LP #42: Trade Liberalization, Not Protectionism, Promotes Inclusive Growth
Author: Gary Winslett
Back in August, U.S. Trade Representative Katherine Tai argued that the Biden administration’s trade protectionism advances the goal of racial equity. Meanwhile, the U.S. International Trade Commission (ITC), at Tai’s request, was preparing a report on this topic. That report, hereafter referred to as the ‘ITC Report’, which was recently released and which you can find here, appears to support Tai’s claim. Respectfully, I think Tai is wrong and the ITC Report makes a number of critical mistakes on its way to its conclusions.
My aim here is to show that the preponderance of evidence suggests that trade liberalization, not protectionism, advances racial equity. This matters a lot for trade policy. If Tai is right, then the Biden administration’s protectionism, while it may cause other problems, has at least one good justification. If, however, I am right, then the Biden administration’s efforts are actually hurting the cause of equity and inclusive growth.
I want to make it clear here that I think Tai’s motives are good and I think the motives of the ITC Report’s authors and contributors are good, but I think they are making serious analytical mistakes that yield trade policies detrimental to the American public.
Workers and the ITC Report
In August, when USTR Tai was making her claim, her main piece of evidence was data showing that non-white workers have seen especially fast wage growth during the recovery. That’s true (and good!) but there’s no evidence that was in any way attributable to the Biden administration’s protectionist policies. Along these lines, of the 74 studies the ITC Report cites overall, 5 appear to back up this claim. Let’s review those though. The first is a 2019 study by Mary Kate Batistich and Timothy Bond. It found that in the late 1970s and early 1980s, cheaper imports from Japanese firms lowered employment among Black workers, but not White workers, but importantly, the study said that the reason this disparity happened was because those Black workers had been forced to attend segregated schools earlier in life, had had their education throttled by discrimination, and had disproportionately not finished high school. By the ITC’s own admission, Batistich and Bond “find only limited evidence for negative effects of race once education levels are taken into account.” (page 124). So, what’s really going on in this study is that Jim Crow education policies hurt Black workers with trade as an epiphenomenal factor. That’s not an argument for protectionism; that’s an argument for eradicating systemic racism.
The second study, by Eric Gould, is not about trade specifically, it is about manufacturing employment decline. This only makes sense if you think international trade is responsible for manufacturing employment decline. There are a lot of reasons that it’s mostly about automation though, not trade (more on this below). Moreover, as Gould says, and as the ITC Report acknowledges in a footnote, racial gaps “precede the era of Chinese trade by decades” (Gould, 6).
The third, fourth, and fifth studies that the ITC Report cites in this vein relates to the China Shock and that trade versus automation question. In a separate roundtable this report was based on, when Williams Spriggs (of the other work just above) says that Black workers get hit hardest by labor market shocks, Timothy Bond (also cited above) agrees. They are right. Black workers do get hit hardest, on average, by labor market shocks. The thing is though, again, this is mostly an automation story, not a trade story.
The China Shock and Its Limitations
Tai’s heart is in the right place (and the same goes for those who agree with her); they want there to be good jobs for high-school educated people in America. I do too! That’s why I think it’s absolutely vital to get the diagnosis right of what is and is not ailing parts of American labor market.
Few research veins in trade politics have gained as much salience as the ‘China Shock.’ It is well-executed, on an important topic, and it arrived right when public anxieties about the decline of manufacturing and the rise of China were beginning to fuse. Their argument went as follows. China’s economic policies were calamitous under Mao. That meant that once China adopted better policies there was a lot of room to grow, especially given its population size. China’s comparative advantage was in labor-intensive manufacturing. China’s manufacturing growth thus constituted a large global supply shock for labor-intensive manufacturing and a large demand shock for raw materials. Different industries tend to cluster in different places; this is true in the United States as it is anywhere else. Because of that, different places in the United States experienced different levels of vulnerability to that supply shock in labor-intensive manufactured goods. As some of Autor, Dorn, and Hanson’s earlier work shows, labor intensive manufacturing is particularly trade exposed and particularly concentrated in a cluster of mostly southern states. The places that were more exposed to this supply shock saw larger declines in manufacturing employment and wages. These negative effects had big local spillovers, i.e. when that labor-intensive manufacturing declined, it undermined demand for other goods and services in that area, corroded tax bases, and contributed to a whole range of other social ills in those areas.
