R-ARIO: Refined Adaptive Regional Input-Output Model

Stanford Urban Resilience Initiative
Original ARIO program by Stéphane Hallegatte, 2014

MATLAB codebase for quantifying post-disaster economic recovery for a single region using the Refined Adaptive Regional Input-Output (R-ARIO) model. The R-ARIO model extends the original ARIO model (Hallegatte, 2014) with three key enhancements: (i) dynamic reconstruction rates based on sector-specific reconstruction time curves, (ii) explicit modeling of housing losses separate from productive capital losses, and (iii) sector-level modeling and uncertainty quantification of behavioral parameters.

License: GNU General Public License v3.0 (see LICENSE)

Citation

If you use this code in your research, please cite:

Issa, O., Zhu, T., Markhvida, M., Costa, R., and Baker, J. W. (2025). "Refined adaptive regional input-output model: Application to the 2016 Kumamoto Earthquake." Natural Hazards Review, 26(4), 04025050. https://doi.org/10.1061/NHREFO/NHENG-2207

@article{issa2025rario,
  title={Refined adaptive regional input-output model: Application to the 2016 {Kumamoto} Earthquake},
  author={Issa, Omar and Zhu, Tinger and Markhvida, Maryia and Costa, Rodrigo and Baker, Jack W.},
  journal={Natural Hazards Review},
  volume={26},
  number={4},
  pages={04025050},
  year={2025},
  doi={10.1061/NHREFO/NHENG-2207}
}

Authors

• Omar Issa — Dept. of Civil and Environmental Engineering, Stanford University
• Tinger Zhu — Dept. of Civil and Environmental Engineering, Stanford University
• Maryia Markhvida — Dept. of Systems Design Engineering, University of Waterloo
• Rodrigo Costa — Dept. of Systems Design Engineering, University of Waterloo
• Jack W. Baker — Dept. of Civil and Environmental Engineering, Stanford University

Key Features

• Dynamic reconstruction rates: Time-dependent, sector-specific reconstruction rates derived from user-provided reconstruction time curves, replacing the constant rate assumption in the original ARIO model.
• Explicit housing losses: Housing damage is modeled as a distinct sector that generates reconstruction demand without distorting productive capital calculations for other sectors.
• Sector-level behavioral parameters: Five behavioral parameters (time to maximum overproduction, time of inventory restoration, maximum overproduction, target inventory level, and production reduction/heterogeneity) are modeled at the sector level with distributional bounds, enabling uncertainty quantification and global sensitivity analysis.
• Sobol sensitivity analysis: Support for variance-based sensitivity analysis to identify the most influential behavioral parameters on predicted indirect loss.

Program Overview

This repository contains the following:

• sr_wrapper.m: Wrapper script governing analysis settings
• fn_run_sr_ario.m: Main wrapper function for the single-region R-ARIO program
• functions/: Functions for simulating economic recovery and auxiliary post-processing tools
• inputs/: Pre-processed input data:
  • Economic inputs (value added, exports, imports, local demand, fixed assets)
  • Damage/loss inputs
  • Input-output table
  • params/: Input files for ARIO behavioral parameter distributions
• output/: Analysis output and postprocessing scripts

fn_run_sr_ario Overview

The function fn_run_sr_ario is the main function governing the R-ARIO analysis, called by sr_wrapper after analysis settings are defined.

• fn_load_default_ario_settings: Loads hardcoded analysis parameters
• fn_initialize_sr_variables: Initializes containers for all timesteps; sets initial pre-disaster values
• Main loop for timestep $k = 1:n$:
  • fn_compute_demand: Computes sector-level recovery percentage, updates reconstruction demand rate and total final demand
  • fn_compute_prod_lim_by_cap: Computes production constrained by production capacity
  • fn_compute_prod_lim_by_cap_sup: Computes production constrained by production capacity and available supplies
  • fn_get_output_econ_metrics: Computes actual satisfied demand and actual supply
  • fn_update_input_econ_metrics: Updates input economic metrics for the next timestep
  • fn_get_output_econ_performance: Updates output economic metrics for the next timestep

Input Files

• sr_ario_building_damage.csv: Individual damage observations/simulations for each building (or building cluster). Used to construct sector-specific recovery curves and control reconstruction rates.
• sr_ario_econ_data.mat: Sector-level economic data (exports, final demand, imports, value added, local demand, fixed assets, total output).
• sr_ario_economic_sectors.csv: List of sectors modeled in the economy (including housing, if available), with sector ID, name, and non-stockable good status.
• sr_ario_IO_data.mat: Raw (non-normalized) input-output table.
• sr_ario_loss_data.mat: Sector-level aggregate losses and reconstruction demand assignments.

Output Files

• sr_ario_results.mat: MATLAB struct (sr_output) containing all analysis results. Postprocess using sr_postprocess.m.

Running an Analysis

1. Place input files (sr_ario_building_damage.csv, sr_ario_econ_data.mat, sr_ario_economic_sectors.csv, sr_ario_IO_data.mat, sr_ario_loss_data.mat) in the inputs/ directory.
2. Configure analysis settings and run sr_wrapper.m.
3. Results are saved to the outputs/ directory.
4. Postprocess using sr_postprocess.m. Example postprocessing functions are included in functions/.

Acknowledgments

This work was supported by Sompo Holdings, Inc. and the National Science Foundation (grant CMMI-2053014), with additional support from the Stanford Urban Resilience Initiative.