining the beneficial nature of a particular exchange accounted for any endogenous or exogenous change that might occur.
I leave this link to a preprint of a peer-reviewed paper that I published in the Journal of the History of Economic Thought in 2021.
Jose Pablo
May 15 2025 at 9:38pm
national average ($32.66).
Conversely, oil & petroleum extraction workers—one of our major exports—earn an average of $28.39/hr
Is the national average below the oil & petroleum extraction average?
Am I getting something wrong?
Jon Murphy
May 15 2025 at 10:24pm
Nope. You’re right. The national average is higher
David Seltzer
May 16 2025 at 3:22pm
Conversely, oil & petroleum extraction workers—one of our major exports—earn an average of $28.39/hr. I suspect, the hourly wage of $28.39/hr is below the average of $32.66 because of tariffs. Lerner’s Symmetry Theorem explains this phenomenon. Of course there are other reasons the extraction workers average hourly wage is below the national average.
Jon Murphy
May 16 2025 at 3:31pm
No. One, these data are from 2024 (most recent). Second, the national average is the national average hourly wage of all workers, from the CEO to the janitor. The manufacturing wages here are just the non-supervisory workers (eg, not including CEOs, managers, janitors, etc)
Pete Smoot
May 16 2025 at 9:28am
You write:
I’ve often wondered: if a worker could get training, acquire skills, and increase their wage, why haven’t they already done this? It seems like the proverbial $20 lying on the sidewalk.
Jon Murphy
May 16 2025 at 9:41am
There are costs to doing so. Yes, the $20 may be sitting on the sidewalk, but if it costs $21 to pick it up, it’ll remain unclaimed.
But more precisely, many do retrain. Career transitions, or promotions within a company, are quite common.
I’m addressing a more precise claim, namely that people laid off from jobs due to trade are necessarily going to live on the public dime.
Thomas L Hutcheson
May 16 2025 at 11:23pm
Reductions in tariffs encourage export industries and for others to export making it worthwhile for workers to retrain to adjust.
Unless of course Congress decides to start cutting taxes w/o reducing expenditures (running deficits), drawing in more foreign capital and strengthening the dollar. Then the tariff reduction effect on import substitutes will be reinforced and the encouragement of export dampened.
Thomas L Hutcheson
May 16 2025 at 11:10pm
Nice.
But it’s worth pointing out that tariffs favor the low (marginal) productivity sectors and _tax_ exports, the high productivity sectors.
This helps t explain (is a concrete example of) why tariffs not only do ont increase total output, but actually reduce its value.
Warren Platts
May 17 2025 at 10:04am
Jon, that is NOT what Trump said. Here are his exact words: “Last year, China made one trillion dollars off trade with the United States.” He didn’t mention the trade deficit at all. And your claim that China only made $300 billion off of us because that’s what the bilateral trade deficit is a prime example of “bad theory.”
Imagine a world consisting of three countries: PRC, ROW, and USA. PRC runs a trillion dollar trade surplus; USA runs a trillion dollar trade deficit; ROW’s trade is balanced (NX=0). However, USA’s bilateral trade balance with PRC is only negative $300 billion. So how is it the case that PRC makes $1 trillion off of USA instead of only $300 billion? Easy peasy: PRC runs a $700 billion trade surplus with ROW, but ROW pays PRC with a $700 billion trade surplus with USA. Thus, it is USA that ultimately absorbs PRC’s entire trillion dollar trade surplus. Trump had it exactly right.
And how do we distinguish “bad theory” from “good theory”? Predictions, I guess. Certainly, it is a prediction of free trade theory that opening up free trade should increase the rate of GDP growth. Yet that did not happen in the USA: post-NAFTA real GDP growth rates are 28% lower than pre-NAFTA growth rates (p=0.90). I have yet to see a free trade economist adequately deal with this anomaly.
You accuse protectionists of citing average wages (“thus they cite average wages”), yet it you who cite average wages. Why? You could just as easily cite median wages; thus one is forced to suspect that you cite average wages because average wages hide the fact that the majority of workers make less than the average wage.
Also, you talk a lot about productivity and (merely) assert that “wages are tied to productivity.” But that’s not really the case (cf. the Baumol effect). As Tyler Watts recently pointed out, there are many jobs that pay better than manufacturing wages, yet are far less productive. Eg. government workers are way better paid, yet only 65% as productive; education workers also better paid, yet manufacturing workers are about 3X more productive. Nor is productivity tied to net exports: we have a surplus in education services despite the dismal productivity. This merely shows we are apparently less bad at education than the ROW on the whole. But that’s how comparative advantage works.