Not only that, but contrary to economists’ conventional wisdom, people did not tend to move; they stayed put and suffered rather than leaving for greener pastures. In other words, the adjustment costs to this shock were deeper and more scarring than was typically appreciated. The negative impacts were felt most strongly by the least-educated, lowest paid workers; those with higher incomes (implying more education and skills) were better able to transition than those with less income, education, and skills. When taken at face value and especially when weaponized by populist politicians, this story painted a damning picture of trade’s impact on middle America.
There are however a number of important caveats to this. The first is that the job losses are probably an overestimation. Autor himself has acknowledged that their estimates are the upper-bound. Other studies, such as this one, have found much smaller effects. Some studies, such as this one, even find that liberalized trade with China likely had a positive impact on wages and employment. Additionally, in terms of mobility, Autor, Dorn, and Hanson’s model assumes literally zero geographic mobility. Even if they’re correct that prior research overestimated mobility, assuming zero mobility is an inaccurate representation of reality and inflates the estimate of jobs lost. The second caveat is that, even if everything about the China Shock is correct, it's inaccurate to attribute Chinese imports to capital-intensive manufacturing decline. As ADH point out, some industries like apparel, textiles, and leather are particularly labor-intensive. Even in areas that are manufacturing oriented, there is a lot of variation in terms of industry exposure to imports; Tennessee was a lot more exposed to trade with China because its manufacturing was disproportionately in furniture while Alabama was less exposed because its manufacturing was in heavy industry.
Michael Hicks and Srikant Devaraj’s research highlights this same differential between industries and suggests that trade only accounted for about 13 percent of manufacturing job losses between 2000 and 2010.
Of the 18 industries they look at, trade was responsible for more jobs lost than productivity increases in none of them and was responsible for more than a quarter of job losses in only two of them (apparel and furniture). As they say in their paper, “had we kept 2000 levels of productivity and applied them to 2010-levels of production, we would have required 20.9 million manufacturing workers. Instead, we employed only 12.1 million.” So even if every word of the China Shock literature is true, the bulk of manufacturing jobs that went away did so because we got better at making more stuff with fewer people, not trade shocks, and that statement becomes even more true if one excludes apparel and furniture. Geographically speaking then, even if it makes sense to attribute the decline of furniture and textile towns in the South to the China Shock, it makes no sense to blame the China Shock for the decline of manufacturing jobs in capital-intensive industry.
Third, the China Shock arguments ignore the benefits of imports from China. According to one Federal Reserve study, for every lost job due to the China Shock, there were approximately $400,000 in consumer benefits. From 2000 to 2006, the peak years of the China Shock, the price of manufactured goods fell by 7 percent. This was not only good for consumers. Because over a third of Chinese imports were intermediate goods, it was good for producers too. It is worth pointing out here that the ITC Report cites the Autor, Dorn, and Hanson research on the China Shock but then does not cite most of this research that pushes back on it. Again, I do not want to impugn anyone’s motives. These might be honest mistakes. But even if they are honest mistakes, they are quite convenient to the protectionist narrative about trade.
Additionally, the United States does not just import from China, it also exports goods and services to China as well. Many of the producers of those goods and services would not have had access to the Chinese market were it not for China’s integration with the global economy. Moreover, discussions around trade shocks often ignore the costs that accompanied the United States’ reactions to it. When President Trump raised tariffs on China and the EU, China and the EU predictably retaliated. American farmers paid the price; agricultural exports targeted by those retaliation were $8 billion lower in 2018 than they had been in 2017, a 27 percent drop. The damage was wider than just farmers. Just for 2020, President Trump’s trade war with China was estimated to cost the average American family over $1200 in higher prices and lost productivity. According to the IMF, the trade war lowered global GDP by 0.8 percent. Antidumping measures on China hurt industries that imported those targeted products (and thus workers) as they raise the cost of production. In other words, they negatively impact employment and wages and don’t actually help the protected industries all that much.
