Donald Trump has repeatedly emphasized the threat from international trade, especially from China. But did Chinese trade help raise the salience of his concerns or even help elect him? Trade may matter even if views on trade don’t drive the public, because trade shocks also affect citizens’ cultural and racial views. James Bisbee finds that citizens living close to businesses affected by Chinese import competition developed more negative attitudes about trade, immigration, and US global leadership. Francesco Ruggieri finds that Chinese import competition did not change attitudes on trade, but did consistently help Republican candidates since 2008 by increasing negative views of racial and religious minorities.
Grossmann: This week on Political Research Digest, did Chinese trade competition increase nativism and elect Trump? For the Niskanen Center, I’m Matt Grossmann. Donald Trump has repeatedly emphasized the threat from international trade, especially China, but did Chinese trade help raise the salience of his concerns or even help elect him? As trade conflict with China continues, I review new research on the local effect of Chinese competition on American public opinion and voting. Trade may matter, even if views on trade don’t drive the public, because trade shocks also affect citizens’ cultural and racial views.
I talk to James Bisbee of New York University about his paper, “What is Out Your Back Door: How Policy Preferences Respond to Local Trade Shocks.” He finds that citizens living close to businesses affected by Chinese import competition developed more negative attitudes about U.S. trade, immigration and U.S. global leadership. I also talked to Francesco Ruggieri of the University of Chicago about his paper, co-authored with Andrea Cerrato and Federico Maria Ferrara, “Why Does Import Competition Favor Republicans?” They find that Chinese import competition did not change attitudes on trade, but did consistently help Republican candidates by increasing negative views of racial and religious minorities.
I encountered both papers at the International Political Economy Society conference, a sign that international relations scholars are increasingly interested in U.S. domestic politics, mostly since Trump’s election. Bisbee says Trump’s election has generated a lot more interest, but his project was stimulated by a longer term difference between economic models and qualitative local knowledge of trade’s impact.
Bisbee: That’s what really got me interested in it, was this sort of disconnect between what I was being taught and what I was learning from, you know, sort of qualitative research. Trump has sort of weirdly been good for my academic career. I used to go to conferences and present my work back in the early days of my PhD, and half of the presentation would be trying to convince the audience that this was a meaningful question. Then Trump comes along and, you know, I don’t even have to spend a slide on that, and all of a sudden everyone’s like, oh yes, obviously this is a big deal, but the route to it was seven years ago where I noticed this sort of disconnect between what theory told us and what the facts on the ground looked like.
Grossmann: Ruggieri says their paper was directly stimulated by Trump, and the similarities they saw between populist uprisings in Europe and the U.S.
Ruggieri: The interest sparked during the 2016 election night, and two of us had been in the United States only for a few months, so we’ll never forget that night, but how our interest evolved was really by looking at the striking similarities between Trump’s victory in the United States and the reasons why British voters voted to leave the European Union and how campaigns evolved in France, Germany, and our country, Italy. These common patterns, both on the supply sides, so looking at elite rhetoric, and also on the demand side, by comparing the mid-west with certain regions of Europe and looking at the European literature made us want to investigate more and more the mechanism linking trade integration with support for right wing parties and increasing political polarization.
Grossmann: Bisbee and Ruggieri both see the field taking a keener interest in domestic politics, since traditional international economic models of trade did not seem to fully explain the politics.
Bisbee: The story I have of how I arrived and how I got interested in this topic was recognizing sort of the failure of traditional economic models to appropriately account for how painful trade shocks can be. I think perhaps that could explain sort of the IPE discipline’s shift towards these more micro level analyses.
Ruggieri: Overall, this conference has shown that, you know, the distinction between comparative political economy and international political economy has become more and more blurred in the past few years, and this is probably due to the fact that the rise of parties and candidates campaigning heavily against trade and globalization has stimulated scholarly interest in the domestic political effects of international economic phenomena such as trade integration. On one hand we can say that comparative political economists have become more acquainted with factoring international trade in their analysis of domestic political environments.
