Recent border carbon measures have relied on the theory that stricter environmental rules in rich countries push pollution-intensive production toward developing economies with weaker regulations. In new research, Irfan Saleem and Giray Gozgor show that the “pollution haven” mechanism is neither automatic nor uniform across industries. Evidence is mixed, often small in magnitude, and highly sensitive to how we measure regulation, model trade, and account for industry mobility.
If a country tightens its environmental policy, policymakers often worry that emissions may “leak” abroad to developing countries as manufacturers and investors shift their operations to jurisdictions with weaker regulation that is cheaper to comply with. Consequently, these efforts can undermine climate goals and harm domestic competitiveness. This concern about carbon leakage is widely cited in discussions of carbon border adjustments, including the European Union’s Carbon Border Adjustment Mechanism (CBAM), which imposes a carbon cost on imports to offset environmental impacts and align trade with domestic climate policy. Many of these debates rest on a deceptively simple claim: companies will reduce compliance when regulation tightens in the North by relocating pollution-heavy production to the South.
That is the essence of the Pollution Haven Hypothesis (PHH). It sounds plausible, and in some contexts, it may be true. But treating it as a default assumption can lead to blunt policy design, especially when effects vary sharply by sector, supply chain position, and the logistical feasibility of industries to shift operational locations.
There is a more evidence-based way to frame the issue: pollution havens are a testable hypothesis, but not a universal law. And the most useful policy insights come from understanding when and where the mechanism is likely to operate.
Why the empirical evidence is mixed
The PHH is difficult to test cleanly for three recurring reasons: (1) measurement and endogeneity problems, (2) aggregation of trade flows that hides sector-level mechanisms, and (3) large differences in mobility across industries.
Environmental regulation is hard to measure and often endogenous, meaning outcomes like carbon emissions can influence and be influenced by regulation. Furthermore, governments tighten rules when pollution rises, when public pressure increases, or when industrial structures change. This all can bias simple correlations between legal stringency and trade flows. This is one reason why classic works, such as the one by Arik Levinson and M. Scott Taylor, emphasize the presence of unobserved heterogeneity, endogeneity, and aggregation, and show that estimated effects can look very different once these issues are addressed.
Aggregation creates another challenge for testing PHH. Pollution haven effects are rarely caused by “macro” phenomena that show up clearly in aggregate trade flows. They tend to be sector-specific, and even within sectors, effects depend on technology, input structure, and the feasibility of relocation. A recurring lesson from empirical literature is that broad, one-size-fits-all tests can wash out the very mechanisms the PHH is meant to capture.
Similarly, firm mobility differs sharply across industries. Many heavy industries, such as mining and construction, are capital-intensive and tied to infrastructure and supply chains. In theory, non-mobile companies are less likely to fulfill the PHH because of high relocation costs. On the other hand, technology and financial companies are considered footloose, and therefore more likely to move to areas with lower environmental regulations. This dynamic motivates theories emphasizing “footlooseness” as a key condition for a strong pollution haven channel.
The overall message is not “the PHH is false.” It is that the PHH is conditionally stronger in some places, times, and sectors than in others.
Carbon border measures raise the stakes for getting the mechanism right.
Because leakage is hard to measure directly, border carbon measures often rely on plausible theory plus modelling. A growing research agenda evaluates how CBAM-type policies affect leakage, competitiveness, and supply chains. Modelling work suggests CBAM can reduce leakage and change incentives, but also highlights distributional and competitiveness effects across sectors and downstream industries.
Take cement and steel, which are capital-intensive and geographically anchored due to infrastructure, logistics, and local demand conditions. In these sectors, relocation is costly and slow. If firms cannot easily move production abroad, leakage may occur not through relocation, but through reduced domestic output, delayed investment, or increased imports of intermediate goods. A CBAM that focuses only on direct imports of finished steel, for example, may miss embedded emissions in downstream products such as machinery or construction components.