Automation and Trade
So then what is ailing manufacturing work? It is not actually output. As Bureau of Labor Statistics show, the United States manufactures more now than it ever has. What has gone down is manufacturing employment.
It is automation that is causing that. Look inside car factories today. Yes, there are still some workers there, but the factories do not have nearly as many people in them as they once did, and that’s because of automation. In the auto sector in 2015, once all of the costs are factored in, a spot welder earned about $25 an hour whereas it only cost about $8 an hour for a robot. That kind of automation is now coming to industries like electric equipment and even to furniture which (as the Hicks/Deveraj work above shows) was long a very difficult industry to automate.
There’s a lot of economic research that backs up this automation thesis. See this paper on Acemoglu and Restrepo and this paper by Autor and Salomons on automation’s displacement of labor. There’s also this paper by Graetz and Michaels that shows that automation improves productivity and accelerates growth but finds some evidence that it reduces demand for low and medium-skilled work. Here’s a study showing that automation negatively impacts manufacturing employment but more than makes up for that by increasing productivity and boosting employment outside of manufacturing. This isn’t just an American phenomenon. Dauth, Findeisen, Südekum, and Wößner find a similar pattern in Germany. Notably, as is the case in the United States, the observed manufacturing job losses there are more than made up for in aggregate. Another piece of evidence that suggests that manufacturing employment is falling because of automation rather than trade is that it started dropping in the 1970s, i.e. well before the ‘China Shock’ and that it has fallen in other countries including Germany and Japan.
Again, it is important to note that these labor-displacing effects in manufacturing are typically far outweighed by the overall increase in labor that automation brings. In contexts as varied as Canada, Spain, Finland, and Japan, more automation has increased overall employment. A meta-analysis literature review found the same. It is worth saying this very clearly: consistent with the historical trends, automation today creates more jobs than it kills. It may be the case that automation negatively affects particular segments of the labor market, but the narrative that automation is going to eat all the jobs or even the milder claim that it will lower overall employment do not seem to be rooted in fact. That is good news!
The bad news though is that particular parts of the labor market really are being hit hard by automation. The same meta-analysis that found an overall employment also said “low-skill, production, and manufacturing workers have been adversely affected by technological change, and effective up- and reskilling strategies should remain at the forefront of policy making along with targeted social support systems.” Moreover, the speed of technological progress may present significant challenges for certain workers, particular given how much change is now happening over the course of a single career. Technologies like AI and machine learning will fundamentally alter how many jobs get done even if they do not eliminate them.
This was also the case for NAFTA specifically. J. Bradford Delong has an excellent explanatory piece on Vox about this. As he says “Did NAFTA drive the fall in the manufacturing employment share? No…. The trend preceded NAFTA, and it would have continued with or without NAFTA.” One thing that DeLong correctly emphasizes is that while trade has bigger benefits than costs, for those who lose out, it is much, much better for them if the overall labor market is tight because that means that they can more easily find a new job. It’s that mobility that’s the key.
This automation trend is especially challenging for Black workers. Vittoria Dicandia (2021) finds that the reason the racial wage gap remains stubbornly high is that Black workers have been disproportionately likely to work in routine-based worked that can be automated and so skill-biased technological change has disproportionately affected Black workers. This Dicandia paper is backed up by a 2021 McKinsey report that also finds that Black workers disproportionately work in automation-vulnerable occupations. At the same time, a new research paper specifically on the China Shock and racial equity finds “no evidence that minority workers are relatively more harmed [by competition with Chinese imports] than white workers in terms of their manufacturing employment.”
This is obviously a very different explanation than the China Shock argument described above. If the technology explanation is correct, then the protectionism advocated for by the China Shock-oriented explanations will hurt Black and Hispanic citizens (and White citizens for that matters) as consumers without actually helping them as workers. What a policy tragedy that would be. We should want to help those workers who have been displaced out of manufacturing, but understanding what caused that displacement is key to getting the policy intervention right. If it was automation rather than trade, then the effective policy response is to help those workers gain new skills, not protectionism.