On the other hand, international political economists have increasingly paid attention to the domestic sources and consequences of inter-state economic relations. The value of bringing ID to the field is a sound understanding of international economic theory, which is certainly the baseline theory to interpret the effects of international trade and finance on domestic politics, but as we actually show in the paper, standard open econ politics views may sometimes be misleading because people may not be fully aware of how trade materially affects their material welfare. It’s important, we believe, to bring ID into the debate of domestic politics.
Grossmann: Both papers help to reconcile the public idea that trade competition from China helped Trump, with the scholarly idea that nativist views mattered more than economic attitudes.
Bisbee: I find basically that people who live in areas or nearby where jobs are lost due to free trade become different in their policy preferences, and different in three specific ways. I show that they become more likely to think free trade is bad for the United States, I show that they’re less likely to think that immigrants benefit the country, and finally I show that these people are less likely to think the United States should be a global leader. These are sort of intuitive findings, but what’s interesting, I think, about them is that these move without movement in, you know, opinions on global warming or abortion or gay marriage, which makes me think it’s sort of a very specific what I’m calling a nativist response to trade shocks.
Ruggieri: Our paper focuses on a recent chain of events, so trade integration, especially since China acts as the World Trade Organization in 2001 has had really disruptive effects on local labor markets, especially those that specialized in textiles or other labor intensive manufacturing industries. This happened both in the United States and in Europe. In the meantime, a growing strand of belligerent political economy has shown that global competition with these Chinese exports triggered sizeable political effects, both in Europe and the U.S., for example, through increased political polarization and support for right wing parties, and so our paper really tries to answer the question what is the link between economic insecurity triggered by trade integration and support for right wing parties, especially Republicans in the United States?
On the one hand, we show that people residing in areas that were affected more severely, directly or indirectly, by trade integration were more likely to express more negative attitudes towards immigrants and racial, ethnic and religious minorities. On the other hand, we observe a change in the way Republican presidential candidates talk about trade. Republican candidates were very consistent in campaigning on anti-immigration, especially legal immigration stances, as well as pointing to issues of criminal justice when they talked about religious and ethnic minorities. At the same time, there was a huge change in the way they spoke about trade. They moved from full support of free trade with McCain in 2008 to protectionism with Trump.
We conclude that trade seems to have little leverage in explaining why regions that were exposed to inter competition consistently supported Republican presidential candidates. We actually combined these two elements of demand and supply to show that support for Republicans in [inaudible 00:08:29] exposed regions is mediated more by a cultural backlash phenomenon, so attitudes towards immigrants and minorities, rather than by the fact that people changed their opinions about limiting foreign imports or international trade agreements.
Grossmann: Bisbee says this matches public assumptions, but shows that there are larger groups of politically influential losers from trade.
Bisbee: For a non-academic audience, this is very intuitive. I mean certainly it sort of checks out with our gut instincts, I think. However, I think the scholarly research, sort of traditionally, has sort of thought free trade or opinions on trade are not really important. Trade hasn’t been thought of as a salient issue for at least recent research on American public opinion. If you look back over the last 30 years or so, a lot of the research on trade and on opinions on trade was very focused on looking at whether what industry you worked in or what occupation you held predicted your opinions on trade. Sort of this research was motivated by basically trying to test these economic models that say, you know, in the United States, if you work in manufacturing, you are “a loser” under free trade, right? If you are low skilled in the United States, you are “a loser” under free trade.
What I think my research contributes, I mean certainly my findings at the very most basic level support previous research in that they find people who are adversely affected by trade hold more negative views of trade. However, what I’ve done is sort of expand or broaden how we define who these losers are. Unlike previous research which simply focuses on only the individual and says what is your occupation, or where do you work, I’ve sort of expanded that definition to say it doesn’t have to be your occupation or where you work. It can also be where you live, right?
My contribution, I guess I would say, is expanding the definition of who we think of when we think of trade’s losers and I think this challenges sort of the conventional wisdom because under the conventional wisdom, you’ve got, you know, manufacturing workers in the United States, maybe some farmers, who we think of as trade’s losers, but they’re a very small minority of the population. What I find is that there is a much larger latent political coalition who is negatively affected by trade by virtue of living in areas where firms shut down, where there’s increasing unemployment, in some cases increasing crime.
