Import competition and firm‐level CO emissions: Evidence from the German manufacturing industry

Lehr, Jakob Article — Published Version Import competition and firm‐level CO emissions: Evidence from the German manufacturing industry Canadian Journal of Economics/Revue canadienne d'économique Provided in Cooperation with: John Wiley & Sons Suggested Citation: Lehr, Jakob (2025) : Import competition and firm‐level CO emissions: Evidence from the German manufacturing industry, Canadian Journal of Economics/Revue canadienne d'économique, ISSN 1540-5982, Wiley, Hoboken, NJ, Vol. 58, Iss. 2, pp. 747-770, https://doi.org/10.1111/caje.70003 This Version is available at: https://hdl.handle.net/10419/323862 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. http://creativecommons.org/licenses/by/4.0/ 747 Import competition and firm-level CO2 emissions: Evidence from the German manufacturing industry Jakob Lehr University of Mannheim Abstract. Using the German census of the manufacturing industry, I analyze the impact of import competition on carbon emissions per unit of deflated sales (emission intensity). I combine precise information on firm-level CO2emissions with sector-level trade flows. Looking at the period 1995 until 2017, I focus on the impact of the rise of Eastern Europe and China while addressing the endogeneity of trade flows with an instrumental variable approach. The baseline results suggest that a 1 pp increase in the import penetration ratio caused a reduction of the average firm’s emission intensity by approximately 0.3%. This result implies that the rise of the joint East between 1995 and 2017 kept the average firm’s emission intensity 6% below the level it would have had in the absence of the East’s rise. I do not find strong indication for reallocation of production towards more efficient firms. Finally, I supplement the analysis by examining the effect of export opportunities due to the East’s rise. The results indicate that exporting to the East increased sales and emissions, with a small, if any, negative effect on emission intensities. Résumé. Concurrence des importations et émissions de CO2des entreprises : l’exemple de l’industrie manufacturière allemande. En utilisant le recensement allemand de l’industrie manufacturière, j’analyse l’incidence de la concurrence des importations sur les émissions de carbone par unité de ventes en termes réels (intensité des émissions). Je combine des données précises sur les émissions de CO2des entreprises avec les flux des échanges commerciaux sectoriels. En examinant la période allant de 1995 à 2017, je me concentre sur l’effet de la montée en puissance de l’Europe de l’Est et de la Chine tout en tenant compte de l’endogénéité des flux commerciaux à l’aide de la méthode des variables instrumentales. Les résultats de base suggèrent qu’une augmentation d’un point du taux de pénétration des importations entraîne une réduction de l’intensité des émissions de l’entreprise moyenne d’environ 0,3 %. Ce résultat indique que la montée en puissance des pays de l’Est entre 1995 et 2017 a maintenu l’intensité des émissions de l’entreprise moyenne 6 % en dessous du niveau qu’elle aurait atteint en l’absence de celle-ci. Je ne constate pas d’indication forte d’une redistribution de la production vers des entreprises plus efficaces. Enfin, je complète l’analyse en examinant l’effet des possibilités d’exportation dues à la montée en puissance de l’Est. Les résultats indiquent que les exportations vers l’Est ont augmenté les ventes et les émissions, avec un effet négatif faible, voire nul, sur l’intensité des émissions. JEL classification: F18, Q54, L60, D22 Corresponding author: Jakob Lehr, [email protected] I thank Katrin Rehdanz and Robert Gold for their helpful comments. The paper also benefitted from discussions at the 15th RGS Doctoral Conference in Economics and the 56th Canadian Economics Association Conference. The paper improved significantly during the review process thanks to the valuable feedback from two anonymous referees and the editor, Jevan Cherniwchan. I acknowledge financial support provided by the Leibniz Association through the Kiel Centre for Globalization (Grant SAS-2016-IfW-LWC) and funding by the German Research Foundation (DFG) through CRC TR 224 (Project B07). I hereby declare that no actual or potential competing interests exist. Any errors are my own. Open Access funding enabled and organized by Projekt DEAL. Canadian Journal of Economics / Revue canadienne d’économique 2025 58(2) May 2025. / Mai 2025. 25 / pp. 747–770 / DOI: 10.1111/caje.70003 c The Author(s). Canadian Journal of Economics/Revue canadienne d’économique published by Wiley Periodicals LLC on behalf of Canadian Economics Association. This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. 748 J. Lehr 1. Introduction How trade and globalization affect environmental performance and, in particular, climate change is an important and widely discussed topic (see Copeland and Taylor 2004, Cherniwchan et al. 2017). While older studies have mostly focused on country or sector-level effects (see Copeland and Taylor 1994, Cole and Elliott 2003), more recent works have emphasized the importance of the underlying firm-level response to trade and globalization (see Cherniwchan 2017, Barrows and Ollivier 2018, Gutiérrez and Teshima 2018, Forslid et al. 2018). For instance, this research has explored the role of foreign direct investment on firms’ energy use (Brucal et al. 2019) or the effect of firms’ exporting status on CO2intensity of production (Richter and Schiersch 2017). However, little is known about how import competition affects firms’ CO2emissions and emission intensity. My paper addresses this gap by analyzing the effect of import competition on CO2emissions per unit of sales in the German manufacturing industry. The role of competition in general and import competition in particular on firm-level productivity has received much attention in the literature. For example, Schmidt (1997) argues that increasing competition threatens firms’ survival and thus forces managers to reduce slack. Indeed, previous empirical research that looks at firms in Europe has established a positive link between import competition and productivity and innovation (see Holmes and Schmitz 2001, Bloom et al. 2015, Shu and Steinwender 2019, Chen and Steinwender 2021). Because the environmental and energy economics literature has identified energy use as particularly inefficient (“energy efficiency paradox”; see DeCanio 1993), improvements in energy efficiency might appear as a “low-hanging fruit” from the perspective of a manager who needs to cut costs to ensure the firm’s survival. Therefore, I expect fierce competition to affect the CO2intensity of production negatively. For the empirical analysis I combine the German census of the manufacturing industry with sector-level trade flows. The census data span the period from 1995 until 2017, cover the universe of manufacturing plants with more than 20 employees (approximately 40,000 plants annually) and provide, among other things, detailed information on plant-level fuel use. This information allows calculating CO2emissions based on fuel-specific conversion factors.1To identify the effect of import competition on emission intensity, I exploit the rise of China and Eastern Europe (which I will jointly refer to as “the East”) as major actors in the world economy. A rising share of imported manufacturing goods in Germany originating from these regions is one manifestation of this process.2 To uncover causal effects, I exploit the across sector variation in exposure to imports from the East, and I address the endogeneity of imports with an instrumental variable approach following Autor et al. (2014) and Autor et al. (2020). The paper relates to several strands of literature. First, it contributes to the literature on trade, globalization and the effects on the environment.3By examining the effect of 1 In recent years, the German manufacturing sector emitted approximately 200 million tons of CO2annually, roughly one quarter of total emissions in Germany, absorbed more than 15% of Germany’s labour force and contributed approximately one quarter to Germany’s gross domestic product. These figures reflect the central role of the manufacturing sector in the German economy. 