The reason that I have gone through this lengthy discussion of the China Shock and the trade versus automation debate is that the ITC Report and USTR Tai both seem to put great stock in the argument that trade has gutted manufacturing employment and that this has been particular bad for Black and Hispanic workers. Indeed, that is the very heart of their argument. What I hope the evidence provided here has shown is that they’ve got the wrong guy. It was automation, not trade, that primarily undercut manufacturing employment and especially for Black workers. As with the omissions related to the China Shock, the ITC Report does not include a lot of this research that is highly relevant to the question they were asking. I do not want to be rude and I acknowledge that everyone makes mistakes but…..if you’re the International Trade Commission, it is your job to get these things right and have your reports accurately reflect the current state of research on international trade. They really should have included this material in their report and it is highly problematic that they did not.
Other Problems in the ITC Report
A lot of what is in the ITC Report is based on a series of roundtables that the ITC hosted. Again, I do not want to impugn anyone’s motives but, a review of the roster of attendees shows that the participants in these roundtables heavily leaned toward individuals and organizations with protectionist viewpoints and there were almost no participants who might have pushed back on that. It’s a bit like hosting a roundtable on the question of “are the New York Yankees bad?” but doing it with Red Sox season ticket holders in a bar outside Fenway Park. You know, or at least you should know, that you’re going to get a certain kind of answer. Again, I am not saying that anyone did anything nefarious or intentionally manipulative, but given who attended these roundtables, it is not at all surprising that their statements trend in a particular direction.
Relatedly, some of the ways in which things have been framed in the report seem tilted to favor protectionist narratives. The McKinsey Report that I mentioned above is used on page 122 in such a way that a reader who isn’t reading very carefully or doesn’t know this academic literature would think it was supporting an anti-trade viewpoint even though that McKinsey report isn’t about trade at all. Another example is in how Michael Waugh’s paper gets used. His paper shows, quite convincingly, that low-income workers were particularly harmed by the drop in exports to China caused by the U.S.-China trade war and that this meaningfully reduced their disposable incomes. In other words, it was saying very clearly, that trade wars hurt low-income Americans. When the ITC report is reviewing his paper on page 176 it gets framed as a critique of trade liberalization, which is the exact opposite conclusion that should be drawn from those findings.
Given the attendees at those roundtables and given this framing pattern, it is perhaps not terribly surprising that consumers interest got short shrift. But, that is a major problem because protectionism is especially terrible for Black and Latino consumers. Tariffs are a highly regressive tax. In part this is because they’re especially high in food and clothing, items that are both highly traded and comprise a higher share of low-income consumers basket of purchases. Additionally, even within product categories, tariffs are higher for mass-market goods than luxury goods.
As the report that table by the Progressive Policy Institute's Edward Gresser is from also points out, those tariffs aren’t even effective at promoting greater employment. So, we’re imposing higher costs on families without even getting any jobs in return for that. It is literally a lose-lose trade policy.
Meanwhile, our long-standing tariffs on Canadian softwood lumber as well as other building materials make it more expensive to build new housing. That is extra bad for Black Americans who need new housing now to build wealth seeing as they were historically locked out of that by redlining and other forms of institutional racism.
So let’s just stop and take note of the fact that protectionism makes it harder for Black and Hispanic citizens to put food on the table, buy their kids back-to-school clothes, and build wealth via housing. That is not equity-informed trade policy. It just isn’t. And, by the way, in this ITC report that is over 250 pages long, the word ‘protectionism’ is not mentioned a single time (hat tip to Simon Lester for catching that). That, by itself, is a strong clue that this report paid insufficient attention consumers’ interest and to the drawbacks of trade barriers. To the extent that it covers those interests at all, that is almost exclusively confined to written submissions that are very briefly reviewed on page 133 and to Edward Gresser’s contributions to Panel 3. That’s it. Given the extent to which ordinary Americans are struggling with high inflation, it surprising how much the report seems to downplay consumers’ interest. The Trump/Biden Section 301 tariffs have now cost American consumers $129 billion since the trade war began, and as noted above, that obviously falls harder on low income Americans, who are disproportionately Black and Latino. One would think an ITC report on trade and equity would discuss this is in considerable detail.