Grossmann: Ruggieri says it shows trade can matter even if people don’t specifically react to it.
Ruggieri: Trade long has been considered a low salience issue, so it has played a role in electoral politics since a few years ago. Recently the effects of trade on electoral politics have been devoted much more attention. We actually build on studies that assess whether Chinese inter competition affect political outcomes, and we confirm the findings of previous papers that, in the United States, competition with Chinese imports systematically supports Republicans.
We challenged the conventional wisdom that this is due to demand for protection. We believe there is a different channel, a more direct one, culture backlash channel, which took place between 2000 and 2016, especially, after China’s accession to the World Trade Organization, in which people systematically increased their support for Republicans because Republicans campaigned more heavily on stances that resonated better in these communities that experienced economic insecurity due to trade integration. That’s how we differ from previous studies.
Grossmann: To dig into the papers, we need to review their local measures of the extent to which an area was negatively affected by trade. Bisbee explains the two measures he combines, one based on applications for compensation for trade-related losses, and the other classic measure used in both papers, based on how Chinese industry level capacity differently competed with industries in each U.S. area.
Bisbee: The Trade Adjustment Assistance Program is this federal program that was started back in the ’60s, and it was very explicitly designed to compensate trade’s losers, right? The way it works is groups of workers file a petition to the Department of Labor asking for additional support in the event that they’ve either lost their jobs or lost some wages. These applications are then investigated by someone from the Department of Labor and then either certified or denied, depending on whether there’s evidence to support the claim that these people’s jobs were indeed lost due to trade competition of some type.
There are sort of two really nice attributes to these data. The first is that it’s actually quite hard to determine whether or not someone has lost their job due to free trade or lost their job for some other reason. Even if you’re just only interested in sort of a labor market story, you still need to make some assumptions about is it industry or is it occupation or is it skill level? How exactly can we create a measure of “trade-related job loss”? This TAA data is really useful for me because it represents, at least in the minds of the workers who have applied for this support, in their mind they think they lost their jobs due to free trade. If I’m interested in sort of making sure that your average Joe on the ground is correctly, or maybe even incorrectly, but as long as they think free trade has been bad to them, that represents a kind of ground truth measure. It’s pretty difficult to get at using other methods.
The second benefit is that it’s very rich data. Each row in this dataset is a petition that contains information on how many people were laid off, as well as the address of the specific site where they were laid off. It’s not, for bigger firms, if someone was laid off at a Starbucks, it’s not the address of the Starbucks headquarters in Seattle, it’s actually the address of the specific Starbucks shop somewhere in the United States. What that allows me to do is get really rich geographic information. I can assign each application a latitude and longitude, so I know exactly where it is.
However, this is also a problematic measure in many ways. The TAA, this program, is a political program, so what that means is you’ve got members of Congress, some of whom like the program and will advocate to their constituents to take advantage of it, and there are others who won’t. That’s a potentially concerning thing if I’m trying to say this is my measure of trade shocks. It might mean that in certain parts of the country where you’ve got members of Congress who don’t promote the program, I might be missing some data there that could confound sort of the conclusions I draw. It’s really rich, it’s geographically dense and it gives me a ground truth insight into the minds of the people who apply, but there is these potential concerns with it, so I don’t trust it entirely to extract a causal claim if I just use it by itself.
What I do is I supplement my analysis with an instrumental variable that is based off of some work by David Autor, David Dorn and Gordon Hanson. These guys wrote a pretty influential paper back in 2013 that created a new type of instrument, and instrumental variable where they assign Chinese imports to counties in the United States using a pretty painstaking and sophisticated way of connecting, say, a bicycle to the particular detailed industry that, in the United States, makes those bicycles, and then says in these counties, this industry is relatively prevalent. You can say for workers in these counties, they are affected more or less by these Chinese imports.
