HAS THE EUROPEAN EMISSIONS TRADING SYSTEM INDUCED CARBON LEAKAGE IN EUROPEAN MANUFACTURING? A SECTOR-LEVEL ANALYSIS

摘要

OverviewFacing the global problem of climate change, climate policy that regulates only one region, e.g. Europe, risks beingineffective: such unilateral policy may lead to a shift of emissions from regulated to unregulated parts of the world –a phenomenon known as carbon leakage – without solving the global climate problem. The main climate policyaffecting the traded goods sectors in Europe is the EU Emissions Trading System (EU ETS). Under the EU ETS twotypes of emissions are regulated for manufacturing firms: Emissions from own power and heat production, as well asprocess emissions. Furthermore, firms may incur indirect costs due to higher prices for electricity they purchase. Ifcarbon leakage were sizable, it would make EU policy efforts ineffective and moreover disadvantage European firmsvis-à-vis their international competitors, potentially leading to job loss and economic downturn. This explains whythe leakage question has become a key topic for research and policy makers.This paper addresses the research question of whether the introduction of the EU ETS has caused carbonleakage. The question is highly controversial from both an academic and a policy perspective. Ex ante approachespredict leakage, while ex post analyses typically fail to confirm these predictions. Ex ante evaluations typically useequilibrium models, finding positive carbon leakage rates of up to a 25% (Felder and Rutherford, 1993). Leakagemay be reduced by using protective measures such as border carbon adjustments or output-based allocation ofallowances (Fischer and Fox, 2007; Monjon and Quirion, 2011; Meunier et al., 2014). Ex post studies typically findthat the EU ETS had a limited impact on carbon leakage (Chan, 2013; Sartor, 2013). Using installation-level surveydata of multinational firms Dechezlepretre et al. (2014) find no evidence of relocation within firms. Martin et al.(2014) conclude that current EU ETS rules on free allocation of allowances to industry over-compensates for leakagerisk in many sectors. In contrast Aichele and Felbermayr (2015), considering the broader context of the KyotoProtocol, find that being a party to the Kyoto Protocol has led to carbon leakage.We contribute to the literature evaluating ex post whether carbon leakage has occurred in the Europeanmanufacturing sector due to the introduction of the EU ETS. We focus on the question of leakage through trade ofmanufactured goods, leaving out green paradox type general equilibrium effects of climate policy on energy prices(Jensen et al, 2015). Our definition of carbon leakage thus includes both relocation of production by European firmsas well as imports from foreign firms. Following this definition, carbon leakage should be evaluated at the sectorlevel. A firm-level analysis may only provide evidence on one aspect of leakage, namely relocation. However, evenif no relocation of production by European firms is found, leakage may occur through an increase in market shares ofcompeting firms that produce in regions with no climate policy. Data allowing to disentangle these two effects aretypically not available.MethodsWe proceed based on the insight that for a given level of consumption, carbon leakage leads to a highershare of imports in total value added. Thus, a sector-level approach identifies whether European climate policy hasled to carbon leakage. We examine the evolution of import intensity at the sector level before and after theintroduction of the EU ETS. According to our identification strategy, if there is carbon leakage, then a greater carboncost of embedded emissions (as a share of a sector's revenue) will, on average, lead to an increase of the import sharein European trade. Note that the focus is not on the level of import intensity, often used as a proxy for vulnerability tocarbon leakage, but on changes in import intensity at the sector level over time in response to the carbon costimposed by the EU ETS. We argue that idiosyncratic determinants of a sector's import share (e.g. cost-pass throughability, exchange rates, transport costs, technological change) are uncorrelated with the sectors' ordering according tothe effect of carbon-cost intensity on changes in import intensity. In our estimations we additionally control forsector-level fixed effects and trends to isolate the cross-sectoral effects from sector-specific events. We reject thehypothesis of no carbon leakage in case of a systematic link between the sector-level relative carbon cost and thechange in the import share.In our analysis we use sector-level European external trade and production data from Eurostat's Comextdatabase at the sectoral PRODCOM level, which we map into four-digit NACE codes. Emission intensity is availableat PRODCOM level from the benchmarking established by the Commission. This benchmark represents for eachsector the average of the 10% most carbon-efficient firms. Price data for European carbon permits is readilyavailable from trading platforms such as the ICE.ResultsBased on preliminary results we do not find indications for carbon leakage. In our data, the cost of embedded carbonvaries between 0.1% and 50% of the final product price, so that there is sufficient “treatment” variation. However,across different measures – ranging from correlations, over rank-correlations to regressions with sector fixed-effects– we do not find any statistical link between the evolution of import-intensity and the relative carbon-cost induced bythe EU ETS. This result is not compatible with the hypothesis of carbon leakage: even if only some sectors suffercarbon leakage whil some other sectors remain unaffected, the average correlation should be positive.ConclusionsBased on our analysis we conclude that the EU ETS has not induced carbon leakage in the European manufacturingsector, even though it is predicted by theory and ex ante simulation analyses. This suggests that barriers againstcarbon leakage exist which are not accounted for in the more stylized ex ante models.