#29: Dr. Joseph Shapiro – The $800 billion Implicit Subsidy for Dirty Industries Due to Trade Policy
Introduction and Transcript
Introduction
Sometimes you hear: “we should stop subsidizing dirty industries.” But are we actually doing that—and how big is it?
In this episode, I’m joined (again) by Professor Joe Shapiro (UC Berkeley) to discuss his paper “The Environmental Bias of Trade Policy” (QJE, 2021). Joe’s core finding is striking: dirty industries tend to face lower trade protection, while cleaner industries face higher trade protection—a pattern that appears across countries, years, and even non-tariff barriers.
Joe then translates that pattern into an intuitive metric: if you interpret existing trade policy as a carbon tariff, it looks like an implicit carbon subsidy of around -$100 per ton of CO₂ (roughly -$85 to -$120/tCO₂) or $550 to $800 billion dollars per year. In other words, goods with higher embedded emissions often face less trade protection—trade policy gets the magnitude “about right,” but the sign wrong.
We unpack what drives this: tariff escalation linked to “upstreamness” (upstream, commodity-like inputs tend to be dirtier and receive lower protection; downstream consumer goods are cleaner and receive higher protection).
Finally, Joe walks through his model-based simulations showing that harmonizing protection between clean and dirty goods could modestly raise GDP while meaningfully reducing global emissions.
In this episode
The key empirical fact: dirty industries have low tariffs; clean industries have high tariffs
The implied magnitude: ~-$100/tCO₂ “implicit carbon tax” embedded in trade policy
The mechanism: upstreamness → tariff escalation → environmental bias
What the simulations show when you “flatten” the bias across sectors
A surprising map result: countries with strong climate reputations can still have trade policy that tilts toward dirtier goods
If you care about CBAM, industrial policy, or the political economy of decarbonization, this episode is a powerful reminder: trade policy can be climate policy—even when nobody intended it.
For questions, comments or suggestions, you can contact me at arvid.viaene.ce@gmail.com
Transcript
Arvid Viaene:
Welcome to another episode. I’m your host, Arvid Viaene. Sometimes you hear the following statement: we should stop subsidizing dirty industries. Because if you want to tackle climate change, subsidizing dirty industries is one of the least productive things you can do.
But whenever I heard something like this, I often found myself thinking: is it true that we are subsidizing them? And if so, how large is the magnitude? Today we are going to discuss those questions, because I think it is vital to understand them. Especially since the implicit subsidies for dirty industries may be much bigger than you might suspect.
To do that, I’m delighted to have Professor Joe Shapiro join me again. He wrote an excellent article on this topic, published in the Quarterly Journal of Economics in 2021, called The Environmental Bias of Trade Policy.
Joe is an associate professor of economics at UC Berkeley and a research associate at the National Bureau of Economic Research and the Energy Institute at Haas. He studies climate change, air pollution, clean energy, and especially water pollution, with links to trade, public finance, and health. He holds a PhD in economics from MIT. And he is also the first repeat guest on my podcast.
So, Joe, welcome back.
Joseph Shapiro:
Thanks. I’m honored to be here again.
Arvid Viaene:
I had a wonderful time the first time. And now we’re talking about the environmental bias of trade policy. Could you explain what your paper is about?
Joseph Shapiro:
Yes. This paper asks: how does trade policy affect greenhouse gas emissions?
Usually when people ask that question, they want to know whether having a high level of tariffs affects carbon emissions. I approached it from a different angle. Most countries have different levels of tariffs across industries, and I wanted to know how those patterns affect carbon emissions.
I really started this paper because I was interested in a policy many countries had talked about, but at the time only one region had implemented. That policy is called carbon tariffs, or carbon border adjustment. Europe has one now. The concern is that if one region has a serious climate policy and other regions do not, then dirty industries will simply relocate away from the climate policy. In that case, global emissions do not actually decrease. People refer to that concern as leakage.