This is what I mean when I was saying earlier it’s hard to actually come up with a measure of trade shocks because in doing this, they’re relying on the assumption that industry is the correct way to separate trade’s winners and losers. This instrument arguably buys me a little more causal identification insofar as it relies on the assumption that changes in Chinese productivity, so over the last 30 years China has grown more productive, it’s sort of unlocked its labor potential, you’ve had massive rural to urban migration. That growth in Chinese productivity can’t impact someone’s opinion on trade or immigrants or U.S. global leadership in the United States except through its impact on U.S. jobs via import competition.
When I combine these two measures together, I get, on the one hand, this really rich geographic information, which allows me to say, okay, I’ve got a survey respondent living within 100 miles of a firm, and another survey respondent who is the same race, the same age, the same educational level, but living 500 miles from a firm. I think I can compare them, but at the same time, I’m also getting more causal purchase. I can say something more causal by using only the variation in these measures that comes from the growth in Chinese productivity.
Grossmann: Ruggieri, Cerrato, and Maria Ferrara also used the Chinese import competition measure, which they say provides causal estimates of direct and indirect influences from trade, but they aggregate it to the district level.
Ruggieri: The main benefit of this index, an approach developed by Dorn, Autor and Hanson, is that it captures both the direct and indirect effects of Chinese labor competition and trade integration more generally. There are really two effects of Chinese labor competition on local labor markets. On the one hand, a direct effect on manufacturing, which declined substantially in those regions that experienced lay offs and competition with foreign imports in general. On the other hand, there is an indirect effect on non-tradable services, and this effect is not through employment, but through compressed wages. We believe that this measure captures the broad consequences of trade integration.
On the other hand, the shortfall may be due to the fact that it masks some heterogeneity within commuting zones, because this index requires some aggregation, which is not ideal, and so while trade-related lay offs may be well localized within commuting zones, this index aggregates a bit too much. This is supported by the fact that anecdotal evidence points to the fact that small and medium sized towns and communities were more severely affected by trade shocks, especially when a large fraction of local employment depended on a firm that shut down, and so our index probably misses something of this, and aggregates a bit too much.
We know that the estimates we have are causal because we believe and it has been shown several times that the two key assumptions of an instrumental variable approach are satisfied. On the one hand, this unprecedented expansion of Chinese manufacturing capacity led to a surge and a simultaneous surge in Chinese exports towards a large number of advanced economies. The growth in Chinese exports towards the United States are highly correlated with the growth in Chinese exports towards many advanced economies.
On the other hand, as far as homogeneity is concerned, it seems implausible to assume that this unprecedented expansion in Chinese manufacturing capacity be directly related to local voting behavior. It’s likely that the channel through which the increase in Chinese inter competition affects voting behavior is through local labor market conditions, such as the one captured by our main independent variable.
Grossmann: Bisbee finds effects on support for trade, immigration and U.S. leadership that are not huge, but not much smaller than the effects of party and education.
Bisbee: They’re significant. They’re not substantively large, so I’m definitely not trying to say this is an epiphanal or enormous effect I’m finding. I mean the more conservative specification, you’re between two and five percentage points more likely to believe that free trade is bad if you’re exposed to trade-related lay offs, so that’s not a huge effect, however to put it into context, it’s commensurate to, it’s about two thirds the size of the coefficient on education, so if you compare college educated to non-college educated respondents, college respondents are more likely to think that free trade is good relative to non-college respondents. It’s also about two thirds the size of the difference between Republicans and Democrats. The U.S. would be seven to 15 percentage points less protectionist than it is with the trade shocks that I use.
Grossmann: He finds larger effects for immigration opinions in areas with larger immigrant increases and larger trade effects when media coverage of trade was higher.
Bisbee: Areas of the United States that back in 1990 were in the bottom quartile of the distribution of the foreign born population, but have grown rapidly over the last 20 years, those are the areas where the xenophobic response is strongest, which suggests effectively immigrants are salient on the minds of people who live in these areas, and they are more sensitive to the potential associated threat when they also see trade-related lay offs. That’s the immigrant story.