Defenders of the ITC Report might respond that the report’s focus was on workers, as its title is “Distributional Effects of Trade and Trade Policy on U.S. Workers” and so consumers’ interest need not be considered. Such a response would be ignoring the fact that every worker is a consumer. Workers do not bury their wages in the ground; they use them to purchase goods and services. Furthermore, most workers in the United States work in non-tradeable services and non-tradeable goods meaning that the primary way that they are impacted by trade is via being a consumer. Thus, to ignore consumers’ interests in trade is to, de facto, ignore most workers’ interest in trade.
It is also worth noting that Black and Hispanic voters consistently express strong support for free trade. 71% of non-white voters see trade as more an opportunity than a threat (that’s 15% higher than whites), and they support NAFTA at higher rates than whites as well. To put it bluntly, Black and Hispanic citizens are pretty clearly saying they want more trade, but the forces of protectionism are marginalizing their voices.
Another trade policy item relevant to equity that the ITC report does not mention at all is the Jones Act. Report after report after report shows that the Jones Act economically strangles Puerto Ricans. Puerto Ricans are Hispanic Americans and on average have lower incomes than other Americans. As Colin Grabow explains, there is a clearly progressive case for reforming the Jones Act. How did this not get discussed at all in a report on trade and equity?
If we look internationally, some of the people most hurt by trade barriers are farmers in the Global South, particularly sub-Saharan Africa. Trade liberalization would help them a lot. Getting the European Union, the United States, and Japan to lower their agricultural subsidies would be immensely helpful for equity globally. Admittedly, considering international interests is not in the ITC’s remit, but if one is going to claim, as USTR Tai does, that this administration’s trade policies are guided by equity considerations, it seems odd to not consider the equity-related impacts that U.S. trade policy has on the Global South. Both parties have reason to not be protectionist here. Both progressives and free-market conservatives should consider market-impeding policies that keep people in the developing countries poor to be antithetical to their values.
To review:
1) International trade is not the primary driver of reduced manufacturing employment. Automation is. So, the main way in which the ITC Report suggests that trade has harmed Black and Hispanic workers is not actually what’s going on.
2) The ITC report is missing a lot of the trade and technology scholarship that would have shown #1.
3) The roundtables that the ITC Report held were very one-sided in terms of participants.
4) Likewise, the framings used around some of the research cited was quite one-sided.
5) The report significantly downplays consumers’ interests and hugely understates the damage that protectionism does to Black and Hispanic consumers.
6) As with the automation versus trade discussion, the report is missing a lot of the evidence that would show these consumer harms.
7) The report is also missing other relevant factors like the harm the Jones Act does to Puerto Rico, Black and Hispanic public opinion on trade, and the implications that U.S. trade policy has for economic growth and poverty alleviation in the Global South.
8) We should want to help workers displaced from manufacturing, but the most effective way to do that is not protectionism, it is helping them gain new skills in occupations that are less vulnerable to automation.
Finally, while I appreciate the focus on equity, I think it’s important to say here that ‘inclusive growth’ is a better way to talk about this than equity. Equity sounds zero-sum. For one person or group to get more, others have to get less. The beauty of international trade is that it is positive-sum. We can all rise together: Black, White, Hispanic, Asian-American, and Native Americans. We are all Americans and we are all in this together. When we create policies that help undo some of the discrimination of the past, that is not taking anything away from anyone, but sometimes that is how the language around equity can sound to some ears. Inclusive growth is probably a clearer way to communicate that positive-sum aspect and so is, to my mind, a better political framing. How do we have robust economic growth and also make sure that everyone has a chance to benefit from that growth? That is what this is about and trade liberalization, not protectionism, is the way to promote that inclusive growth.