Sign-Dependent Spillovers of Global Monetary Policy

This paper examines the sign-dependent international spillovers of Federal Reserve and European Central Bank monetary policy shocks. Using a consistent high-frequency identification of pure monetary policy shocks across 44 advanced and non-advanced economies and the methodology of Caravello and Martinez-Bruera, 2024, we document strong asymmetries in international transmission. Linear specifications mask these effects: contractionary shocks generate large and significant deteriorations in financial conditions, economic activity, and international trade abroad, while expansionary shocks yield little to no measurable improvement. Our results are robust across samples, identification strategies, and the framework proposed by Ben Zeev et al., 2023.

💡 Research Summary

This paper investigates how monetary policy shocks from the Federal Reserve (Fed) and the European Central Bank (ECB) are transmitted internationally, using a unified high‑frequency identification of pure policy shocks and a sign‑dependent local projection (LP) framework. The authors construct a monthly panel covering 44 advanced and emerging economies from January 1999 to December 2023. Pure monetary shocks are extracted from 30‑minute windows around policy announcements, following the event‑study and decomposition approach of Jarociński and Karadi (2020) to separate the monetary component from contemporaneous information effects. For the Fed, shocks are based on U.S. interest‑rate futures; for the ECB, they rely on the Euro‑Area Monetary Policy Event‑Study Database.

The econometric specification augments the standard LP of Jordà (2005) with both the shock itself (εₜ) and its absolute value (|εₜ|):

yᵢ,ₜ₊ₕ = αᵢ,ₕ + β_sign,ₕ·εₜ + β_abs,ₕ·|εₜ| + Γₕ·Xᵢ,ₜ₋₁ + uᵢ,ₜ₊ₕ

where yᵢ,ₜ₊ₕ denotes a macro‑financial variable for country i at horizon h, and Xᵢ,ₜ₋₁ contains lagged domestic and global controls. Country‑by‑calendar‑month fixed effects absorb seasonality, and standard errors are clustered in time to account for common shocks. Under the assumption of a symmetric shock distribution, the absolute‑value term isolates sign‑dependent non‑linearities; a statistically significant β_abs,ₕ therefore indicates asymmetric responses to tightening versus easing. The impulse responses for a tightening (positive) shock and an easing (negative) shock are constructed as IRF⁺ₕ = β_abs,ₕ + β_sign,ₕ and IRF⁻ₕ = β_abs,ₕ – β_sign,ₕ, respectively.

The paper first presents linear (symmetric) LP results, which show that both Fed and ECB tightenings are associated with foreign‑currency depreciation, higher inflation, lower industrial production, and falling equity prices abroad—findings consistent with the existing literature. However, the linear model forces symmetry and implicitly assumes that easings simply reverse these effects.

When sign dependence is allowed, the authors uncover striking asymmetries. Tightening shocks generate large, persistent deteriorations in global financial conditions, output, and trade. For example, after a Fed tightening, major emerging‑market currencies depreciate by 5‑7 % against the dollar over 6‑12 months, inflation abroad rises by 0.5‑1 percentage points, and industrial production falls by 1‑2 percentage points, with effects lingering for up to three years. By contrast, easing shocks are largely statistically insignificant; in several cases they even produce opposite (i.e., improvement) effects that are economically modest. The ECB exhibits a similar pattern, with its easing shocks showing negligible international impact.

Channel‑by‑channel analyses reveal that the asymmetry is present across trade flows, sovereign spreads, equity markets, and domestic lending rates. Tightening amplifies trade deficits and widens spreads, while easing does not generate commensurate improvements.

Robustness checks include: (i) alternative high‑frequency identification schemes (different windows, alternative information‑effect separations); (ii) the interacted local projection approach of Ben Zeev et al. (2023); (iii) sample extensions to additional countries and sub‑samples of advanced economies; and (iv) tests for size‑non‑linearity by adding quadratic shock terms. All checks confirm the core finding of pronounced sign‑dependent asymmetry.

The contribution of the paper is threefold. First, it provides the first comprehensive, high‑frequency‑based evidence that international monetary transmission is fundamentally asymmetric, challenging the prevailing symmetric assumptions in most macro‑econometric studies. Second, it demonstrates that linear models substantially understate the magnitude of spillovers from tightening and may even mis‑sign the effects of easing, with important implications for policymakers and international financial institutions. Third, it showcases the practical usefulness of a parsimonious sign‑dependent LP specification that retains statistical power while allowing for non‑linearities, offering a template for future macro‑financial research.

Limitations are acknowledged. The absolute‑value term captures only sign non‑linearity, not size‑dependent asymmetry; large easings could have different effects than small ones, which the current framework cannot disentangle. Moreover, the analysis remains reduced‑form: it does not explicitly identify the structural channels (credit, capital‑flow, expectations) that drive the observed asymmetries. Finally, while the high‑frequency identification is state‑of‑the‑art, residual contamination from information effects cannot be ruled out entirely.

In conclusion, the study convincingly shows that global monetary policy spillovers are heavily driven by contractionary shocks, with expansionary shocks having muted or even opposite effects. This finding calls for a re‑evaluation of policy simulations and international coordination mechanisms that rely on symmetric transmission assumptions, and it underscores the importance of incorporating sign‑dependent dynamics in macro‑economic modeling.