Bisbee: I also do a little bit of the same thing looking at national media coverage of trade. During periods where the national media is covering trade more intensely, these protectionist responses are also stronger. There’s two pieces of evidence there, to me suggesting part of this story is a salience story.
Grossmann: Bisbee also finds an effect on voting for Trump. Though not huge, an extreme change in trade competition would have been enough to help Clinton win three more states in the election.
Bisbee: We find a significant positive relationship. It’s only roughly a two percentage point shift toward Trump relative to their support for Romney, but it is non-trivial. An important thing to keep in mind here is that these are coefficients that, you know, I estimate them using the entire data, but when you actually again, so I do another counter factual to what I say, what would have the 2016 election looked like had there been no trade-related lay offs? Again, this sort of very extreme though experiment. When I do that, I show, what is it, that Pennsylvania, Iowa and Wisconsin would have flipped for Clinton according to the marginal effect that trade had on shifting support for Trump.
There’s two things I want to be sort of clear about here. One, I’m not saying Trump was elected due to trade. The marginal effect that I document is relatively small, albeit significant, and the counterfactual in which Clinton would have won is an extreme counterfactual that would require there to be no trade shocks over the preceding four years.
Grossmann: Ruggieri looked at three elections, 2008. ’12 and ’16. They first showed that Republicans had been consistently more nativist in their rhetoric, whereas only Trump had embraced protectionism.
Ruggieri: Our text analysis shows that Republican and Democratic presidential candidates were much more consistent than one may think in terms of rhetoric regarding issues of criminal justice, ethnic minorities and immigration. Of course we see from our main findings that there is increasing polarization on these topics, but the relative stances of Republicans and Democrats are not that different between 2008 and 2016. Of course the change in trade policy platforms was stark between 2008 and 2016, and McCain’s campaign was ideally in favor of free trade, Romney was less protectionist than Obama, and McCain as well. Trump changed everything. This may be kind of difficult to digest as of now, but there is plenty of anecdotal evidence that points to the fact that Obama was in fact slightly more protectionist than both Republican candidates he campaigned against in 2008, 2012. There are several clips on YouTube that confirm this rhetoric against jobs being chopped and shipped overseas.
Grossmann: They also found clear but not huge effects on immigration and minority group attitudes.
Ruggieri: Our analysis, first of all, confirms that the traditional demographic determinants of voting behavior were also strong predictors of individual attitudes towards minorities and immigrants. Race, ethnicity, education are all there. They are still strong and we do not worry about that. At the same time, there is an additional effect due to competition with Chinese imports, and this effect is quite sizeable for immigrants [inaudible 00:27:04] to a mean around 3%, and also on religious minorities, especially Muslims seem to be targeted quite a lot in the order of 2% to 3% compared to the mean score.
We use a dependent variable which is called feeling thermometer, which is a cardinal measure which is not easy to interpret, but it’s an idea that gives a sense of how people appreciate certain categories. Let’s say that a one standard deviation increase in penetration, so if a person were to move from a certain area to an area with a one standard deviation higher degree of inter competition would be 2% less likely to have a high appreciation for Hispanics, and around 3% more likely to oppose illegal immigration and believe in the negative effects of immigration.
Our analysis confirms that, you know, the traditional determinants of voting behavior, race, ethnicity, education, are also strong predictors of individual attitudes towards minorities, religious out-groups and immigrants. Then we use a dependent variable which is a cardinal feeling thermometer, which is not very easy to interpret, but given its meaning what we can say is that if a person were to move from a certain commuting zone to a commuting zone which has a one standard deviation higher index of penetration, then that he or she would be close to 2% less likely to support immigration into the United States, and the effect is even more sizeable in terms of opinions against Muslims or in terms of favoring religious in-groups, such as Christian fundamentalists, so the effects are quite sizeable, especially when they are compared to the mean score for the interviewed immigrants.
Grossmann: They did not find an effect on trade opinions.
