A central concern in the climate policy debate is whether ambitious mitigation efforts simply shift emissions abroad. This column addresses estimates carbon leakage rates and their drivers across a wide range of climate policy instruments, countries, and sectors. Leakage rates are moderate overall but considerably higher in small open economies and carbon-intensive sectors. Market-based instruments, such as carbon taxes, raise imported emissions through higher and dirtier imports, while regulations and technology support policies result in cleaner imports. These results highlight the potential of well-designed climate policy mixes to limit leakage and support global decarbonisation.

Existing evidence on carbon leakage is mixed. Early ex-post studies of carbon pricing typically found little evidence of leakage, largely because of low carbon prices and generous compensation schemes (Venmans et al. 2020, Dechezleprêtre et al. 2022). As carbon prices have risen, more recent empirical work, such as Teusch et al. (2024), has found an average leakage rate of 13% across the cement and steel sectors. Ex-ante modelling studies have long projected meaningful leakage rates, with computable general equilibrium (CGE) models frequently estimating effects of 5–30%. Studies of policies beyond carbon pricing have yielded even more heterogeneous conclusions. Some reported moderate leakage linked to cross-country energy price differences, while others identified no leakage or even negative leakage effects, including evidence that climate legislation abroad reduces domestic emissions (Misch and Wingender 2024, Eskander and Fankhauser 2023).

Against this backdrop, our analysis provides novel estimates of carbon leakage rates for 14 manufacturing sectors, 49 countries, and a broad set of climate policy instruments, employing a unified empirical framework (Hemmerlé et al. 2025). Using a gravity model with international and domestic trade and emissions flows, we estimate how unilateral increases in domestic climate policy stringency affect imported emissions, import volumes, and the carbon intensity of imports relative to their domestic counterparts. Crucially, we distinguish the effects of market-based (e.g. carbon taxes), non-market-based (e.g. regulations), and technology-support measures (e.g. subsidies) – each of which may shape carbon leakage in distinct ways.

Climate policy stringency is rising and becoming more heterogeneous

Climate policy stringency has increased in recent years, but not uniformly across countries. We track these developments using the OECD Climate Actions and Policies Measurement Framework (CAPMF) – the most comprehensive harmonised database of climate policy stringency available to date – which scores 56 policy instruments across 49 countries on a 0–10 scale. To capture the stringency of countries’ overall mitigation policy mix, we construct a composite indicator at the country-year level by averaging the scores across policy instruments. This composite indicator shows that countries with already stringent policies in 2010 strengthened their policies the most by 2020 (Figure 1).

Stringency has also risen across the three CAPMF policy subindices – market-based, non-market-based, and technology-support instruments – along with a widening dispersion. As countries move forward with different mitigation policy mixes, the leakage risks they face depend not only on whether they act, but also on how they act.

Figure 1 Overall climate policy stringency across countries in 2010 and 2020

Note: The figure shows overall climate policy stringency by country in 2010 (black diamonds) and 2020 (blue bars). Overall stringency is measured using the composite CAPMF index, constructed by averaging stringency scores across policy instruments. The composite index aggregates three policy sub-indices: market-based, non-market-based, and technology-support policies. Stringency scores of policy instruments range from 0 (least stringent) to 10 (most stringent).
Source: Authors.

Climate policies raise imported emissions through higher import volumes

Imported emissions can rise because a country imports more (scale effect), because the carbon intensity of imports increases (technique effect), or because the composition of imports shifts towards carbon-intensive sectors or products (Aichele and Felbermayr 2015, Rottner and von Graevenitz 2024). For a unilateral one-unit increase in the composite CAPMF index – roughly what an average country accumulates over nine years – we find, over 2000–2020, that:

• imported emissions increase by about 10%;
• import volumes increase by around 11%, indicating that the scale effect explains the increase in imported emissions; and
• the carbon intensity of imports falls by nearly 6%, suggesting that imports become cleaner.

Sectoral results reinforce these findings. Climate policy tightening raises imported emissions more strongly in high and medium-emission intensity industries, such as basic metals and rubber and plastics, because import volumes in these sectors respond more sharply (i.e. a composition effect). Yet, all sectors experience a slight decline in the emission intensity of imports. This is consistent with Levinson (2023), who shows that the import baskets of high-income OECD economies have shifted toward cleaner products over time.

Different policy instruments yield different leakage effects

Market-based instruments – including carbon taxes, emissions trading systems, and fossil fuel excise taxes – increase imported emissions the most. They raise import volumes and, unlike other policy types, increase the average carbon intensity of imports. This aligns with recent work of Wang and Kuusi (2024) and Coster et al. (2025), who find that the EU Emissions Trading System (EU ETS) increased the carbon intensity of European imports.

