Replication materials: Attributing Climate Change Impacts in the Fiscal Budgets of Local Municipalities

Abstract

This repository contains the R code, processed data, fitted models, and supporting files required to reproduce the tables, figures, and main analyses presented in the associated paper. The analysis constructs a municipality-year panel of Italian local government budget accounts from OpenBDAP/ARCONET for 2017–2023, combines these data with factual and counterfactual climate indicators, estimates municipal fiscal responses to temperature and hydrological hazards, and attributes part of these responses to anthropogenic climate change. Repository contents The replication package includes: 48 R scripts covering data processing, descriptive analyses, econometric estimation, counterfactual attribution, robustness tests, heterogeneity analyses, and alternative specifications. The main processed municipality-year analysis panel used for all reported results. Alternative datasets used for robustness analyses, including alternative SPEI thresholds and a stricter classification of climate-relevant budget items. Pre-estimated expenditure, revenue, and local-projection models used in the attribution analysis. Municipal income, population, hazard, vulnerability, and spatial data required by the analysis. Supporting lookup files and municipal boundary shapefiles. A detailed README.md, LICENSE, and CITATION.cff. The main analysis panel is fused_data_spei2.Rdata. Alternative panels are also provided for the robustness exercises. Reproducing the results Download and unpack replication_package.zip and set the extracted directory as the R working directory. The file fiscalcostscc_attribution_italy/sourcer.R provides the intended order in which the analysis scripts should be run. For reproduction of the results reported in the paper, users can start directly from Stage 4. Stages 1–3 reconstruct the processed analysis panel from large publicly available raw datasets that are not included in the Zenodo deposit. Their output, fused_data_spei2.Rdata, is provided, so the analyses reported in the paper can be reproduced without downloading or rebuilding the original raw inputs. Please note that sourcer.R is intended as a guide to the analysis workflow and should not itself be sourced. Individual scripts are designed to be run interactively and are largely self-contained. Data not included Several large raw or intermediate datasets are not deposited because they are publicly reconstructible and are not required to reproduce the analysis from Stage 4 onward. These include the original OpenBDAP/ARCONET budget extracts, ISIMIP3a factual and counterfactual climate fields, ISPRA hazard rasters, and large intermediate climate-processing files. The repository instead provides the processed datasets and compact lookup files required for reproduction of the reported results. Reproducibility check The archive was tested after extraction into a clean directory without access to the authors' original file paths. Selected descriptive, regression, and attribution scripts reproduced the corresponding manuscript outputs, with numerical results matching the archived versions. Sample period The regression analysis covers 2017–2023, which is the longest period over which the required municipal budget and climate data are jointly available. The climate-change attribution analysis is evaluated over 2017–2021, reflecting the availability of the ISIMIP3a counterfactual climate simulations used to construct the attribution estimates. Software The analysis requires R 4.2 or later. Required R packages and additional system dependencies are documented in the accompanying README. Spatial packages including sf, terra, and raster require GDAL, GEOS, and PROJ. Code repository: https://github.com/giacfalk/fiscalcostscc_attribution_italy Licence: GPL-3.0-only For citation information, please see CITATION.cff included in the repository.