Overview

The code in this replication package conducts the statistical analysis and produces the results presented in Linsenmeier et al. 2023. The package consists of one script for the empirical statistical analysis written in Stata, a program for the Monte Carlo simulations written in C++, and several scripts for visualisation of the results written in Python 3.

Data Availability

All data are publicly available at no cost.

Data on carbon pricing policies:

• Carbon Pricing Dashboard of the World Bank: https://carbonpricingdashboard.worldbank.org/
• World Carbon Pricing Database: https://github.com/g-dolphin/WorldCarbonPricingDatabase

Data on country characteristics:

• World Development Indicators of the World Bank (WDI): https://databank.worldbank.org/source/world-development-indicators
• World Governance Indicators (WGI): https://info.worldbank.org/governance/wgi/
• Greenhouse gas emissions (Minx et al. 2021): https://doi.org/10.5281/zenodo.5566761
• Reserves of fossil fuels from the Energy Intelligence Agency (EIA): https://www.eia.gov/
• Global Debt Database (GDD): https://www.imf.org/external/datamapper/datasets/GDD
• Government Finance Statistics (GFS): https://data.imf.org/?sk=a0867067-d23c-4ebc-ad23-d3b015045405
• Expenditure by Function of Government (COFOG): https://data.imf.org/?sk=5804c5e1-0502-4672-bdcd-671bcdc565a9
• Democracy Index (Polity 5): https://www.systemicpeace.org/polityproject.html
• Public belief in climate change (Gallup): https://news.gallup.com/poll/117772/awareness-opinions-global-warming-vary-worldwide.aspx

Computational requirements

Software Requirements

• Stata 17
  • estout
• Python 3.10.9
  • numpy 1.23.4
  • pandas 1.5.1
  • scipy 1.10.0
  • statsmodels 0.13.5
  • geopandas 0.12.0
  • newtorkx 3.0
  • matplotlib 3.6.1
  • seaborn 0.12.0
• C++ compiler (for simulations)

The file requirements.txt lists these dependencies, please run pip install -r requirements.txt as the first step. See https://pip.readthedocs.io/en/1.1/requirements.html for further instructions on using the requirements.txt file.

Memory and Runtime Requirements

Approximate time needed on a standard (2023) desktop machine:

• empirical analysis : 1 hour
• simulations: 7 days (can be shortened by reducing the number of Monte Carlo simulations)

Description of individual scripts

• stata_all.do: This script estimates all proportional hazard models and stores the results (estimated coefficients and predicted effects) in the folder results.
• p01_make_latextables.py: This script uses the estimated coefficients and produces all regression tables shown in the paper and SI.
• p02_examine_nonlinearities.py: This script visualises the predicted effects of the non-linear models and then fits an inverse hyperbolic sinus to the model with cubic splines.
• p03_quantify_emission-reductions.py: This script uses the results of the Monte Carlo simulations and quantifies the direct and indirect emission reductions from policy diffusion.
• p04a_visualise_emission_reductions.py: This script visualises the direct and indirect emission reductions.
• p04b_visualise_centrality.py: This script calculates network centrality measures for all countries, regresses indirect emission reductions on those measures, and visualises the statistical associations with scatter plots.
• p04c_visualise_coverage.py: This script visualises the results of the Monte Carlo simulations including the sensitivity analysis in terms of the share of countries/global emissions with a carbon pricing policy for scenarios with and without policy diffusion.
• p04c_visualise_effectivenes.py: This script visualises the indirect emission reductions for different assumptions about the effectiveness of carbon pricing policies.

License for Code

The code in this repository is licensed under a CC-BY-NC license.

Instructions to Replicators

• Run all scripts in the order indicated by the file names (i.e. p01, p02, p03, ...). This can also be achieved with the Makefile in the repository (make clean; make all).
• Some of the scripts store intermediate results in the folder results.
• Once all scripts have finished, all tables and figures can be found in the respective folders tables and figures.
• For the simulations, see the separate Makefile in the folder simulations.