2 For example, between 1995 and 2017, the share of German imports from the East rose from 10% to almost 30% (see table A6 in the online appendix). 3 See Cherniwchan (2017) and Copeland et al. (2021) for recent literature reviews. Import competition and emissions 749 import competition on CO2emission intensity in Germany, it closely relates to Gutiérrez and Teshima (2018), who found that intensified import competition due to tariff cuts improved energy efficiency among firms in Mexico. It also complements recent work by Leisner et al. (2023), who document that import competition from China lowered emissions and sales among Danish manufacturing firms but did not reduce emission intensities except in the most emission-intensive quartile. My paper extends this literature by focusing on manufacturing in Germany—Europe’s largest economy, its largest emitter and a country with a relatively energy-intensive industrial structure.4Leisner et al. (2023) also analyze the effects of offshoring, i.e., importing intermediate inputs. Similar to Akerman et al. (2021) and Dussaux et al. (2023), who examine Swedish and French manufacturing firms, respectively, they find that offshoring leads to lower emission intensities by increasing sales more than emissions.5In related work on the effects of trade liberalization through NAFTA, Cherniwchan (2017) attributes reduced emissions of local pollutants from US manufacturing plants to the increased availability of emission-intensive intermediate inputs and new export opportunities for American manufacturers. The relation between exporting and firms’ CO2emission intensity has been studied by Richter and Schiersch (2017) and Barrows and Ollivier (2021) for Germany and India, while Kong et al. (2022) and Bombardini and Li (2020) examine the impact of exporting on local pollutants in China. The paper further relates to the literature on the determinants of energy efficiency and the so-called “energy efficiency paradox” (see DeCanio 1993, Jaffe and Stavins 1994, Gerarden et al. 2017), which refers to seemingly suboptimal energy use by firms. The literature has identified several potential explanations for inefficient energy use, such as managerial inability (see Bloom et al. 2010, Martin et al. 2012), capital constraints (see Levine et al. 2018, De Haas et al. 2021) or market conditions such as size (Forslid et al. 2018). By linking changes in the level of competition to changes in CO2intensity, I investigate a further determinant of energy efficiency. More broadly, my paper contributes to the literature investigating the effect of import competition from China and Eastern Europe on the manufacturing sector in western industrialized countries.6For instance, Bloom et al. (2015) document “trade induced technical change,” i.e., technological upgrading, more patenting and higher total factor productivity (TFP) revenue among European firms that operate in sectors more exposed to Chinese imports. Chen and Steinwender (2021) also find positive effects of import competition on productivity among initially unproductive family-owned firms in Spain.7In contrast, Autor 4 For example, during the first three trading periods of the EU–ETS (2005–2020) approximately 20% of verified emissions from industrial installations came from Germany (see European Environmental Agency 2023). 5 While this is an arguably important channel through which trade integration can affect German firms, I cannot study offshoring because I lack relevant information on imports. 6 An abundant literature studies the labour market consequences, both on the regional and individual level, e.g., Autor et al. (2013), Autor et al. (2014)andAcemogluetal.(2016)look at the US and Dauth et al. (2014) analyze the case of Germany. These papers document negative effects of import competition on employment. The China shock appears to be most relevant for the US, whereas for Germany the rise of Eastern Europe was more critical. 7 They provide evidence that competition forces unproductive firms to improve material usage and eliminate what they call “X-inefficiencies,” which describes suboptimal firm behaviour regarding the objective to maximize monetary profits. 750 J. Lehr et al. (2020) estimate a negative effect of increasing import competition from China on R&D expenditure and patenting among US manufacturing firms. Indeed, the literature on import competition and innovation summarized by Shu and Steinwender (2019) finds “largely positive evidence for such [import competition increasing innovation] in Europe and mixed evidence for such in Northern America.” I start the analysis with a decomposition of the three-digit sector-level emission intensity (see Olley and Pakes 1996), showing that total sector-level emission intensity is negatively related to imports from the East, driven by within-firm changes. Reallocation of market shares towards more productive firms, if at all, played only a minor role. The main firm-level analysis confirms the result from the sectoral decomposition. Baseline estimates imply a decrease of emission intensity in the range of 0.3% to 0.4% inresponsetoa1ppincreaseintheshareofimportsfromtheEastrelative to baseline absorption (import penetration ratio). By examining firms’ CO2emissions and sales separately, I show that both decline due to increased import competition, with a stronger response in emissions, resulting in a fall in emission intensity. The effect is centred on firms with above-median emission intensity—underlining the relevance of the effect for aggregate emissions—and among firms operating in sectors with low import penetration in 1995. The results are robust to alternations of the regression specification and to “controlling” for changes in sectoral exports to the East, for which I constructed an instrument analogous to the one for imports. Respective estimates suggest that new export opportunities lead to a nearly proportional increase in emissions and sales. I further analyze the effect of import competition on emissions per unit of value added, which can be calculated for a subsample of firms yielding quantitatively similar estimates. The baseline results are also robust to an alternative identification strategy based on gravity residuals presented in the paper’s online appendix. The remainder of the paper is structured as follows. In section 2, I outline the empirical approach. Section 3introduces the data set, shows descriptive statistics and provides first results at the sectoral level to motivate the main analysis. The main results from the firm-level analysis are presented in section 4together with robustness checks and effect heterogeneities. Section 5presents additional results to complement the main analysis, and finally, section 6discusses the findings and concludes. 2. Empirical approach To estimate the effect of import competition on firm-level outcomes consider the following regression specification: yitz =β0+αIPREast zt +νi+itz.(1) The dependent variable yitz can be any outcome of firm iin year toperating in sector z. The coefficient of interest αcaptures the effect of an industry’s exposure to imports from the East defined as total imports from the East in year tscaled with initial absorption (see Autor et al. 2020). Concretely, the “import penetration ratio” (IPR) is defined as follows: IPREast zt ≡ImpEast zt Yz,1995 +Impz,1995 −Expz,1995 . Finally, itz in equation 1is a random error term. I follow Bloom et al. (2015) by taking long differences (four years) which eliminates the firm fixed effect νi. The differenced equation reads as follows: Δyitz =β0+αΔIPREast zt +Δitz.