Ruggieri: Trade is a complex phenomenon, and the complexity of this phenomenon helps justify why we’ll not find the sizeable effect of inter competition on opinions about trade. It may be the case that in local communities, small towns where then opinions might be affected more, but at least in our analysis, which is based on local areas such as commuting zones, this effect is almost non-existent.
Grossmann: Despite that key difference between the paper’s findings, both agree that economics can help change racial attitudes. Bisbee says economic conditions can exaggerate racial concerns.
Bisbee: I think it’s sort of a false dichotomy to say it’s either economic resentment or it’s racial resentment. I think what I show is that sort of humans’ worst instincts, our xenophobia, our racial resentment, these are not exogenous, they’re not handed down by God when we’re born. We are not born with a certain level of animosity, rather they’re exaggerated or muted by economic conditions. I think there’s space in the current debate to acknowledge that the racial resentment story is potentially itself endogenous to an economic decline.
Grossmann: Ruggieri says people look for easy group targets when they suffer economically.
Ruggieri: One individual’s experience, economic insecurity due to, in this case, economic downturn due to foreign competition, our argument is that they look for potential targets that are easily identifiable who can be directly or indirectly blamed for the economic insecurity. Of course the fact that these categories bear absolutely no relevance to the economic downturn is irrelevant, and the most easily identifiable targets seem to be ethnic and religious minorities as well as immigrants. This happens, in our view, because the short and especially the long term consequences of international trade may be quite difficult to grasp for ordinary citizens.
It’s true that if a factory shuts down in a local community, everybody will understand its causes, but this does not imply that people would change their opinions and would have capability of changing their opinions about international trade in a way that would affect voting behavior. Culture backlash seems to be a channel that consistently predicts support for Republicans in areas subject to tighter inter competition. It started all in the south with the presence of manufacturing was substantial, starting from 2008, and this was enriched by the mid-west in 2016. Nativism is really a consequence of how people change their opinions about targets that are easily identifiable.
Grossmann: Ruggieri found that most of the negative effects of trade on Democratic voting are mediated by their effects on racial attitudes.
Ruggieri: What we’re finding that is I think very interesting is the idea the direct effect of inter competition on voting for Republicans, what we actually see in the data, is actually smaller in magnitude than the mediated effect. What it means is that if the effect of inter competition on voting for Republicans were only channeled through these mediators, such as opinions about ethnic minorities and religious out-groups, this effect will be even stronger than the actual one we observe in the data by a factor of two or even three. There are actually some factors which seem to mitigate the effect of inter competition on the support for Republicans, and this may be due, for example, to the economic benefits from trade.
Grossmann: Another key difference is that while Bisbee looked at changes from Romney to Trump voting, Ruggieri showed both the trade effect and the cultural backlash mechanism helped Republicans in 2008 and 2012 as well.
Ruggieri: A lot of the effect of the pro conservative effect of inter competition is still there in 2008 and 2012. Trump really turned the table in the mid-west and did something unique in the region that had never been Republican. Even if we analyze the three elections separately, we see an effect, a pro conservative effect. The effect of inter competition on individual attitudes towards ethnic and racial minorities and immigrants was still there in 2008. Probably it was a bit confounded by the anti-incumbent Obama effect in 2012, but we definitely find a strong effect of penetration on [inaudible 00:33:25] attitudes in 2008 and 2016, which confirms previous evidence by other research, which points to the pro conservative effect of inter competition even in Congressional elections in 2008 and 2016.
Grossmann: Bisbee says the differences are likely attributable to the trade question they use and how close the local trade shock measures are to the individuals surveyed.
Bisbee: Oh, and they show that it’s mostly resentment is towards Asians and Hispanics, but not toward African-Americans, so they think this is evidence of more recent ethnic minorities is consistent with an anti-immigrant story, but not a broader racial antipathy story. They also find no new effect in their measure for opinions on trade, and I think there’s probably two reasons why we come up with the same findings for xenophobia, but different findings for trade. I think the first is simply the question that I use, again, it’s this Pew Research survey question that says do you think free trade has been good or bad for the United States. Insofar as trade is a relatively complicated policy, there’s tariffs, there’s non-tariff barriers, there’s quotas, there’s all these different policy tools that can be used. An easier question like the Pew one is more likely to elicit responses from survey respondents. Francesco and his co-authors, think they used three different measures, one of them asking about imports, it’s about imports, it’s about trade agreements. There is slightly more detailed questions which may not elicit the same sort of gut reaction that my respondents seemed to. That’s potentially the first explanation.