In contrast, technology support policies, such as RD&D subsidies and financing mechanisms for green technologies, reduce the carbon intensity of imports. A possible explanation for this finding is that lower costs and greater availability of low-carbon technologies encourage firms and consumers to shift demand, on average, towards cleaner products and inputs, including those sourced from abroad. At the same time, these policies increase import volumes, likely reflecting an income (or rebound) effect as they can lower production costs and raise output, thereby increasing demand for imports. The positive effect on imports and the negative effect on the carbon intensity of imports offset each other to some extent, resulting in a modest increase in imported emissions.

Non-market-based policies – such as performance standards, technology standards, and energy efficiency mandates – also reduce the carbon intensity of imports while having little effect on import volumes. A possible explanation is that stricter standards shift demand towards products and suppliers that are, on average, more efficient or ‘greener’ and have upgraded technologies to comply with these rules. In addition, regulations can act as de facto international standards: firms operating in multiple markets may harmonise production practices to the strictest regulations in their main export destinations. This can promote the cross-border diffusion of cleaner technologies and production processes (Vogel 1997, Cherniwchan & Scott Taylor 2022). At the same time, these instruments can lead to higher compliance costs than comparable market-based instruments.

So how large is the carbon leakage rate?

A widely used measure to calculate carbon leakage rates is the ratio of the increase in imported emissions to the reduction in domestic emissions induced by climate policy. Because our baseline model identifies effects on imported relative to domestic emissions (following the recent literature), we modify it to estimate policy effects on each separately and calculate leakage rates for each country–sector pair. These leakage rates reflect both how strongly emissions respond to policy changes and how heavily sectors in each country rely on imported rather than domestically produced goods. Across all country-sector pairs, we find the following:

• The median leakage rate is around 3%, meaning that for every 100 tonnes of domestic emission reductions, about 3 tonnes leak abroad.
• The arithmetic average leakage rate is, however, around 14%, reflecting higher leakage in small open economies and in carbon-intensive sectors.
• The basic metals and other non-metallic minerals sectors (including cement) face the highest leakage rates (Figure 2). Within these sectors, some countries show higher leakage than others because they rely more on imports to meet domestic demand.

Figure 2 Cross-country carbon leakage rates in high-emission intensive sectors

Note: Each boxplot shows the distribution of estimated carbon leakage rates across countries for the indicated sector. The black line represents the median. The figure includes all countries used in the analysis and excludes outliers. The carbon leakage rate is calculated as the ratio of the policy-induced change in imported emissions over the change in domestic emissions (expressed as percentage change), multiplied by the ratio of imported to domestic emission bases (expressed in physical units). The ratio of foreign to domestic emission bases is calculated using data for the year 2019. The carbon leakage rate estimates capture short-term and linear effects; leakage rates could differ in the longer term and be non-linear.
Source: Author’s computations

The impact of carbon leakage on global emissions crucially also depends on the size of the domestic sector. A high leakage rate for a small sector will result in small absolute emission leakage, while a moderate leakage rate for a sector with a large emission base will instead lead to large absolute leakage. Calculating country-wide carbon leakage rates using sectoral emission bases as weights accounts for this and better represents the leakage risks each country faces. The average carbon leakage rate calculated this way is around 9% across countries, with small open economies – such as Latvia, Costa Rica, and Malta – facing larger leakage rates than large economies such as China and India (Figure 3). Small open economies face higher leakage rates because a large share of their domestic demand is met through imports.

Figure 3 Country-wide weighted average leakage rates

Note: The figure shows the country-wide leakage rate following a unit increase in the composite CAPMF. The country-wide leakage rate is calculated as a weighted average with sectoral imported and domestic emission bases as weights. The sectoral imported and domestic emissions bases are based on data for the year 2019. The carbon leakage rate estimates capture short-term and linear effects; leakage rates could differ in the longer term and be non-linear.
Source: Author’s computations.

Implications for policy

Our results have several implications for national and international climate policy frameworks:

• The moderate overall leakage rates and the absence of a shift towards more carbon-intensive imports under the overall climate policy mix suggest that unilateral action can still contribute to global emission reductions. However, the scale-driven rise in imports and higher leakage rates in some emission-intensive industries may raise concerns about competitiveness and effectiveness.
• Climate policies differ in the leakage risks they create. Market-based instruments tend to contribute most to carbon leakage. Regulations that apply equally to domestic and foreign producers are less prone to inducing leakage but can involve higher costs than comparable market-based approaches. Technology support policies can complement these measures by helping shift demand toward cleaner inputs.
• Taken together, the evidence suggests that combining market-based, non-market-based, and technology support instruments can balance concerns over cost effectiveness, competitiveness, and global emission outcomes.
• Border carbon adjustment mechanisms can mitigate leakage risks from market-based instruments in directly regulated sectors, but their design should avoid counteracting the cleaner import patterns induced by technology support and non-market-based policies and be consistent with rules-based trade.

Source : VOXeu