(2) Import competition and emissions 751 The regressions I take to the data may or may not include additional sets of fixed effects, e.g., year dummies or year by emission intensity decile dummies, to control for annual shocks or shocks occurring along the energy-intensity distribution.8 To estimate a causal effect of the trade exposure of a sector on firm-level outcomes, requires addressing the endogeneity of trade flows. For instance, demand conditions in Germany are expected to affect imports and domestic firms’ behaviour simultaneously.9 I address the endogeneity similar to Autor et al. (2013) and Dauth et al. (2014)by instrumenting the changes in German imports from the East’s industry zwith changes in trade flows from the same industry zin the East to a set of other countries: IPROther←East zt ≡ImpOther←East zt Yz,1995 +Impz,1995 −Expz,1995 . The idea is that part of the variation of German imports from the East is due to a rising comparative advantage of the East or lower trade costs. The instrument is relevant for this part of the variation as the rise of the East also affects trade flows to the other countries. The other (endogenous) part of the variation is due to domestic conditions in Germany and needs to be separated out. The instrument succeeds in separating the exogenous component of trade flows to Germany from the endogenous under the assumption that conditions in Germany are orthogonal to those in the chosen set of other countries. The exclusion restriction further demands that trade flows between the set of other countries and the East have no direct effect on German firms. These considerations guide the selection of an appropriate set of countries for the instrument group. I follow Dauth et al. (2014), who included Australia, Canada, Japan, Norway, New Zealand, Sweden, Singapore and the United Kingdom, all of which are high-income countries but neither directly borders Germany nor is any of them a member of the European Monetary Union (EMU). Dauth et al. (2014) argue that demand conditions among neighbouring countries are too similar and that the fixed exchange rate within the EMU might cause a violation of the exclusion restriction if changes in trade flows between other countries and the East directly affect German industries.10 Finally, for the instrument to work, it needs to be relevant, which, however, can be tested and is indeed confirmed by the first-stage results reported in section 4. 8 Ideally, I would like to use lagged absorption from the period before the rise of the East in the denominator of the instrument. However, due to data limitations this was not possible. The statistical office provides production data at the economic sector level, based on the sector classification from 1993, only since 1995. Before 1995 information on sectoral production is available for the sector classification from 1979. The statistical office could not provide a mapping between the classifications. 9 Suppose the demand for some goods, e.g., heat pumps and solar panels, suddenly increases in Germany. That will simultaneously affect the imports of respective goods and domestic producers of those goods. For example, domestic producers’ production might increase, and hence, they require more energy inputs. Moreover, their energy intensity might also increase because expanding production beyond the efficient level will still be profitable given that high demand drives up prices. 10 A further concern in the context of this application might relate to global energy price shocks simultaneously affecting a specific sector in Germany and exports from the same sector in the East to the instrument countries. Table A8 in the online appendix correlates changes in sectoral energy prices in Germany with export flows from the East to the instrument countries. Conditional on year fixed effects sectoral energy prices in Germany and exports from the East to the instrument countries do not co-move. 752 J. Lehr 3. Data, descriptive statistics and motivating exercises 3.1. Data The main data source is the German census of the manufacturing industry called AFiD (Amtliche Firmendaten für Deutschland), which covers the universe of German industrial plants with more than 20 employees. The data consists of different “modules” of which I combine “AFiD Modul Industriebetriebe” (industrial plants module) with “AFiD Modul Energieverbrauch” (energy use module). The industrial plants module contains economic variables such as sales, sales abroad, number of employees and investment. To deflate these monetary values, I use the four-digit sector-specific producer price indices published by the German statistical office.11 The energy use module details plant-level energy use by fuel type. Energy use is reported in physical units (kWh) and can thus be converted to CO2emissions based on fuel-specific conversion factors (see Petrick et al. 2011, Richter and Schiersch 2017).12 To account for indirect emissions resulting from the generation of electricity that firms buy from the grid, I take the average carbon content of the German electricity mix, which varies by year (see Umweltbundesamt 2018). The sum of direct and indirect emissions is total emissions. A major caveat with the energy data is a break in the reporting between 2002 and 2003. The time series before and after 2003 are both internally consistent. In my estimation, I make sure to exclude variation that results from the break in the reporting by excluding years from the sample in which subtracted lags, used in difference calculations, predate 2003.13 The final dataset is an unbalanced panel aggregated at the firm level, covering 1995 until 2017. I supplement the main data with the “cost structure survey,” an unbalanced panel at the firm level that typically includes firms for four consecutive years.14 The cost structure survey provides—among other things—information on intermediate input expenditures, allowing for an estimation of the effect of import competition on the emission intensity of value added. Unfortunately, more granular information on intermediate inputs, such as by product and origin, is unavailable, making it impossible to study offshoring. I rely on the BACI database for information on bilateral trade flows, which is constructed from the United Nations Commodity Trade Statistics Database (Comtrade) and provided by CEPII. The database reports trade flows at the six-digit product level from the Harmonized System (HS) nomenclature. To merge the trade information to the firm-level data, I aggregate from the product level to the three-digit economic sector level using the classification 11 See “Statistisches Bundesamt, Index der Erzeugerpreise gewerblicher Produkte (Inlandsabsatz) – Lange Reihen der Fachserie 17 Reihe 2.” 12 For the conversion of energy from primary sources, I draw upon the emission factors provided by the Umweltbundesamt (a federal agency). A table with the relevant information is available at www.umweltbundesamt.de/themen/klima-energie/treibhausgas-emissionen, last retrieved November 18, 2020. The table gives the fuel-specific time-varying CO2content per terajoule. This unit can be converted to CO2per kWh. We then multiply the fuel use in kWh with the respective conversion factor to obtain the CO2emissions. 13 In the main specification that uses four-year differences, this implies the omission of the years 2003–2006. For further background on the energy use module as well as the change in reporting, see Petrick et al. (2011). 14 This sampling procedure also dictates the choice of differencing, making it infeasible to take differences longer than four years. Only firms with more than 500 employees are included annually in the cost structure survey. Import competition and emissions 753 from 1993 (equivalent to NACE industry codes).15 I fix firms’ economic sector to the sector from the first year in which the firm appears in AFiD. The final data set is cleaned from outliers. Specifically, I drop firms with either CO2emission intensity, the four-year change in CO2emission intensity or sales below the 1st percentile or above the 99th percentile. I also omit the economic sectors “manufacture of office machinery and computers” (WZ 30) and “manufacturing of radio, television and communication equipment and apparatus” (WZ 32). Over the period 1995 to 2017, both sectors have been subject to rapid technological changes, quality upgrading and falling prices. Because of this development, the producer price indices in these sectors are outliers making a comparison of deflated sales as measures of physical output over time difficult.16 In a robustness check I show results obtained from a sample that includes both sectors. 