The second is the unit of analysis. They’re aggregating up, they’re doing Congressional districts, which is absolutely the appropriate unit for the questions that they’re trying to answer, whereas I’m much smaller. I’m either doing commuting zones, counties or, in sort of, I’ve got this section of my paper that says let’s forget about geographic units completely. Let’s just measure distance between every respondent and every firm and apply these different weights to assess how fast the decay, like how far away do you have to move before you’re no longer reacting to these trade shocks?
Grossmann: Ruggieri agrees that the differences in findings are likely based on the trade survey question and the measure of trade shocks.
Ruggieri: Our papers, of course, share the same interest in shedding light on the mechanism linking international trade and mass public opinion in the United States. There are some differences in terms of, you know, research design and of course, as a consequence, results. Two of the main differences I think are the dependent variable, which is basically how questions are being asked. Bisbee uses data from the Pew Research Center, while we rely on the American National Election Studies and actually questions asked are quite different. The wording of the survey is different. Our questions are mainly based on specific policy related to international trade, while Pew questions are closer to general opinions about trade.
The main difference I would say is in the independent variable. Using county level trade-related lay offs from the TAA is actually quite different from using a measure of inter competition at the commuting zone level, which as I mentioned before, really captures both the direct effect of trade integration and indirect effect on the service, non-tradable sector, it turns out. These differences can explain why we have no findings on specific questions about international trade policy and in fact he finds a significant effect. These differences may be partly due to these differences in surveys and the dependent variables, but I think they can be reconciled. There is some evidence supporting the fact that there is bundling in how people have opinions, form their opinions about immigrants, minorities and international trade. Maybe the fact that there is a stronger effect on immigration [inaudible 00:37:45] out-groups, but at the same time, this carries over the effect on opinions about international trade.
Grossmann: Is this a specific story about U.S. politics, or a broader trend? Ruggieri sees lots of similarities with European politics.
Ruggieri: The channel is definitely there even in Europe. Also there is current research done by [Colantoni Estenic, 00:38:05] at Bocconi University which actually is very closely related to our findings. They showed that opinions about immigrants, minorities, authoritarian values are triggered by trade integration, and voters in those regions actually seem to behave quite similarly to Republican voters in the mid-west and the south. The similarities in terms of geography and elite rhetoric are really striking. The interesting question will be how will governments react to this?
Grossmann: He says we don’t know if the findings are specific to Chinese competition or to trade more generally.
Ruggieri: In order to have a plausible exogenous variation due to the unprecedented expansion of a big country’s manufacturing capacity, we are using China as a benchmark. It’s hard to tell whether this would be true if any other country’s manufacturing capacity expanded. Definitely the surge in Chinese exports to the entire world is a unique phenomenon that provided a fantastic natural experiment for researchers to study the political consequences, and not only the political consequences of international trade.
The thing is that China is used as a tool to study these effects. The external validity is, of course, a big question mark, as in all research that is rigorous and empirical economics. It’s hard to tell, but definitely there’s a lot specific to China about this research because the research design is based on a very specific and unprecedented, as I mentioned, surge in Chinese exports throughout the entire world.
Grossmann: What comes next? Bisbee says Republican opinion has shifted back and forth on trade, making it not obvious whether Congress will follow Trump towards protectionism.
