The paper describes the core WiNDC build stream and an empirical application on
state level carbon leakage. Data and results are constructed as follows:

      1) The core WiNDC database can be constructed by running run.gms in
      submission/windc_build. Preset options need to be changed to generate the
      energy-economy-environment dataset. Set %satdata% to "bluenote" and
      %sdisagg% to "yes". Keep year as 2016 as this represents the most recent
      version of the dataset. Note that reading the raw data files has been
      shown to be sensitive to the local machine's configuration. Because of
      this, all constructed data is included in the model directories for
      replication.

      2) Embodied carbon estimates and plots are reproduced by running run.gms
      in submission/embodiedcarbon. Note that this routine relies on WiNDC
      datasets that have been pre-generated and included in the embodied carbon
      directory. They can also be re-generated using the build directory.

      3) Model results from the energy-economy model are generated by first
      building bilateral trade matrices in the submission/leakage/build
      directory. Run databuild.gms. This will produce both a trade elasticities
      data file in submission/leakage/build/stata which can be used to generate
      elasticities (run submission/leakage/build/stata/stata.do) and the GDX
      file needed to run the model (in
      submission/leakage/model/data/WiNDC_bluenote_2016_leakage.gdx). Model
      results can then be reproduced by running run.gms in the model directory.

** In order to run R and Stata code, the working directory must be changed to
   match the local machine.