3.2. Descriptive statistics Table 1shows summary statistics from the estimation sample for firm-level variables pooled over the period 1995 until 2017. The average firm in the data has close to 116 employees and generates approximately 23 million euros in annual turnover, of which—on average—18% are exported. Additional information on value added from the cost structure survey is available for about 40% of the observations. The average firm’s value added amounts to 24 million euros, more than the average sales in the primary dataset, reflecting the oversampling of large firms in the cost structure survey. TABLE 1 Descriptive statistics – Firm-level information Variable Mean Std. dev. p10 p50 (median) p90 N Number of employees 116 188 24 54 255 744,062 Sales 22,923.2 53,710.3 1,729.5 6,705.33 51,559.5 744,062 Export share 0.18 0.24 0 .07 0.56 744,062 Value added 23,726.5 56,173 1,668.8 7,398.2 58,492.91 297,241 Value added share 0.60 0.17 0.37 0.61 0.82 297,241 Total energy (in MWh) 11,048.7 103,877.5 155.8 933.5 12,400 744,062 Total CO2emissions (in t) 3,722.4 31,634.2 65.9 381.7 4,823.46 744,062 Total electricity (in MWh) 3,619.55 29,104.7 64.4 423.3 5,351.5 744,062 Import penetration ratio 8.85 9.94 1.04 5.26 20.44 744,062 NOTES:The table shows the average of respective variables from the period 1995–2017. Deflated sales and value added are in 1,000 euro, energy use (total and electricity) is in MWh and CO2emissions in tons. SOURCES:Research data centres of the Federal Statistical Office and the Statistical Offices of the Länder: AFiD-Panel Industriebetriebe, 1995–2017; own calculations. 15 To be precise, I first convert the product-level information from HS92 to SITC3 (conversion table was downloaded from https://unstats.un.org/unsd/trade/classifications /correspondence-tables.asp) and then I map from SITC3 to the three-digit industry classification using the same mapping as Dauth et al. (2014). The industry classification from 1993 was in place until 2008 with minor modifications in 2003. Therefore, I omit all firms from the analysis that were first observed only after 2008 because the economic sector based on the classification from 1993 is unknown for these firms. 16 The average producer price indices in the manufacturing sector increased moderately from 94 in 1995 to 107 in 2017 (indexed to 100 in 2010), while the average producer price indices in WZ 30 collapsed from 307 to 81 and in WZ 32 from 217 to 89 in the same period. 754 J. Lehr In addition to presenting economic performance indicators, table 1summarizes total energy use, electricity use and CO2emissions. For instance, the table shows that mean energy use amounts to 11,048 MWh annually, resulting in 3,722 tons of CO2emissions per firm. Finally, the last row of the table shows the average firm’s import penetration ratio from the East at approximately 9 pp. This average masks strong dynamics over time and differences across sectors as can be seen from table A6 in the online appendix, which reports summary statistics for imports (shares) from China and Eastern Europe from 22 two-digit industries for 1995 and 2017. In the initially least-exposed sector (Manufacture of tobacco products, NACE Rev.1 16), imports from the East accounted for merely 1.4% in 1995 but for 27% in the most exposed sector (Manufacture of wearing apparel; dressing and dyeing of fur, NACE Rev.1 18). Twenty-two years later, import shares ranged from 11% (manufacture of chemicals and chemical products, NACE Rev. 1 24) to 58% (manufacturing of office machinery and computers, NACE Rev.1 30).17 China’s accession to the WTO and the EU’s eastward enlargement played an important role in the increase in import shares, which indeed increased in all 22 sectors during the time window under consideration. 3.3. Motivating exercise: Sectoral decomposition of emission intensity Before turning to the firm-level analysis, I describe the evolution of aggregate sectoral emission intensity from 1995 until 2017. Following Olley and Pakes (1996) and Brucal et al. (2019), I decompose aggregate sector-level emission intensity in an unweighted mean and a covariance term. The latter captures the association between firms’ market shares and their emission intensities. The following expression describes the decomposition: Wzt = i∈Z sitlnEit   Weighted CO 2Intensity in Sector Z =lnEzt  Unweighted avg. Intensity + i∈Z (sit −st)(lnEit −lnEt)   Covariance ,(3) where sit is the share of sales by firm iin total sales in sector z(is market share) at time t,lnEzt is the average emission intensity from all firms in sector zand stis the average market share in sector z. The aggregate weighted emission intensity in sector z(Wzt) can be re-written as the average emission intensity from all firms in sector z(lnEzt) and the covariance term. A negative covariance implies higher market shares for more carbon-efficient firms (the inverse of carbon intensity) and thus reflects a more efficient allocation of economic activity across firms. Changes in this term capture a reallocation of market shares across firms with different carbon efficiency, e.g., a decrease implies a reallocation of market shares towards firms with a lower carbon intensity (higher carbon efficiency). Changes in the unweighted average emission intensity reflect changes within firms. Figure 1shows the evolution of the weighted average, the unweighted average and the covariance term averaged across three-digit economic sectors. One can see that the weighted average decreased by approximately 40% between 1995 and 2017 as indicated by the solid 17 Aggregated across the sectors, imports from China amounted to approximately US$8 billion (approximately 22*350 MIO USD) in 1995, corresponding to an import share of 2%. Until 2017, this share rose to 10% (imports worth US$89 billion and approximately US$4 billion from the average two-digit sector). Similarly, imports from Eastern Europe rose from US$32 billion (approximately 8% of total imports) in 1995 to US$200 billion in 2017, corresponding to an import share of nearly 20%. Import competition and emissions 761 TABLE 4 Effects of import competition and export opportunities IV OLS (1) (2) (3) (4) PanelA.LogofCO 2emission intensity Δimports −0.0026*** −0.0029*** −0.0012*** −0.0001 (0.0007) (0.0009) (0.0005) (0.0005) Δexports −0.0008 0.0037*** −0.0034*** −0.0013** (0.0010) (0.0012) (0.0005) (0.0005) Number of observations 403,367 403,367 403,367 403,367 Kleibergen–Paap F-statistic 106.78 60.91 PanelB.LogofCO 2emissions Δimports −0.0079*** −0.0027*** −0.0024*** 0.0004 (0.0012) (0.0008) (0.0005) (0.0004) Δ exports 0.0041*** 0.0033*** 0.0024*** 0.0022*** (0.0009) (0.0010) (0.0005) (0.0005) Number of observations 403,367 403,367 403,367 403,367 Kleibergen–Paap F-statistic 106.78 60.91 Panel C. Log of gross output Δimports −0.0053*** 0.0001 −0.0011 0.0005 (0.0013) (0.0011) (0.0007) (0.0007) Δ exports 0.0049*** −0.0004 0.0058*** 0.0035*** (0.0013) (0.0014) (0.0007) (0.0007) Number of observations 403,576 403,576 403,576 403,576 Kleibergen–Paap F-statistic 106.68 60.90 CO2intensity–decile–year dummy Yes Yes Yes Yes Sales-decile-year-dummy Yes Yes Yes Yes Export share–decile–year dummy Yes Yes Yes Yes Sector dummy Yes Yes NOTES: *p<0.1, ** p <0.05, *** p <0.01. Columns (1) and (2) show results from 2SLS estimations and columns (3) and (4) from an OLS estimation. The dependent variable is the four-year change in the log of firms’ CO2emission intensity (panel A), the log of firms’ CO2emissions (panel B) and the log of firms’ deflated sales (panel C). The explanatory variables are the four-year changes in the sectoral IPR from the East and changes in sectoral exports from Germany to the East scaled with domestic absorption in 1995. The instruments are the four-year changes of exports from the East to other countries (to instrument changes in German imports) and the sectoral change in exports from other countries to the East (to instrument the German exports). Standard errors were clustered both at the firm and at the three-digit industry–year level. SOURCES: Research Data Centres of the Federal Statistical Office and the Statistical Offices of the Länder: AFiD-Panel Industriebetriebe, 1995–2017, own calculations. emissions (panel B). Surprisingly, the point estimate for the effect of export opportunities on emission intensity is positive due to a positive effect on emissions but a null effect of export opportunities on output (panel C).22 In addition to the main goal of the analysis presented here—to control for export opportunities as a potential confounder—the findings about the impact of export opportunities on firm-level emissions are of significance by themselves. Section A3 in the online appendix 22 An alternative is to simply control for firms export shares in the baseline IV estimation, with the caveat that this does not lend to a causal interpretation of the export share coefficient. From table A9 in the online appendix, one can see that a 1 pp increase in firms’ export share is associated with a reduction in emission intensity by 0.23%, driven by an increase in sales and a less-than-proportional increase in emissions. The main effect on the effect of import competition on emission intensity remains unchanged. 