Donor-Adjusted Post-Shock Forecasting

Implements donor-adjusted methods for forecasting conditional means and variances after structural shocks. Historical donor episodes are weighted using covariates observed before each shock, and their estimated post-shock effects are combined with forecasts from a target-series model. The methods build on Lin and Eck (2021) . The package supports donor balancing weights, structured donor pools, autoregressive integrated moving average models, and generalized autoregressive conditional heteroscedasticity models with external regressors.

postshock

postshock provides donor-adjusted methods for forecasting conditional means and variances after structural shocks. Historical donor episodes are weighted using information observed before each shock, and their estimated post-shock effects are combined with forecasts from a target-series model.

The package builds on the methodology of Lin and Eck (2021) and extends it with practical tools for donor weighting, conditional mean forecasting, volatility forecasting, structured donor pools, and automatic GARCH-X model selection.

Main Features

Donor balancing weights (dbw())

Estimate donor weights from pre-shock covariates so that weighted donor characteristics resemble those of the target series.

Features include:

• optional scaling and centering
• lower and upper weight bounds
• L1 and L2 regularization
• uniform-weight benchmark

Conditional mean forecasting (SynthPrediction())

Generate donor-adjusted forecasts by

• fitting an ARIMA/SARIMA model to the target series,
• estimating donor-specific post-shock effects,
• combining donor effects using balancing weights, and
• adjusting the target forecast.

Structured donor pools

Support multiple donor pools representing different economic or substantive mechanisms, allowing donor effects to be estimated and combined within predefined groups.

Automatic GARCH-X model selection (auto_garchx())

Automatically search across GARCH-X and GJR-GARCH specifications using information criteria.

Supported features include

• external regressors,
• robust covariance estimation,
• backcast handling, and
• automatic order selection.

Conditional variance forecasting (SynthVolForecast())

Forecast post-shock conditional variances by combining

• a target-series GARCH-X model,
• donor-estimated volatility adjustments, and
• balancing weights estimated from pre-shock information.

Covariate construction (build_control_xreg())

Construct control variables and external regressors for use in forecasting workflows.

Installation

Once the package is available on CRAN, install it with

install.packages("postshock")

Getting Started

Load the package

library(postshock)

The package vignette provides a complete workflow covering

• donor balancing weights,
• conditional mean forecasting,
• structured donor pools,
• volatility forecasting, and
• practical guidance for applied analyses.

Open the vignette with

vignette("postshock-workflow", package = "postshock")

Main Functions

Reference

Lin, Y. and Eck, D. J. (2021).

Post-shock forecasting using synthetic controls.

International Journal of Forecasting.

https://doi.org/10.1016/j.ijforecast.2021.03.010

License

MIT License.