Jon Murphy
May 18 2025 at 8:32am
Mr Platts,
There are numerous errors of fact, errors of theory, and contradictions in this comment, but I’ll only address two. You claim
Actually, the evidence is quite strong that NAFTA and trade increases GDP growth. Just about any textbook will cover the evidence, as does Doug Irwin’s Free Trade Under Fire. The CBO, BEA, just about anyone who has done studies finds the same conclusion. You are making the classic “post hoc ergo propter hoc” fallacy. Yes, GDP growth has slowed, but there many, many, many factors that have contributed to that (the rise of regulation, miscounting service, etc). Trade is not one of them. You need to show evidence that all else held equal trade has failed. I mean, even you agree this because you cite that your claim is statistically insignificant (p=.90 is statistically insignificant. P<0.05 is a good rule of thumb to use for statistical significance, although there are other cutoffs you can use. But .90 is way too high).
Second, you write:
You should not rely on maliace when simple facts will do. If you follow the link to the data I cite, you’ll see median wages are also there. And you will see the story does not change. As I said, I used a simple mean because that is the statistic protectionists use. I am showing that by their own metrics their argument fails.
David Taborsky
May 18 2025 at 5:27pm
I am NOT an economist, sadly only an engineer. If I were a country, I would encourage foreign investment that gave me knowledge and ability in a given industry. Then I would make my own products because I had grown more efficient than anyone else. At that point, I would become a free market proponent. Maybe I am naive, but how else does a country that has little domestic demand go from no ability to being the best?
Warren Platts
May 20 2025 at 2:18pm
That’s an apt description of what’s going on in China.
Warren Platts
May 19 2025 at 6:40pm
Sorry, that’s a typo. I was thinking in my mind that there was a 90% chance that the observed pattern was NOT the result of random variation: i.e., that p=0.10; thus I’m sure you will cut me a little slack, Jon, as I’ve noticed a few typos in your own writings over the years!
The question is: Shall we consider p=0.10 significant? Yes, it’s greater than the popular p=0.05 cutoff, but there’s nothing magical about the p=0.05 cutoff. The proper cutoff will depend a lot on the subject matter. If you’re doing a clinical trial on an expensive new drug that potentially has a marginal effect on patient outcomes and you’ve got large sample size, then yes, you’ll want a p=0.05 cutoff. But if you’re a particle physicist who thinks you’ve found a new particle, you better have six or nine sigmas. Since we’re talking economics — and despite all the fancy math, economics is not the most exact science ever — and since in this particular test the sample size isn’t very big, and since there’s a lot of variation, then a p=0.10 should be considered to be reasonable. Moreover, it cannot be denied that a 1 pp drop in real GDP average growth rate is highly economically significant. Therefore, the pattern is certainly worth looking into.
As you say, Jon, it is certainly the case that “give someone enough assumptions, they can prove anything they want.” So if historically, GDP growth is relatively fast under protectionism, then slows down under free trade, it is always possible to concoct an historical narrative along the lines that the fast growth happened not because of protectionism, but in spite of protectionism; and if growth slows down when free trade is initiated, then it’s lucky that free trade was instituted right when it was because the GDP growth rate would be even lower than observed!
On the other hand, what is the theory that says that free trade for USA in the early 21st century is actually causing a headwind to growth? And again, neither I nor my fellow American protectionists are making any general claims about trade. The claims are specific to this particular time & place. So: (1) in a consumption driven economy, spending depends on wages, yet the percentage of the national income that’s gone to wages as opposed to the owners of capital has declined drastically in the free trade era; (2) economic growth in the long run depends on technology more than anything else, yet we’ve been exporting more technical know-how than we’ve been importing in the free trade era — that’s for sure; (3) there’s not just free trade going on, there’s also the trade deficit going on: this represents a net export of USA’s aggregate demand; therefore, the quantity demanded for domestic production must go down.
Again, I have no doubt that it’s the case that in a perfect world, global free trade would maximize global economic growth. But this world, as it is, is far from ideal. What is happening now is that America is being altruistic in the biological-evolutionary sense to the ROW. That is to say, we are making an unwitting expenditure of our resources to help out unrelated peoples. In general, this is not considered a winning evolutionary strategy. That is, free trade, as USA is experiencing it now, is not symbiosis — it is parasitism — on us, because we are the ones supplying the blood..
Comments are closed.