762 J. Lehr expands on this analysis and further analyzes the effect of export opportunities on emission intensity. 4.2.3. Energy prices as potential confounder A concern briefly touched on in the introduction to the empirical approach refers to the exclusion restriction in the particular context of this analysis. If shocks in the global oil and/or gas market simultaneously affect the instrument countries’ imports of goods from the East’s sector zand the energy intensity of production in sector zin Germany, the exclusion restriction would be violated. Information about total energy expenditure in the cost structure survey combined with physical energy use allows for constructing a measure for sectoral energy prices. To estimate the effect of import competition on emission intensity conditional on energy prices, I include changes in the log of average energy prices at the three-digit sector level as a control variable. Table A10 in the online appendix reports the results, which show that the primary effect of imports on emission intensity remains unchanged conditional on sectoral price changes. The results further indicate a negative correlation between changes in sectoral prices and emission intensity, as expected. 4.3. Heterogeneity The subsequent paragraphs shed light on heterogeneities. First, I distinguish between imports from Eastern Europe and imports from China. Second, I analyze heterogeneities depending on firm characteristics, e.g., firms’ emission intensity, export share or size. 4.3.1. Eastern Europe vs. China Table 5reports separately the effects of changes in the IPR from China (columns (1) and (2)) and Eastern Europe (columns (3) and (4)). As in the main results table, the dependent variables are the logs of emission intensity (panel A), emissions (panel B) and sales (panel C). One can see that the effect of imports from China closely resembles the main effect reported in table 3. The coefficient in column (1) indicates a decrease in emission intensity by 0.37% due to a 1 pp increase in the IPR. This effect drops to 0.18% after accounting for sectoral trends (column (2)). Panels B and C show that the effect on CO2emissions is negative and significant in both specifications, while the effect on sales is negative and significant in column (1) but becomes insignificant in column (2). For Eastern European imports, I find a large and highly significant negative effect on emission intensities in both specifications. However, these effects must be qualified: first, the first stages are weak, with F-statistics below 10. Second, when examining the effect on sales in panel C, I find positive point estimates, which become larger after conditioning on industry trends.23 Dauth et al. (2014) emphasize the role of intra-industry trade between Germany and Eastern Europe and the correlation between industry-level imports from and German exports to Eastern Europe. Therefore, the positive effect on sales, which reduces emission intensity, might partly result from correlated German exports. The integration of Eastern European economies into German firms’ value chains, for example, through FDI from Germany, further challenges the identification. For instance, productivity improvements 23 Specifically, the point estimate without industry trends is 0.0015 with a standard error of 0.0049. After conditioning on industry trends, the point estimate is 0.0085 with a corresponding standard error of 0.0049, making it marginally significant. The point estimate with CO2emissions as the dependent variable is −0.0046 without and −0.0002 with industry trends. Import competition and emissions 763 TABLE 5 Eastern Europe and China separately China Eastern Europe (1) (2) (3) (4) PanelA.LogofCO 2emission intensity Δimports −0.0037*** −0.0018* −0.0062*** −0.0088** (0.0010) (0.0011) (0.0024) (0.0040) Number of observations 403,367 403,367 403,367 403,367 Kleibergen–Paap F-statistic 205.98 182.10 6.88 5.13 PanelB.LogofCO 2emissions Δimports −0.0098*** −0.0026** −0.0046 −0.0002 (0.0015) (0.0009) (0.0040) (0.0046) Number of observations 403,367 403,367 403,367 403,367 Kleibergen–Paap F-statistic 205.98 182.10 6.88 5.13 PanelC.Logofsales Δimports −0.0061*** −0.0008 0.0015 0.0085* (0.0015) (0.0013) (0.0049) (0.0049) Number of observations 403,576 403,576 403,576 403,576 Kleibergen–Paap F-statistic 206.02 182.13 6.88 5.13 CO2intensity–decile–year dummy Yes Yes Yes Yes Sales–decile–year dummy Yes Yes Yes Yes Export share–decile–year dummy Yes Yes Yes Yes Sector dummy Yes Yes NOTES: *p<0.1, ** p <0.05, *** p <0.01. The table show results from 2SLS estimations. The explanatory variable is the four-year change in the IPR from China (columns (1) and (2)) and Eastern Europe (columns (3) and (4)) at the three-digit industry level. The instruments are the respective four-year changes of exports from China/Eastern Europe to other countries. The dependent variables are the four-year changes in the logs of firms’ CO2emission intensity (panel A), CO2emissions (panel B) and sales (panel C). Standard errors were clustered both at the firm and at the three-digit industry–year level. For each specification, I report the Kleibergen–Paap F-statistic. SOURCES:Research Data Centres of the Federal Statistical Office and the Statistical Offices of the Länder: AFiD-Panel Industriebetriebe, 1995–2017; own calculations. in Germany might increase exports from German foreign affiliates in Eastern Europe. Given the weak first stage in the IV specification for Eastern Europe and the limitations mentioned above, the magnitude of the effect of imports from Eastern Europe should be interpreted with caution. 4.3.2. Firm characteristics Table 6examines how the effects vary based on firm characteristics. To do so, I interact the trade shock with above-median-dummies for CO2intensities, export shares and size. Additionally, I interact the trade shock with dummy variables representing single-product firms and firms in sectors with high import shares in 1995.24 Column (1) reports heterogeneities by emission intensity, showing a statistically insignificant main effect and a negative and highly significant interaction effect.25 Despite 24 All dummy variables were determined using first-period values to prevent potential endogeneity issues with the trade shock. The choice of control variables depends on the specific interaction to avoid absorbing too much useful variation. All models are “fully interacted,” meaning that all fixed effects/controls were also interacted with the above-median dummies. 25 I omit the CO2intensity–decile fixed effects because they would absorb all the variation in the interaction effect and condition on start-off period values instead. 