Innovation and adaptation to tropical cyclones

Folder structure

• code
  • data processing : run the following files in the right order, to process raw data into a final dataset we can use for our regressions
    • 1-compute_country_dist_to_coastline.R : Compute, for selected countries, raster files with distances to the coastline. Generated data is in the data/processed/distance_to_coastline folder
    • 2-tce_dat_surge_affected_pop.R : Load the generated coastline distances, and combine them with the spatially explicit TCE-DAT dataset to get improved indicators for pop/assets exposed (e.g. pop/assets located up to 5km from the coast)
    • 3-build_patent_dataset.do : Combine patent data from Simon into a single Stata dataset with the right indicators (patents for storms, hvi/all, stock/count)
    • 4-make_datasets.py : Combine the processed pop/assets indicators and patents datasets into a single "storm/patent" dataset at a country-year level, that we can use for our regressions
  • regressions script : file do_regression.do
• data
  • raw
  • processed
  • doc

Reproduce results

Installation

A recent version of R should work (I used R 3.6.0), and the following libraries should be installed prior to running the scripts:

sf, readr, tidyverse, raster, rnaturalearth, tmap, rgdal, countrycode, here

A recent version of python 3, with the libraries pandas and numpy installed should work. I used versions 1.0.3 and 1.18.1 respectively (on python 3.8.2).

The regressions made with Stata MP 15.1

Run the scripts

You can run the data processing scripts in the order, you just need the root variable at the beginning of the 3-build_patent_dataset.do file to the absolute path of your project folder.

You should then be able to run the do_regression.do file.