So, for example, if Europe seriously regulates carbon and other regions do not, then carbon could leak away from Europe. Europe might just start importing dirty goods from China, India, Mexico, and other trading partners. For many years, countries have said: if that is a problem, let’s create a tariff proportional to the carbon emitted in producing those goods — a high tariff on dirty goods, a low tariff on clean goods — and that will solve the leakage problem.
When I talk to undergraduates, they refer to this as a kind of whack-a-mole problem. Europe hits carbon at home, but then it pops up somewhere else. So I thought a long time ago: a carbon tariff is just a tariff by another name. It is simply a tariff with a high rate on dirty goods and a low rate on clean goods.
And then I thought: I wonder if countries already have this and nobody noticed. Maybe they already have high tariffs on dirty goods and low tariffs on clean goods. That would be plausible, because dirty industries spend a lot of resources on lobbying. One of the goals of lobbying is often to obtain protection from foreign competition. So I thought it was plausible that dirty industries could have high tariffs and clean industries lower ones. And I thought: let’s just look at the tariff data and see.
Arvid Viaene:
And I think, like you say, a carbon border adjustment mechanism is basically a tariff. It is a tariff between regions. So what did you find when you analyzed this problem?
The Surprising Trade Pattern
Joseph Shapiro:
I found exactly the opposite of what I expected, which is not what happens in most papers. I found that dirty industries have low tariffs and clean industries have high tariffs. That is true in the US, where I started. It is also true in most other countries in the world where I then expanded the research. It is also true for other trade policies like quotas and product standards.
People collectively refer to all of those as non-tariff barriers, because they are barriers to trade besides tariffs. I also found that this pattern was present in all the years of tariff data available for the US. In sciences like physics or biology, people often want to estimate parameters out to six decimal points. In economics, sometimes we are happy if we can just get agreement on the sign of something, let alone the magnitude.
So it is pretty rare to find a pattern like this that shows up in different datasets, different countries, different years, and different measures of policy. It seemed like a very systematic and quantitatively important pattern. And then the paper went on to try to understand why this pattern exists and what it means for the environment and the global economy.
Arvid Viaene:
So you started with the prior that maybe there was lower trade protection on greener goods, but then you found the opposite. And then you found it was consistent across countries, measures, and time periods. I imagine you were pretty surprised that it just kept showing up.
Joseph Shapiro:
I was, but it was revealed to me in stages. I actually thought of the idea when I was in graduate school, but at the time I never looked at the data. I assumed it would just be a simple scatterplot of tariffs against carbon.
Then, five or six years later, I had to take a train every day from New Haven to New York to use restricted-access data at a data center in New York. It was a two- or three-hour round trip with no internet. So each Tuesday, I would pick a paper idea that had been gathering dust and spend the day on it. Being who I am, I had tariff and carbon data on my laptop. So I spent a couple of hours on the train looking at this.
Just for the US, it jumped off the page that clean industries had high tariffs and dirty industries had lower tariffs. I thought: well, that is kind of interesting. That summer, I was going to several trade conferences. At the time, I was not deeply involved in trade papers. But when economists talk in hallways at conferences, they always ask what you are working on.
So I just started describing this in conversations. And basically everyone said: that is really interesting — why is it happening, and what does it mean? Then that fall, I was applying for a grant at the National Science Foundation. I was going to describe three papers. Two were already fairly advanced, and then I decided to include this as the third.
I thought I should look at countries beyond the US. I realized I could use input-output data and national accounts to do this for many countries. So I did the same analysis internationally. At that point, I realized this was happening in most countries, not just the US. Then I got comments back from the grant reviewers, and they basically said: papers one and two are fine, but paper three sounds really interesting — you should definitely pursue that one.
And then I thought, okay, everybody is telling me this thing I had initially dismissed is actually worth writing about. Then I looked at non-tariff barriers and found the same pattern.