764 J. Lehr TABLE 6 Interaction effects (1) (2) (3) (4) (5) Main effect −0.00049 −0.00332*** −0.00258*** −0.00303*** −0.00373*** (0.00074) (0.00095) (0.00084) (0.00080) (0.00120) Interaction – CO2intensity −0.00395*** (0.00079) Interaction – Export intensity 0.00080 (0.00095) Interaction – Size −0.00114 (0.00083) Interaction – Single product 0.00105 (0.00091) Interaction – High import shares 0.00329** (0.00151) Number of observations 403,367 403,367 403,367 370,981 403,367 Kleibergen–Paap F-statistic 56.19 77.07 116.19 70.60 70.77 CO2intensity–decile–year dummy Yes Yes Yes Yes Sales–decile–year dummy Yes Yes Yes Yes Export share–decile–year dummy Yes Yes Yes Yes NOTES: *p<0.1, ** p <0.05, *** p <0.01. The table reports main and interaction effects from 2SLS estimations. The dependent variable is the four-year change in the log of firms’ CO2emission intensity. The fixed effects are reported at the bottom of the table and vary depending on the interaction variable. All fixed effects were also interacted with the “above-median dummy.” The dummies for the interaction effects were determined based on first-year values. The regression in column (1) further includes the start-off period value of the dependent variable. The explanatory variable is the four-year change in the sectoral IPR from the East instrumented with the respective four-year change of exports from the East to other countries. Only single plant firms were used in column (4). Standard errors were clustered both at the firm and at the three-digit industry–year level. SOURCES:Research Data Centres of the Federal Statistical Office and the Statistical Offices of the Länder: AFiD-Panel Industriebetriebe, 1995–2017; own calculations. considerable variation in firms’ emission intensity above the median (see table 1), this result still suggests that the effect of import competition is quantitatively relevant for aggregate emissions. Column (2) shows the interaction between import competition and firms’ integration in the global economy, measured by their export shares. The estimated main effect is negative, and the interaction effect is positive but small and insignificant. For a German firm selling primarily domestically, an increase in IPR might be more consequential compared to one selling in the competitive world market, providing a plausible rationale for the stronger response of firms with below-median export shares. The third column reports the interaction effect between import competition and firm size (measured by the number of employees), indicating a larger effect among larger firms. Column (4) interacts the trade shock with a dummy for single product firms.26 The corresponding estimate provides suggestive evidence for a larger efficiency improvement among multi-product firms. Finally, column (5) interacts the change in import exposure with a dummy for three-digit sectors with high import shares in 1995. One might expect that the competitive environment is less affected by increasing imports from the East in sectors that were already relatively open. Consistent with this idea, the main effect is negative and significant, while the interaction effect is positive. This implies that the reduction in emissions was concentrated among firms in sectors with lower import exposure in 1995. 26 I am using single plant firms only here because I do not know the number of products for multi-product firms. Import competition and emissions 765 When interpreting the heterogeneity presented above, account should be taken of the fact that part of the variation is driven by across sector differences. For instance, the most energy-efficient steel producer will still emit more CO2per unit of sales than the least-efficient textile company. Thus, the heterogeneity regarding emission intensity speaks more to the relevance of the effect of import competition on emissions for aggregate emissions and less to differential effects depending on firms’ “CO2productivity.” To analyze heterogeneities within three-digit sectors, I split the sample into within-sector quintiles of emission intensity, export shares and size. Figure A1 in the online appendix plots results for the baseline specification (left subfigure) and the specification that accounts for industry trends (right subfigure). Starting with the subsamples formed based on different emission intensity quintiles, one can see that the effect is relatively stable. All point estimates in the left Subfigure are negative and of similar magnitude; only the effect among the firms in the most emission-intensive quintile is slightly larger than the rest. Results from specifications that account for sectoral trends also hint at a slight increase in the effect’s size for more emission-intensive firms—the ones that presumably have the most room for improvement.27 When examining differences in firm size within sectors, there exists some indication for a more significant decline in emission intensities among larger firms, aligning with the results presented in table 6. As for the export share quintiles within sectors, the point estimates are less stable, possibly due to varying numbers of observations within quintiles, e.g., firms with lower export shares appear to have a lower survival rate. Results in figure A1 suggest that firms with higher export shares relative to other firms in their three-digit sector respond stronger to increased import competition. In contrast, the emission intensity of firms with an above-median export intensity showed a muted response (see table 6). 5. Additional results 5.1. Leakage An obvious concern with this analysis is “carbon leakage.” For instance, the increase in imports from the East could be high carbon content inputs in domestic firms’ production process. If these inputs used to be produced by the domestic firms themselves and outsourced once the opening up of the East allowed firms to do so, the emission intensity of sales would decline. To (partially) address this concern, I estimate the effect of import competition on the emission intensity of value added. In the scenario described above, value added would decline, so the emission intensity of value added would remain unchanged or even increase despite a decrease in the emission intensity of sales. Recall that this analysis is feasible only for firms included in the cost structure survey, which oversamples large firms (i.e., it includes firms with >500 employees yearly and additional firms in four-year blocks). Because the difference specification further distorts the sample towards large firms, table 7includes estimates from a specification that controls for time-invariant firm-level unobservables through firm fixed effects (similar to columns (5) and (6) in table A2 in the online appendix), in addition to results from the baseline difference specifications. The baseline results in panel A of table 7indicate that a 1 pp increase in the IPR induces an approximate 0.3% to 0.23% decrease in firms’ emission intensity of value added. In contrast to the main results, conditioning on industry trends leads to a statistically 27 These results should be interpreted cautiously as a larger effect among firms with higher intensities could also be related to mean reversion. 766 J. Lehr TABLE 7 Additional results – Value added IV IV–FE OLS OLS–FE (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) Panel A. CO2intensity of value added Δimports −0.0029*** −0.0025*** −0.0023** −0.0008 −0.0037*** −0.0018** −0.0026*** −0.0003 −0.0034*** −0.0002 (0.0009) (0.0009) (0.0009) (0.0010) (0.0007) (0.0008) (0.0006) (0.0006) (0.0005) (0.0005) Number of observations 75,499 75,499 75,499 75,499 285,581 285,581 75,499 75,499 285,581 285,581 Kleibergen–Paap F-statistic 198.96 201.61 195.03 144.00 568.76 418.40 Panel B. Share of value added Δimports −0.0111 −0.0182 −0.0120 −0.0140 0.0054 0.0075 −0.0292*** −0.0300** −0.0222** −0.0211* (0.0152) (0.0151) (0.0148) (0.0163) (0.0125) (0.0154) (0.0106) (0.0121) (0.0092) (0.0110) Number of observations 75,514 75,514 75,514 75,514 285,581 285,581 75,514 75,514 285,581 285,581 Kleibergen–Paap F-statistic 198.97 201.62 195.08 144.00 568.95 418.24 Panel C. Log material intensity Δ imports 0.0005 0.0006 0.0005 0.0007 −0.0001 −0.0001 0.0009*** 0.0009*** 0.0007** 0.0007* (0.0005) (0.0005) (0.0004) (0.0005) (0.0004) (0.0005) (0.0003) (0.0004) (0.0003) (0.0004) Number of observations 75,419 75,419 75,419 75,419 285,360 285,360 75,419 75,419 285,360 285,360 Kleibergen–Paap F-statistic 198.91 201.56 194.98 143.90 