{getToc} $title={Table of Contents}
Summary
The paper proposes a large non-Gaussian structural VAR model identified by higher moments without imposing economically motivated restrictions. It scales well to higher dimensions, allowing the inclusion of many variables, and develops an efficient Gibbs sampler for estimation. The model is applied to study the effects of a monetary policy shock.
Highlights
- Proposes a large non-Gaussian structural VAR model identified by higher moments.
- Does not require economically motivated restrictions for identification.
- Develops an efficient Gibbs sampler for model estimation.
- Scales well to higher dimensions, allowing many variables.
- Applies the model to study monetary policy shock effects.
- Finds that prices and output respond with a large delay to monetary policy shocks.
- Demonstrates good estimation properties through experiments with artificial data.
Key Insights
- The model exploits information in higher moments to achieve identification without relying on traditional economically motivated restrictions, offering a novel approach to structural VAR modeling.
- By not imposing restrictions based on economic theory, the model allows for a more data-driven approach to understanding the relationships between variables.
- The use of higher moments for identification enables the model to capture non-Gaussian features of the data, such as skewness and kurtosis, which can be crucial for accurately modeling economic phenomena.
- The Gibbs sampler developed for the model provides an efficient means of estimation, making it feasible to apply the model to large datasets with many variables.
- The application of the model to study the effects of a monetary policy shock highlights its practical relevance and demonstrates its ability to provide insights into complex economic issues.
- The finding that prices and output respond with a significant delay to monetary policy shocks underscores the importance of considering the dynamic effects of policy interventions.
- The model's good estimation properties, as evidenced by experiments with artificial data, suggest its reliability for empirical analysis.
Mindmap
Citation
Prüser, J. (2024). A large non-Gaussian structural VAR with application to Monetary Policy (Version 1). arXiv. https://doi.org/10.48550/ARXIV.2412.17598