So yes, the big surprise first came on the train, when I saw the US result had the opposite sign from what I expected. But I initially assumed it was probably a fluke. Once I looked across countries, I realized it was systematic. Academics can be slow to update their priors. It took me a while to become convinced this was a genuine scientific pattern. And I would say that understanding why it was happening was also important for persuading me that it was real.
The Magnitude of Implicit Subsidy for Dirty Industries
Arvid Viaene:
I want to get to the reasons in a moment, but first I wanted to give listeners a sense of the magnitudes. One thing is that you found the pattern consistently. The other question is whether it is substantial.
Joseph Shapiro:
I summarize the magnitude as an implicit carbon tax. What I mean by that is the following: if a region like the European Union imposes a carbon border adjustment, it is basically a tariff on carbon emissions. They have to choose a carbon price — maybe $50 a ton, or $100, or $200.
So I wanted to ask: if you look at the trade policies countries already have — not policies called carbon tariffs, just ordinary tariffs and non-tariff barriers — what carbon tax rate would those imply if you interpreted them as carbon tariffs? I calculated that, on average around the globe, trade policy was creating a carbon tax rate in the ballpark of negative $100 per ton. The range was roughly minus 85 to minus 120 dollars per ton.
The minus sign matters, because it means that the more carbon is emitted, the lower the level of trade protection. So imagine two bundles of goods arriving at a port. If one of those bundles emits an additional ton of carbon dioxide in production, it faces about $100 less in tariffs and tariff-equivalent barriers.
Is that large or small? Today, leading estimates suggest the optimal carbon tax is probably around $200 per ton. At the time I was writing the paper, many estimates were more in the range of $50 to $100 per ton. So my summary was that trade policy accidentally got the magnitude about right — it just got the sign wrong. Another way I tried to show the importance of the magnitude was to sum it globally and ask: if this were a carbon tariff, how much revenue would it generate?
The answer is over $500 billion per year. I think the range was roughly $550 to $800 billion. Some macroeconomists say, “Call me when you get to trillions.” But this is still a very large number. One benchmark is direct fossil fuel subsidies, which policymakers often worry about. India has historically subsidized kerosene. Venezuela subsidized diesel. The IMF and World Bank have repeatedly written about these direct subsidies.
If you sum those globally, they are in roughly the same ballpark — around $500 billion per year. So I took that comparison to mean: this implicit subsidy to carbon emissions, embedded in trade policy, is of the same order of magnitude as the direct fossil fuel subsidies that attract so much attention.
Now, of course, it may have a very different effect on actual carbon emissions, because it is a more indirect subsidy. But the scale is surprisingly large.
Arvid Viaene:
That was one of the things that startled me when I read the paper. There is already a lot of attention on direct subsidies. But if you take your results seriously, you almost have to add a second layer of subsidy on top of that — these implicit subsidies coming from different tariff rates.
Joseph Shapiro:
Yes, though I would add a grain of salt to that comparison. Even though the dollar magnitudes are similar, they are very different kinds of subsidy. Giving someone a dollar to burn a gallon of kerosene is probably going to increase kerosene use more directly than giving an extra dollar to somebody importing a bundle of goods with higher carbon content.
Partly for that reason, the final part of the paper uses a mathematical model of trade, energy, and emissions to calculate what this all adds up to in environmental terms.
Why This Happens: Upstreamness vs Downstreamness
Arvid Viaene:
So we have the magnitude, which is striking. They got the magnitude close to the social cost of carbon — but with the wrong sign, which is a huge deal. So what is driving these implicit subsidies?
Joseph Shapiro:
First, let me rule out a reason. I do not think trade policymakers sat around a table and said, “Let’s burn up the planet. How can we best do that through trade policy?” There is no rational reason to do that.
And also, I talked to a lot of trade policymakers around the world in the course of writing this paper, and I found zero who were aware this was happening. Several actually reached out and said: this is news to us — can we talk to you about it? I ended up working as a consultant to Europe’s Directorate-General for Trade, and I talked to European parliamentarians, politicians, and US trade policymakers. So this was definitely not something people had on their radar.