568.97 418.02 CO2intensity–decile–year dummy Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Sales–decile–year dummy Yes Yes Yes Yes Yes Yes Yes Yes Yes Export share–decile–year dummy Yes Yes Yes Yes Yes Yes Yes Yes Sector dummy Yes Yes Yes Yes NOTES: *p<0.1, ** p <0.05, *** p <0.01. Columns (1) to (6) show results from 2SLS estimations and columns (7) to (10) from an OLS estimation. Columns (1) to (4) and (7) and (8) report estimates from the four-year difference specification, while columns (5), (6), (9) and (10) show results from a fixed effects estimation in which each firm received a separate fixed effect for 1995 to 2002 and 2003 to 2017 to accommodate the break in the reporting of energy variables. The dependent variable in panel A is the log of CO2emissions per unit of deflated value added, in panel B the share of value added (i.e., value added divided by sales) and in panel C the log of material intensity. The explanatory variable is the IPR from the East in each three-digit industry instrumented with respective exports from the East to other countries. For each specification, I report the Kleibergen–Paap F-statistic. Standard errors were clustered both at the firm and at the three-digit industry–year level. SOURCES:Research Data Centres of the Federal Statistical Office and the Statistical Offices of the Länder: AFiD-Panel Industriebetriebe, 1995–2017; own calculations. Import competition and emissions 767 insignificant coefficient of −0.0008 (column (4)). The estimates in columns (5) and (6) based on the within estimator yield somewhat larger point estimates and match the main results closely concerning effect size. Not only are these estimates based on a larger f but also the sample is also arguably less distorted towards large firms. To further investigate the role of import competition on the depth of value added, I relate the share of value added over total sales to the IPR. Results are shown in panel B of table 7. A declining share could point to a fragmentation of firms’ value chains. However, the estimates in panel B of table 7do not indicate that import competition from the East contributes to this because all point estimates are small and indistinguishable from zero across specifications.28 Taken together, the results presented in panels A and B do not support the hypothesis that the effect of import competition on the emission intensity of sales results from carbon leakage. Of course, this analysis does not imply that the imports from the East do not also include intermediate inputs. For example, imports from the East could displace imports from other countries. Also imports from the East in the three-digit sector xcould still be inputs in the three-digit sector y. Finally, panel C of the table reports the effect on the log of material intensity, which is closely related to the results in panel B. Again, I do not find evidence for an increase in material intensity, which one would expect if upstream production steps were sourced out. 5.2. Additional outcomes Table A11 in the online appendix shows results for additional outcomes such as emissions per worker, number of employees, investment and measures of firms’ product choice. This analysis aims to understand better how companies adjust to import competition. Based on this, speculations can also be made about the mechanisms underlying the main results on emission intensity. First, and in line with Dauth et al. (2014), I document a negative effect of increasing competitive pressure on the number of employees, e.g., the baseline estimate implies that the average firm’s number of employees decreased by approximately 0.35% in response to a 1 pp increase in the IPR. Like energy inputs, the OLS estimates that capture the association between the number of employees and imports show a positive bias. Despite a negative effect on both production inputs—labour and energy—I find that emissions per employee fall significantly in the baseline specification (see panel B), which provides suggestive evidence for an improvement of energy efficiency relative to potential efficiency improvements in the use of other inputs. Panel C reports the estimates with inverse hyperbolic sine transformed investment as the dependent variable.29 On the one hand, investment to upgrade technology appears as a natural means for firms to increase their energy productivity. On the other hand, in an environment in which competition is becoming more intense and market shares are falling, a general decline in investments would be expected. The point estimates in panel C rather align with the second hypothesis, suggesting a substantial reduction in investment by 1.4% in response to a 1 pp increase in the IPR (relatively imprecisely estimated though). Again, the OLS estimates show a positive bias, e.g., firms invest more when their products are in high demand. Finally, panels D and E focus on firms’ product mix responses. The dependent variables are the number of products (panel D) and an indicator for single-product firms (panel E). 28 The dependent variable is on a scale between 0 and 100. 29 Investment is quite lumpy; other than the log, the inverse hyperbolic sine transformation is defined for zero values. 768 J. Lehr While changes in the market environment have been shown to affect firms’ product mix (see Goldberg et al. 2010), Barrows and Ollivier (2018) highlight the product mix’s role as a determinant of firm-level emission intensity. For example, when firms concentrate their economic activity further on their core competency, this could decrease the firms’ emission intensity. Panels D and E provide some suggestive evidence in line with this idea: The baselineresultsinpanelDsuggestthata1ppincreaseintheIPRdecreases the average firm’s number of products by 0.005 and increases the chance of being a single-product firm by 0.1%. 6. Discussion and conclusion In this paper I analyze the effect of import competition on firm-level emission intensity. To do so, I combine comprehensive firm-level data from the German manufacturing industry with sector-level trade flow. I focus on the rise of China and Eastern Europe between 1995 and 2017. Using an instrumental variable strategy, I provide evidence that increasing import competition is associated with fewer CO2emissions per unit of sales. Within-firm changes drive this effect. I do not find strong indication for between-firm reallocation. The baseline regression specification implies that firms’ emission intensity of production decreases by 0.3% in response to a 1 pp increase in the import penetration ratio. Between 1995 and 2017 the import penetration from the East increased by approximately 20 pp. Hence the point estimate suggests that the increase in competition kept emission intensity about 6% below the level it would have had in the absence of the rise of the East. The results are robust to examining emission intensity of value added and accounting for new export opportunities to the East. The latter analysis indicates that new export opportunities increased sales and emissions, i.e., firms scaled up, with only a small effect on emission intensity. The main results align with parts of the international trade literature, which tends to find positive effects of import competition on European firms’ productivity (see Shu and Steinwender 2019 and Chen and Steinwender 2021). Given that energy consumption is often viewed inefficient (“energy-efficiency paradox”), it appears plausible that improving energy efficiency is an easy cost-cutting measure to stay competitive under tougher conditions. The results further align with findings by Gutiérrez and Teshima (2018), who show that Mexican manufacturing firms’ energy intensity decreased in response to intensifying import competition. More recently, Leisner et al. (2023) find evidence for a negative impact of import competition from China on emissions intensities among Danish firms, but only among the most emission intensive ones. This result fits my heterogeneity analysis, which also showed that the more emissions-intensive companies drive the effect. Against the background of a comparatively emissions-intensive industry in Germany, different