To run all code in this directory, do ./0_make_dataset.sh.

Please note, the raw data used here is not publicly available.

• Therefore, only 2_construct_regression_ready_data.do, 3_unit_root_test_and_plot.do, and 4_plot_ITA_other_energy_regimes_timeseries.R in this folder (the codes that use intermediate data as an input) can be run by a user outside of the Climate Impact Lab.
• The intermediate dataset, IEA_merged_long.dta, is outputed by 1_construct_dataset_from_raw_inputs.do, and cannot be run.

climate - code and shapefiles for constructing, cleaning and assembling climate data

coded_issues - encoded issues and documentation used to construct reporting regimes, clean the energy consumption data, and construct the climate data

energy_load - code for cleaning raw energy consumption data

pop_and_income - code for cleaning population and income data

merged - code for cleaning the merged dataset

Because the raw data is currently unavailable, code in climate, energy_load, and pop_and_income cannot be run and is only present for reference.

Codes in this folder accomplish the following tasks:

• Construct an intermediate dataset including population, energy consumption, climate, and income data.
  • /DATA/regression/IEA_merged_long_GMFD.dta:
  • We clean IEA_merged_long.dta in two different ways to produce regression ready datasets for our main specification (Methods Equation 2; Appendix Equation C.4) and robustness models;
• Construct regression ready data:
  • DATA/regression/GMFD_TINV_clim_EX_regsort.dta: used for estimating the excluding imputed data model (Appendix I.2)
  • DATA/regression/GMFD_TINV_clim_regsort.dta: used for estimating the main model
• Save information on each country-year's income and climate covariates, which is used as an input to plotting code
  • /DATA/regression/break_data_TINV_clim_EX.dta: used for plotting output for the excluding imputed data model
  • /DATA/regression/break_data_TINV_clim.dta: used for plotting output for the main model
• Testing for the existence of unit roots in our outcome variable, motivating the need to use first differenced variables in our empirical analysis

1_construct_dataset_from_raw_inputs.do produces IEA_merged_long.dta throught the following steps:

1. Clean and construct population, income, climate and energy consumption datasets
2. Merge population, income, climate and energy data by country and year

• unavailable raw data

• /DATA/regression/IEA_merged_long_GMFD.dta

2_construct_regression_ready_data.do can produce both GMFD_TINV_clim_EX_regsort.dta and GMFD_TINV_clim_regsort.dta through the following steps:

1. Construct reporting regimes and drop data according to encoded data issues
2. Match product specific climate data with product
   • climate is product specific due to the encoded data issues. Please reference this climate/README.md for more information on the topic.
3. Find income spline knot location to model a nonlinear effect of income on energy temperature sensitivity
4. Perform Final Cleaning Steps before first differenced interacted variable construction
   • Classify countries within 1 of 13 UN regions -- these UN regions are used to construct one of the fixed effects used in the analysis
   • Classify countries in income deciles and groups -- merge constructed income groups from (3) into main dataset
5. Construct First Differenced Interacted Variables used in the analysis section

Note: at the top of 2_construct_regression_ready_data.do set the global macro model to TINV_clim to produce regression ready data for the main model and to TINV_clim_EX to produce regression ready data for the excluding imputed data model.

• /DATA/regression/IEA_merged_long.dta

• /DATA/regression/GMFD_TINV_clim*_regsort.dta

As well as producing the regression ready datasets, 2_construct_regression_ready_data.do can produce both break_data_TINV_clim.dta and break_data_TINV_clim_EX.dta. These are intermediate datasets, that are outputted for 3x3 array plotting. These datasests contain covariate information for each country-year, including:

• Income:

  • Decile of overall income distribution of our observations (gpid)
  • Tercile of overall income distribution of our observations (tpid)
  • Income groupings based on location of knot (largegpid_*), note, these vary by product.
    • See the Paper Appendix Section C.3 for discussion of what these knots are.
  • Average values of the long run income covariate, within each CDD tercile (avgInc_tgpid)
  • Maximum values of the long run income covariate within each income group (maxInc_largegpid_other_energy and maxInc_largegpid_electricity)

• Climate

  • Tercile of the distribution of long run CDDs (tpid)
  • Average value of the long run HDD covariate, within each income tercile (avgHDD_tpid)
  • Average value of the long run CDD covariate, within each income tercile (avgCDD_tpid)

Note: at the top of 2_construct_regression_ready_data.do set the global macro model to TINV_clim to produce regression ready data for the main model and to TINV_clim_EX to produce this covariate information for the excluding imputed data model.

• /DATA/regression/IEA_merged_long.dta

• /DATA/regression/break_data_TINV_clim.dta
• /DATA/regression/break_data_TINV_clim_EX.dta

Through this data testing we motivate the first differencing completed in Step 5 of Constructing Regression Ready Dataset

3_unit_root_test_and_plot.do takes the regression ready dataset created in 2_construct_regression_ready_data.do, and tests for the existence of unit roots in the load_pc variable.

• The code implements the tests described in Section Appendix A.1 of the paper.
• The figures outputted are those in the paper as Appendix Figure A.2

4_plot_ITA_other_energy_regimes_timeseries.R takes the regression ready dataset created in 2_construct_regression_ready_data.do, and plots a simple visualisation of the time series for ITALY Other fuels.

• The figure outputted is in the paper as Appendix Figure A.1

• /DATA/regression/GMFD_TINV_clim_regsort.dta

• /OUTPUT/figures/fig_Appendix-A1_ITA_other_fuels_time_series_regimes.pdf
• /OUTPUT/figures/fig_Appendix-A2_Unit_Root_Tests_p_val_hists_electricity.pdf