Replication package for Fillon & Guivarch, "Valuing climate subsystems: an application to the Amazon rainforest"
Details for replication are in replication_package.pdf

Data
Most data are in analysis_figures/data/
Some heavy datasets are available only from this Google Drive and should then be placed in analysis_figures/data with these names rainfall, temperature and earthdata_carbon2010
These datasets can also be downloaded from ISIMIP and EarthData

Step 1 - calibration of Amazon dynamics
Step 1A
run analysis_figures/calibration_dynamics/estimation.R
this step uses temperature and precipitation data from ISIMIP 2B
compute for each climate model the coefficient of climate shock

Step 1B
run analysis_figures/calibration_dynamics/calibration.R
this step uses previous coefficients to match expert estimates
three calibration: one benchmark, two counterfactuals

Step 2 - solve optimal intertemporal program
We run 6 specificiations as benchmark:
run 1: deterministic (stochastic=0), without temperature damages, exogenous paths of variables without control
run 2: deterministic (stochastic=0) with temperature damages, without Amazon
run 3: deterministic (stochastic=0) with temperature damages, with Amazon
run 4: stochastic=1 (one risk dT/dS only) with temperature damages, without Amazon
run 5: stochastic=1 (one risk dT/dS only) with temperature damages, with Amazon
run 6: stochastic=2 (two risks dT/dS and dA/dT) with temperature damages, with Amazon
run 7 to 11 : first counterfactual - same specifications as above, but with a different calibration from expert elicitations
run 12 to 16 : second counterfactual - same specifications as above, but with a different calibration from expert elicitations

Step 2A - interpolation of value function
open study.csv and define the run number
run 1 then 2 should be run first because state space for all runs is created at run 2
run meta_run1_interpolation.py

Step 2B - simulation of stochastic paths
only for runs 4 to 6 (for counterfactuals - 9 to 11, 14 to 16)
open meta_run2_simulation.py once chebyshev meta_run1_interpolation.py is done for this run
define the run number
run meta_run2_simulation.py