So why is it happening? I realized in the process of writing this paper that trade policy is a somewhat distinct field from trade economics more generally. Trade policy puts much more emphasis on game theory, political influence, and patterns across industries.
Trade policy researchers have, for decades, tried to understand why different industries have different levels of trade protection. I took a list of those explanations — from a handbook chapter by Dani Rodrik — and asked: which of these explanations can account for why dirty industries have low protection and clean industries high protection?
There were about 15 explanations. I looked at them first in the US and then internationally. And going in, I assumed perhaps 10 of the 15 would matter and we would get some messy combination. That was totally wrong. One explanation really jumped off the page. The term is one economists made up about a decade ago: upstreamness.
Upstreamness means that some industries mostly sell their output to other firms, whereas other industries mostly sell directly to final consumers. For example, hot-rolled coils of steel are not something you buy at a grocery store. They are sold to other companies, which then transform them into final products. That makes steel a very upstream industry.
By contrast, bubble gum is a very downstream industry. It is sold directly to consumers. It turns out that for well over half a century, upstream commodities have had very low tariffs, while more downstream processed goods have had higher tariffs. That pattern is called tariff escalation.
In the mid- to late 20th century, it was central in both policy and academic discussions of trade policy. There is even an old paper in the Journal of Political Economy by Corden that says tariff escalation is so well known that detailed substantiation is hardly needed.
That was the state of knowledge in the 1960s: everybody knew upstream goods had low protection and downstream goods had high protection. It was somewhat forgotten over the following half century. When I was doing this research, many trade and trade-policy researchers either only vaguely remembered it or thought it was irrelevant.
But it turns out to be one of the most powerful empirical forces determining tariffs and non-tariff barriers across countries. Upstream goods have low protection, downstream goods have high protection.
And the key observation is that upstream goods are systematically dirty, while downstream goods are systematically clean. The real lightbulb moment for me came when I looked at a paper measuring upstreamness in the US economy. The most upstream industries were steel, cement, refined petroleum, chemicals, and aluminum. Those are exactly the industries you see in environmental research as carbon-intensive sectors.
So I thought: if those are the most upstream industries, and we know for decades that upstream industries have low trade protection, and it is obvious they are the dirtiest industries, then this has to be the smoking gun. Once you account for an industry’s upstreamness — how far it is from final consumers, and how much it sells to firms rather than households — the relationship between carbon intensity and trade policy almost disappears.
In fact, the first graph in the paper is basically an X-shape: as you move from upstream to downstream industries, tariffs go up and carbon intensity goes down. I then made that graph for the global economy, and then separately for each country. Astonishingly, almost every country showed a similar X-pattern. Some graphs were neater than others, but the pattern was highly systematic around the world.
Arvid Viaene:
I would actually recommend people go look at that graph, because it really is compelling. So then the next question is: what drives that result? Why do downstream firms get so much more protection?
Why Tariff Escalation Happens
Joseph Shapiro:
That is a great question. Writing this paper felt a little bit like peeling an onion. First: do we have a carbon tariff? Yes, but it is negative. Why? Because of upstreamness. Then: why does upstreamness matter?The paper does talk a bit about this, but I felt this was really a good question for trade policy more generally. I did not think I could fully answer it in this paper.
So I offered what I thought was a sufficient explanation, though not necessarily the only one. The explanation I found most persuasive is upstream versus downstream lobbying. Industries want cheap inputs. So they lobby legislators for low trade protection on the goods they use as inputs.
Many industries in advanced economies use steel, directly or indirectly. Car manufacturers use steel directly. Producers of pipes, machines, and many other goods use it directly or further up the supply chain. So when you ask who is arguing for low trade protection on steel, it is a huge share of industries.
Now compare that to bubble gum, which is a downstream consumer good. Who argues for low protection on bubble gum? Basically just final consumers — millions of dispersed individuals. It is easy to organize three or five large manufacturers who use steel. It is very hard to organize hundreds of millions of consumers to argue for low protection on a final good.
