🤖 AI Summary
When macroeconomic shocks exhibit a composite structure—comprising multiple distinct underlying components—the local-projection instrumental-variable (LP-IV) estimator aggregates impulse responses across components using weights that can be negative, violating the exclusion restriction and undermining causal interpretability. Method: This paper provides the first rigorous demonstration that LP-IV weights are not inherently nonnegative; it proposes a novel set-identification framework incorporating sign restrictions or fine-grained auxiliary information to achieve causal decomposition of composite shocks. Contribution/Results: Empirically, the study confirms that monetary policy shocks are indeed composite, and estimates a non-defense government spending multiplier significantly greater than one. By exposing a fundamental flaw in conventional LP-IV logic, the work redefines the foundations of macroeconomic IV identification and delivers an operationally feasible, structurally interpretable paradigm for structural estimation under composite shocks.
📝 Abstract
Macro shocks are often composites, yet overlooked in the impulse response analysis. When an instrumental variable (IV) is used to identify a composite shock, it violates the common IV exclusion restriction. We show that the Local Projection-IV estimand is represented as a weighted average of component-wise impulse responses but with possibly negative weights, which occur when the IV and shock components have opposite correlations. We further develop alternative (set-) identification strategies for the LP-IV based on sign restrictions or additional granular information. Our applications confirm the composite nature of monetary policy shocks and reveal a non-defense spending multiplier exceeding one.
