Heterogeneous Agent Kalman-Filtering Framework for Dynamic Fiscal Multiplier Decomposition Under Incomplete Markets and Stochastic Volatility Regimes

Abstract

This study proposes a structural estimation framework that integrates heterogeneous agent modeling with Kalman-filter-based signal extraction to decompose state-contingent fiscal multipliers under incomplete asset markets and stochastic volatility. Leveraging a Bewley-Huggett-Aiyagari backbone augmented with Markov-switching variance processes, the model disentangles aggregate demand externalities from precautionary saving channels in government expenditure transmission. Using quarterly panel data from eighteen OECD economies spanning 2000–2023, we apply Bayesian sequential Monte Carlo inference to recover latent volatility states and household-level marginal propensities to consume. Empirical results confirm that fiscal multipliers exhibit strong regime-dependency, exceeding unity during high-volatility contractions while remaining sub-unitary in tranquil expansionary phases. These findings carry direct implications for counter-cyclical fiscal consolidation design and sovereign debt sustainability assessments within post-pandemic macroeconomic stabilization frameworks.