average effects can be well explained. The question about potential mechanisms naturally emerges from my results. Throughout the analysis, I find consistent evidence for a negative effect of import competition on sales and employment. Moreover, Subsection 5.2 provides suggestive evidence for firms adjusting their product mix, e.g., a slight increase in the number of single-product firms and a decrease in the number of products in response to an increase in the IPR. These adjustments could reflect firms concentrating on their productive core activities—a plausible explanation for the efficiency improvements (see Barrows and Ollivier 2018). While the estimates on the effect of the trade shock on the emission intensity of value added and the value added share do not support the hypothesis that carbon leakage drives the results, future work using more granular input data could examine this potential channel Import competition and emissions 769 more closely. Similarly, improved access to intermediates could enhance the overall productivity of downstream firms. Indeed, a small number of recent papers (see Akerman et al. 2021, Dussaux et al. 2023, Leisner et al. 2023) have examined the effect of improved access to intermediate inputs on emission intensities and found consistent evidence for a negative effect due to a disproportionate increase in output. From a policy perspective, one key message of this paper is that pro-competitive policy can lead to environmental benefits via more efficient energy use and hence fewer emissions, a message likely to be relevant beyond trade policy. Supporting information Supplementary material accompanies this article. The data and code that support the findings of this study are available in the Canadian Journal of Economics Dataverse at https://doi.org/10.5683/SP3/TAJMH0. References Acemoglu, D., D. Autor, D. Dorn, G. H. Hanson, and B. Price (2016) “Import competition and the great US employment sag of the 2000s,” Journal of Labor Economics 34(S1), S141–S198 Akerman, A., R. Forslid, and O. Prane (2021) “Imports and the CO2emissions of firms,” CEPR discussion paper no. 6090 Autor, D., D. Dorn, G. H. Hanson, G. Pisano, and P. Shu (2020) “Foreign competition and domestic innovation: Evidence from US patents,” American Economic Review: Insights 2(3), 357–74 Autor, D.H., D. Dorn, and G. H. Hanson (2013) “The China syndrome: Local labor market effects of import competition in the United States,” American Economic Review 103(6), 2121–68 Autor, D.H., D. Dorn, G. H. Hanson, and J. Song (2014) “ Trade adjustment: Worker-level evidence,” Quarterly Journal of Economics 129(4), 1799–860 Barrows, G., and H. Ollivier (2018) “Cleaner firms or cleaner products? How product mix shapes emission intensity from manufacturing,” Journal of Environmental Economics and Management 88(C), 134–58 (2021) “Foreign demand, developing country exports, and CO2emissions: Firm-level evidence from India,” Journal of Development Economics 149, 102587 Bloom, N., M. Draca, and J. Van Reenen (2015) “Trade induced technical change? The impact of Chinese imports on innovation, IT and productivity,” Review of Economic Studies 83(1), 87–117 Bloom, N., C. Genakos, R. Martin, and R. Sadun (2010) “Modern management: Good for the environment or just hot air?,” Economic Journal 120(544), 551–72 Bombardini, M., and B. Li (2020) “Trade, pollution and mortality in China,” Journal of International Economics 125, 103321 Brucal, A., B. Javorcik, and I. Love (2019) “Good for the environment, good for business: Foreign acquisitions and energy intensity,” Journal of International Economics 121, 103247 Chen, C., and C. Steinwender (2021) “Import competition, heterogeneous preferences of managers, and productivity,” Journal of International Economics 133, 103533 Cherniwchan, J. (2017) “Trade liberalization and the environment: Evidence from NAFTA and US manufacturing,” Journal of International Economics 105, 130–49 Cherniwchan, J., B. R. Copeland, and M. S. Taylor (2017) “Trade and the environment: New methods, measurements, and results,” Annual Review of Economics 9(1), 59–85 Cole, M.A., and R. J. Elliott (2003) “Determining the trade–environment composition effect: The role of capital, labor and environmental regulations,” Journal of Environmental Economics and Management 46(3), 363–83 Copeland, B.R., J. S. Shapiro, and M. S. Taylor (2021) “Globalization and the environment,” NBER working paper no. 28797 770 J. Lehr Copeland, B.R., and M. S. Taylor (1994) “North-south trade and the environment,” Quarterly Journal of Economics 109(3), 755–87 (2004) “Trade, growth, and the environment,” Journal of Economic Literature 42(1), 7–71 Dauth, W., S. Findeisen, and J. Suedekum (2014) “The rise of the East and the far East: German labor markets and trade integration,” Journal of the European Economic Association 12(6), 1643–75 De Haas, R., R. Martin, M. Muûls, and H. Schweiger (2021) “Managerial and financial barriers to the net-zero transition,” BOFIT Discussion Papers, no. 6/2021 DeCanio, S.J. (1993) “Barriers within firms to energy-efficient investments,” Energy Policy 21(9), 906–14 Dippel, C., R. Gold, S. Heblich, and R. Pinto (2021) “The effect of trade on workers and voters,” Economic Journal 132(641), 199–217 Dussaux, D., F. Vona, and A. Dechezleprêtre (2023) “Imported carbon emissions: Evidence from French manufacturing companies,” Canadian Journal of Economics 56(2), 593–621 European Environmental Agency (2023) “European Union Emissions Trading System (EU ETS) data from EUTL” Forslid, R., T. Okubo, and K. H. Ulltveit-Moe (2018) “Why are firms that export cleaner? International trade, abatement and environmental emissions,” Journal of Environmental Economics and Management 91, 166–83 Gerarden, T.D., R. G. Newell, and R. N. Stavins (2017) “Assessing the energy-efficiency gap,” Journal of Economic Literature 55(4), 1486–525 Goldberg, P.K., A. K. Khandelwal, N. Pavcnik, and P. Topalova (2010) “Multiproduct firms and product turnover in the developing world: Evidence from India,” Review of Economics and Statistics 92(4), 1042–49 Gutiérrez, E., and K. Teshima (2018) “Abatement expenditures, technology choice, and environmental performance: Evidence from firm responses to import competition in Mexico,” Journal of Development Economics 133, 264–74 Holmes, T., and J. Schmitz (2001) “A gain from trade from unproductive to productive entrepreneurship,” Journal of Monetary Economics 47, 417–46 Jaffe, A.B., and R. N. Stavins (1994) “The energy paradox and the diffusion of conservation technology,” Resource and Energy Economics 16(2), 91–122 Kong, D., G. Ma, and N. Qin (2022) “Trade liberalization and firm toxic emissions,” Review of International Economics 30(4), 971–1002 Leisner, J., J. R. Munch, A. T. Nielsen, and G. Schaur (2023) “The impact of offshoring and import competition on firm-level carbon emissions,” IZA discussion paper no. 16556 Levine, R., C. Lin, Z. Wang, and W. Xie (2018) “Bank liquidity, credit supply, and the environment,” NBER working paper no. 24375 Martin, R., M. Muûls, L. B. de Preux, and U. J. Wagner (2012) “Anatomy of a paradox: Management practices, organizational structure and energy efficiency,” Journal of Environmental Economics and Management 63(2), 208–23 Olley, G.S., and A. Pakes (1996) “The dynamics of productivity in the telecommunications equipment industry,” Econometrica 64(6), 1263–97 Petrick, S., K. Rehdanz, and U. J. Wagner (2011) “Energy use patterns in German industry: Evidence from plant-level data,” Jahrbücher für Nationalökonomie und Statistik 231(3), 379–414 Richter, P.M., and A. Schiersch (2017) “CO2emission intensity and exporting: Evidence from firm-level data,” European Economic Review 98, 373–391, ISSN 0014-2921 Schmidt, K.M. (1997) “Managerial incentives and product market competition,” The Review of Economic Studies 64(2), 191–213, ISSN 00346527, 1467937X Shu, P., and C. Steinwender (2019) “The impact of trade liberalization on firm productivity and innovation,” Innovation Policy and the Economy 19, 39–68 Umweltbundesamt (2018) “Entwicklung der spezifischen kohlendioxid-emissionen des deutschen strommix in den jahren 1990–2017,” ISSN 1862-4359