Bisbee: The degree to which rank and file GOP members of Congress will go along with Trump and break with sort of the traditional party platform is always going to be a function of whether they think that will help them electorally. It’s always going to be a function of whether their constituents are on board. One thing that is really fascinating to look at is … The Pew Research Center asks, I’ve used their data in my paper, and they ask the question that I use as my outcome variable is do you think free trade has been a good thing or a bad thing for the United States? Their most recent plots … after the period that I analyzed, they continued to ask this question in 2017, in 2018. It’s this amazing partisan divergence, where Republicans were towing the party line, pro free trade right up until 2016, and then absolutely nose dived, literally just fell off a cliff, which is, you know, you can’t get clearer evidence of the impact of a leaked communication on public preferences than that.
What is even more interesting, potentially … That’s sort of the main year of the shift was in I think March or May of 2017, and then mid 2018, they released another survey asking the same question, and Republicans are rebounding, right? They’ve gone from, I think, 60-ish percent support of free trade, down to a low in 2017 of I think 36%, and now they’re coming back up sort of reverting mean reverting back to the traditional sense. I think that is very important to keep in mind when we’re trying to guess how effective is this platform for Trump going to be? A lot of the conversation about the first two years of his presidency has been, again and again, GOP leaders falling in line with him. I don’t think there’s a fundamental change in the calculus there. I think the GOP leaders were reading the tea leaves among their constituents, and I think potentially this mean reversion of support for free trade among Republican voters could suggest that the GOP members of Congress are not going to follow along with Trump’s trade plan as it goes forward.
Grossmann: Ruggieri agrees that Republicans are divided, even their voters.
Ruggieri: One important factor to bear in mind for the Republican party is highly heterogeneous, and we saw that in the recent mid-term election in November. For example, plenty of anecdotal evidence from Orange County, California, points to the fact that Republican voters are not just rural or living in manufacturing intensive areas, but may be well educated, well off and living in suburban areas. Those voters are unlikely to be supporting protectionist stances. We believe this will make it very difficult to form a coalition on divisive topics such as protectionist policy.
Grossmann: There’s, of course, a lot more research to do. Bisbee wants to look at the local environment of trade shocks. Can people really see it in their city?
Bisbee: The transmission mechanisms are the really interesting thing now, both in terms of scholarly interest and in terms of, you know, if you’re a policy maker or a politician, an advocate, whatever, understanding exactly how these opinions move is crucial. A couple of things I’ve been doing, I already mentioned looking at mediation analysis using designated market areas. I’m also collecting data at the moment that uses Google, what is it, Street View, where I’m going to, for more recent lay offs in the TAA dataset, I want to use Google Street View to actually look at and use image analysis to look at the environment around which these lay offs are being recorded, to try and assess whether sort of social/urban decay is at all prognostic of the strength of the response toward the lay offs, right? Again, this is going to speak to this question of is it a leak communication, or is it your day to day life, what you see as you go about your business?
Grossmann: Ruggieri wants to look at whether governments can do anything to compensate the losers from trade, and whether elites are driving the cultural backlash or just responding to it.
Ruggieri: There are lots of interesting avenues of research stemming from this literature. I think it’s surprising how little we know about the political effects of welfare state are aimed at compensating the losers from globalization and seeing how and to which extent state and local governments can compensate these losers and attenuate, if we can say that, the effects of globalization on the rise of radical right movements and increasing polarization, so more research we think is needed on the geography of the political effects of wealth transfers in the United States, especially exploiting the state and local heterogeneity in this context.
Another interesting avenue is disentangling demand and supply in the effect of inter competitional voting. In our text analysis, we show a change in the equilibrium condition over time, show that Republicans switched from support of free trade to protectionism, and we also know that people residing in tighter [intercompetitionaries, 00:45:33] were more likely to support a Republican candidate, but we do not know whether this is a bottom up or a top down phenomenon. Who’s driving this fact? Who’s driving this change in the equilibrium condition? Is it elite rhetoric mainly, or bottom up approach stemming from the fact that people experience economic insecurity and the rhetoric matches this demand?
Grossmann: There’s a lot more to learn. Political Research Digest is available bi-weekly from the Niskanen Center and on iTunes. I’m our host, Matt Grossmann. Thanks to Francesco Ruggieri and James Bisbee for joining me. Please check out their papers and then join us next time.