sabman / Leppert-et-al-2021-Replication-Code-Data

This repository contains replication code and data for the paper 'Accounting for Geographic Basis Risk in Heat Index Insurance: How Spatial Interpolation Can Reduce the Cost of Risk'. We use a variety of spatial interpolation methods to reduce geographic basis risk in weather index insurance.

Replication Data & Code for 'Accounting for Geographic Basis Risk in Heat Index Insurance: How Spatial Interpolation Can Reduce the Cost of Risk'

This repository contains R code and data used in our forthcoming paper. Researchers who wish to replicate our results are invited to consult the README-file. We encourage use of our code by others, cite: Leppert D., Dalhaus, T. and Lagerkvist, C. (2021) "Accounting for Geographic Basis Risk in Heat Index Insurance: How Spatial Interpolation Can Reduce the Cost of Risk".

We use weather station observations from NOAA global historical climate network dataset from Illinois and Iowa, 1980 - 2019, available here: https://www1.ncdc.noaa.gov/pub/data/ghcn/daily/readme.txt. To acknowledge the specific version of the dataset used, please cite:

Menne, M.J., I. Durre, B. Korzeniewski, S. McNeal, K. Thomas, X. Yin, S. Anthony, R. Ray, R.S. Vose, B.E.Gleason, and T.G. Houston, 2012: Global Historical Climatology Network - Daily (GHCN-Daily), Version 3. [indicate subset used following decimal, e.g. Version 3.12]. NOAA National Climatic Data Center. http://doi.org/10.7289/V5D21VHZ [access date].

Compute cumulative monthly values of variable 'TMAX' for each station and convert to CDDs > 84 degrees F. Alternatively, we provide the complete data files.

We use county-level survey corn yield data from the U.S. Department of Agriculture NASS QuickStats database available here: https://quickstats.nass.usda.gov/. Alternatively, we provide the complete data file.

To reconstruct our interpolated (Inverse-Distance Weighting, Ordinary Kriging, Regression Kriging) and nearest-neighbor indices, follow the steps below: Download GHCN-Daily data into working directory, and co17_d00.shp from this repository into your RStudio workspace. Alternatively we also provide a file with pre-processed station data: stations_data.RData.

1. Run indices_code.R as instructed which produces the four CDD indices data files.
2. Run contract_code to calculate results presented in the paper.
3. Load results and run EU_RP_diffs.R to calculate relative changes in risk premiums between insurance contracts

To reconstruct the indices using different station sample sizes, change the sample within the indices_code.R script before running.

To replicate plots from the paper: Consult plots.R upon replicating data