Influence of the Carbon Border Adjustment Mechanism on the imports of Passenger Vehicles to Germany
Description
This dataset accompanies the study “Influence of the Carbon Border Adjustment Mechanism (CBAM) on the Imports of Passenger Vehicles to Germany”, which analyzes the political and economic implications of the EU’s CBAM within the framework of the “Fit for 55” package. The research applies a gravity model to investigate how bilateral trade flows respond to CO₂ pricing, technological efficiency, and institutional governance factors between 2008 and 2024. The empirical results show that CBAM exerts a mild but statistically significant trade-restrictive effect, which intensifies with the CO₂ intensity of exporting countries. This effect is stronger for non-EU nations and weaker in large, densely populated markets that benefit from economies of scale. The dataset combines information from several authoritative sources: Car import data from the German Federal Ministry of Transport (annual bilateral import volumes, 2008–2024); Macroeconomic indicators (GDP per capita, total population, exchange rate, political stability) from the World Bank’s World Development Indicators; Inflation rates from the International Monetary Fund (IMF); CO₂ intensity data from the European Commission’s EDGAR database; Technological efficiency measures from the Global Innovation Index (WIPO), used to construct a custom CBAM impact indicator that accounts for differences in carbon pricing, production efficiency, and embedded emissions; Additional constructed variables include a dummy for EU membership, bilateral geographic distance (CEPII GeoDist dataset), and derived indicators such as population density ratios (BevRho_ij) and relative inflation and governance differences between trading partners. All datasets were harmonized by country and year, cleaned for consistency, and merged into a balanced panel structure covering 47 partner countries across 17 years. Variables were log-transformed or z-standardized where appropriate to reduce heteroskedasticity and ensure comparability. The dataset is provided in Excel (.xlsx) format, consisting of multiple sheets, each representing one of the key variables used in the regression analysis (e.g., GDP, imports, CO₂ intensity, distance, EU membership, etc.). Each sheet contains country–year data in a wide structure, which can be easily converted into a long (panel) format for econometric replication in software such as Python, Stata, or R.
Files
Steps to reproduce
The dataset was compiled from multiple publicly accessible sources and processed using a reproducible analytical workflow in Python. Car import data were obtained from the German Federal Ministry of Transport, while macroeconomic variables such as GDP, population, exchange rates, and political stability originated from the World Bank. Inflation data were collected from the International Monetary Fund (IMF), and CO₂ intensity data from the European Commission’s EDGAR database. To account for technological efficiency and domestic carbon pricing, a custom CBAM impact indicator was constructed by combining CO₂ emission data with the Global Innovation Index (WIPO). All data were harmonized, merged by country and year, and transformed into logarithmic form where appropriate. Missing and extreme values were handled carefully to preserve time-series variation, and all variables were standardized for cross-country comparability. The econometric analysis was implemented using the linearmodels and statsmodels libraries in Python. Several model specifications were estimated to ensure robustness: Random Effects (RE) Model – identified as the most appropriate baseline estimator based on the Hausman test. Feasible Generalized Least Squares (FGLS) – correcting for heteroskedasticity and cross-sectional dependence. Clustered Standard Error Model (CEM) – adjusting for within-country autocorrelation and panel heterogeneity. Mixed Linear Model (MLM) – integrating random country intercepts and year fixed effects. Outlier Sensitivity Test – re-estimating the model excluding the top 5% of CO₂ intensity observations to test robustness. Heterogeneity and Subgroup Analyses – conducted to capture structural differences between (a) EU vs. non-EU countries, (b) high- vs. low-emission groups and (c) high- vs. low-population-density markets. Across all models and subgroups, results consistently showed that the CBAM effect on imports is mild but significant, becoming strongly negative in high-emission countries and negligible or slightly positive in low-emission groups. Non-EU exporters experienced a stronger restrictive effect compared to EU members, while densely populated and institutionally strong countries were less affected due to economies of scale and governance stability. The complete workflow—including data preparation, transformations, and model estimations—is fully reproducible in Python. The corresponding scripts are available from the author upon reasonable request to facilitate transparency and replication.
Institutions
- Nanjing University of Science and Technology School of Economics and Management