So you end up with lots of industries lobbying for low trade protection on upstream goods. And the only people who might argue for lower protection on downstream goods are final consumers, who are much harder to organize. That is one sufficient explanation for tariff escalation. It may not be the only one, but it is a persuasive one. And it helps explain why tariff escalation shows up not just in tariffs, but also in quotas, product standards, and other non-tariff barriers.
Arvid Viaene:
Yes, exactly. The same logic applies to quotas and the rest. Those five concentrated firms can coordinate, whereas final consumers are much more dispersed. One thing I do want to ask: you mentioned that in the 1960s this broad pattern was more widely understood, even if the term upstreamness was not used. But when you published this paper, many economists and policymakers still seemed surprised. Have you noticed increasing awareness over time?
Has Awareness Increased?
Joseph Shapiro:
I think so. I have only a limited orbit, of course, but when I talk to people in the trade-policy and climate-policy space now, I do think this pattern is known.
When countries choose trade policy, there are many interest groups involved. There is the industry itself, downstream users, consumer groups, and so on. But environmental groups are usually not in that hallway, unless we are talking explicitly about solar, wind, or other clean-tech goods. That is not really what this paper is about.
So one idea I had was: when countries are deciding whether to put tariffs on steel, bubble gum, or fossil fuels, environmental groups actually have reasons to be in that conversation. I am not directly inside those negotiations. But my sense is that much of the conversation linking trade and the environment still focuses on carbon border adjustments and leakage.
That said, I do know that some regions have considered the kind of issues this paper raises when thinking about broader trade reform. For example, after COVID, the European Commission was discussing whether it should undertake major trade-policy reform, and this kind of issue was on the agenda. They have a trade-policy day each year, and I joined their trade-policy committee in Parliament and spoke about this.
So it is in the conversation. I do not know exactly how it will play out, but I think of research as starting a snowball rolling downhill. Hopefully it gathers momentum, but it is hard to predict how, where, and when.
Arvid Viaene:
Yes, exactly. And I partly hope this conversation helps spread that awareness too. Because like you say, once you see it, it is kind of hard to unsee it. It is so systematic — across countries, across quotas, across non-tariff measures, across the graphs.
And I think you are right: when it comes to the EU ETS or a carbon border adjustment mechanism, environmental groups are clearly in the room. But if it is “just” a tariff on a product, they may not be around that table.
Joseph Shapiro:
I think more broadly, I have been really excited by the rapid growth in research at the intersection of trade and the environment over the last five years. There have been a lot of creative papers trying to identify ways trade policy might improve environmental outcomes, even if those ideas are not yet central to policymaking.
Some are clearly environmental policy tools, like carbon tariffs or climate clubs. Others sit somewhere in the middle. I think it is exciting that frontier economic research is building new tools, methods, and ideas that are also useful for policy.
This paper is one contribution to that conversation, but far from the only one. At a broad level, I also took from this research the lesson that trade policy can have quantitatively important environmental effects.
That idea, I think, has become more accepted. One reason for all the discussion is that climate change is a global public good with a severe free-rider problem. If one city pollutes another, maybe a state can coordinate. If one state pollutes another, perhaps a national government can coordinate.
But with climate change, countries are polluting each other. And we do not have many strong sticks or carrots to coordinate them. Trade policy is one natural tool that can provide such sticks or carrots.
Of course, the ideal climate policy would just be a carbon price. Then trade policy would have no special role. But the world is nowhere close to that optimum. Only about a quarter of global carbon emissions face any price, and those prices are often low, fragmented, and inconsistent.
In that kind of second-best setting, there may be some scope for trade policy to improve environmental outcomes. Of course, there is also scope for trade policy to make environmental problems worse. And many trade researchers reasonably worry that trade policy itself has had a rough few years, and that tying trade and climate together could make both harder rather than better.
