Bayesian inference for dynamic spatial quantile models with interactive effects

2025-03-02

AI Summary

This paper addresses the challenge of simultaneously modeling unobserved individual effects, high-dimensional cross-sectional dependence, and heterogeneous regression coefficients in large-scale dynamic spatial panel data. To this end, we propose the first Bayesian dynamic spatial panel quantile model. Methodologically, we introduce quantile randomization, design a structure-aware Gibbs sampler, and improve inverse Gaussian random number generation to ensure tail stability; we further establish Bayesian consistency under a double-asymptotic framework. Efficient Bayesian inference is implemented via MCMC, and Monte Carlo simulations demonstrate superior estimation accuracy and robustness. An empirical application to the gasoline market uncovers quantile-specific co-movement structures, offering a novel tool for assessing asymmetric policy effects.

Technology Category

• Reasoning under Uncertainty: Relational Probabilistic Models
• Machine Learning: Calibration & Uncertainty Quantification
• Search and Optimization: Sampling/Simulation-based Search

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Economics, Online Markets and Human Computation: Incentives in network design for Web infrastructures and ecosystems
• User Modeling, Personalization and Recommendation: Practical large-scale studies of user experience

Abstract

With the rapid advancement of information technology and data collection systems, large-scale spatial panel data presents new methodological and computational challenges. This paper introduces a dynamic spatial panel quantile model that incorporates unobserved heterogeneity. The proposed model captures the dynamic structure of panel data, high-dimensional cross-sectional dependence, and allows for heterogeneous regression coefficients. To estimate the model, we propose a novel Bayesian Markov Chain Monte Carlo (MCMC) algorithm. Contributions to Bayesian computation include the development of quantile randomization, a new Gibbs sampler for structural parameters, and stabilization of the tail behavior of the inverse Gaussian random generator. We establish Bayesian consistency for the proposed estimation method as both the time and cross-sectional dimensions of the panel approach infinity. Monte Carlo simulations demonstrate the effectiveness of the method. Finally, we illustrate the applicability of the approach through a case study on the quantile co-movement structure of the gasoline market.

Problem

• Develops dynamic spatial quantile model with unobserved heterogeneity
• Proposes Bayesian MCMC algorithm for model estimation
• Analyzes quantile co-movement in gasoline market case

Innovation

• Bayesian MCMC for dynamic spatial quantile models
• Quantile randomization and new Gibbs sampler
• Stabilizes inverse Gaussian tail behavior