By contrast, in more mobile manufacturing segments, such as certain chemicals or basic manufacturing inputs, firms may find it easier to shift marginal production abroad. In these sectors, CBAM may reduce leakage more effectively, but only if coverage is broad and enforcement is credible. Narrow product coverage can create incentives to reroute trade through lightly processed intermediates or third countries.
Implementation design also matters. If a CBAM relies on default emissions benchmarks rather than firm-specific data, it may over- or under-price carbon content, distorting trade without accurately targeting high emitters. Similarly, if foreign producers can claim exemptions under nominal climate policies lacking enforcement, leakage risks may persist despite formal compliance.
Legal and political constraints matter too. Policy and parliamentary discussions emphasize that CBAMs are designed to address leakage risk, but implementation choices and international compatibility shape feasibility and legitimacy.
This is exactly why the pollution haven narrative needs to be handled carefully. If policymakers assume a strong, generalized relocation mechanism, they may overestimate leakage risk in sectors where relocation is structurally difficult, underestimate other adjustment margins (abatement, innovation, product shifting), and design measures that are too blunt for supply-chain reality.
A South Asia example of why nuance matters
These general lessons are especially clear in a region many would expect to be a “textbook” pollution haven: South Asia trading with the OECD. It looks textbook because the PHH is most plausible when there is a large and persistent gap in environmental stringency between trading partners, tighter rules in rich importing economies, and weaker enforcement in developing exporting economies, combined with deep trade links that could transmit compliance costs through supply chains.
At first glance, the descriptive evidence can seem to fit that storyline. South Asia’s mix of exports has shifted over time, and some pollution-intensive categories have expanded, which can be read as “dirty industry moving south.” Our paper shows why that inference is too quick. Once we move from descriptive patterns to more rigorous testing using industry-level data, the results are far less supportive of a simple relocation mechanism. The apparent “pollution haven” patterns are weaker and less robust, suggesting that changes in South Asia’s pollution-intensive exports are driven at least as much by other forces (comparative advantage, upgrading, demand shifts, and supply-chain dynamics) as by regulation-induced relocation. The takeaway is not that pollution havens never occur, but that even in a setting that looks tailor-made for PHH, the mechanism is conditional rather than automatic.
What an evidence-based approach implies for policy
If we want climate-and-trade policy to be effective and defensible, we should treat the PHH as a starting hypothesis and design around the conditions under which leakage is most plausible.
First, make policy sector-specific rather than narrative-driven. Recent empirical work emphasizes that effects differ across sectors. That means policy tools should be targeted, reflecting industry mobility, emissions intensity, supply-chain position, and substitutability across sources.
Second, distinguish relocation from other margins of adjustment. Leakage can occur through relocation, reduced domestic output without relocation, import substitution, changes in product mix, and delayed investment. If relocation is constrained in heavy industry, we should expect a larger role for these other channels.
Third, pair border measures with upgrading pathways. If developing exporters face competitiveness pressures from tightening standards, whether domestic or foreign, then support for cleaner upgrading becomes central. Otherwise, “trade and climate” risks become a conflict rather than a transition strategy.
Finally, build legitimacy through transparent evidence. Border policies that reshape trade need credibility. That requires transparent measurement, careful sector design, and realistic assumptions about how firms and industries respond, not slogans.
Bottom line
The PHH remains a powerful political story, and in some contexts, particularly for mobile industries and multinational production, it may be economically meaningful. But decades of research also show that the mechanism is conditional, often difficult to identify, and far from universal.
For today’s climate-and-trade agenda, the key takeaway is simple: design policy for the world we have—heterogeneous industries, complex supply chains, and multiple adjustment margins—rather than for a one-line narrative about “dirty industry flight.”
Authors’ Disclosures: The authors report no conflicts of interest. You can read our disclosure policy here.
Articles represent the opinions of their writers, not necessarily those of the University of Chicago, the Booth School of Business, or its faculty.
Subscribe here for ProMarket’s weekly newsletter, Special Interest, to stay up to date on ProMarket’s coverage of the political economy and other content from the Stigler Center.