So it is not guaranteed that linking trade and environmental policy will improve either one. It requires proceeding carefully.
What the Simulations Show
Arvid Viaene:
Exactly. That brings me to the last point I wanted to raise. In the paper, you also run simulations to think about possible policy reform. Could you talk about those?
Joseph Shapiro:
Yes. I set up a mathematical model of production, consumption, and trade for most countries in the world. Then I asked: what if we reformed trade policy so that clean and dirty goods faced roughly the same level of protection?
In other words, we are not increasing or decreasing tariffs on average. We are just harmonizing trade policy between clean and dirty goods. I found that if most countries implemented that reform, there would be a very slight increase in global GDP, but also a meaningful decrease in global greenhouse gas emissions — on the order of magnitude of the estimated impact of Europe’s cap-and-trade market in the mid-2010s.
The reason is that current trade policy is sending the wrong price signal. It encourages production, consumption, and trade in dirty goods. Once you take away that policy distortion, there is no longer an incentive to specialize in fossil-fuel-intensive production. Policy is no longer putting its finger on the scale and tipping the global economy toward international sourcing of carbon-intensive goods.
What Joe Is Working on Now
Arvid Viaene:
That is really interesting. One thing I also wanted to ask is: since writing this paper, have you continued in this field? Are you working on related things?
Joseph Shapiro:
Yes.
The paper I wrote most immediately afterward looked quite different on the surface — it was about institutions and the environment — and I published it a year or two ago. But if you compare the two papers, they share a common principle: a commodity tax that differs across industries can encourage the global economy to shift toward one set of industries rather than another. I am also working on a few other papers on how trade policy affects carbon emissions.
Some are with collaborators — Eki Garcia-Lembergman, Natalia Raimondo, and Andrés Rodríguez-Clare.
One of those projects is about trade and value chains. It tries to understand how different it is if you tax carbon at the point of fossil-fuel extraction, versus at the point of producing goods that use fossil fuels, versus at the point of import and export.
Another looks at international finance and how investment in carbon-intensive or clean production abroad can help transfer clean technology around the world. And one question there is whether the $100 billion a year in transfers envisioned under the Paris Agreement could partly happen through foreign investment and multinational production in clean industries.
Arvid Viaene:
That is really cool. And like you said earlier, this is clearly a growing research area. I feel like there are more and more papers coming out at the intersection of trade and climate. So maybe there will have to be a part three. Before we wrap up, is there anything else you still wanted to add about this paper that we have not covered?
Countries with Strong Climate Regulations Can Have Higher Subsidies
Joseph Shapiro:
Yes, just one other thing. The paper includes a map showing the extent to which each country is imposing higher trade protection on cleaner goods — in other words, the implicit subsidy to carbon emissions embedded in trade policy. And it has some pretty surprising patterns.
The countries where trade policy is doing the most to accelerate climate change include places like France, Germany, and Norway. The countries where trade policy is doing the least to accelerate climate change, in this cross-industry sense, include places like Saudi Arabia, Egypt, and Iran. That seems very strange at first. It is not because Norway is a super-polluter and Saudi Arabia is super-green. Stereotypically, it is the opposite. But that actually reinforces the idea that countries are not intentionally doing this. They develop trade policy for political-economy reasons, and unfortunately it ends up being strongly correlated with carbon emissions.
To the atmosphere, though, it does not matter whether carbon is emitted from Ohio or Shanghai, or whether it is emitted because of trade policy, diesel subsidies, or anything else. Carbon is carbon.
Arvid Viaene:
Exactly. That is actually a really strong point. Countries with strong climate reputations, like Norway or France, may be very focused on climate policy in other domains and yet still not be aware of this bias. So I really do hope this conversation contributes to increased awareness.
And to listeners: I really recommend checking the paper out. Joe, thanks so much.
Joseph Shapiro:
Thanks, everybody. Great to talk to you. It was fun to discuss these topics.
Arvid Viaene:
Yeah, thanks. You too.
