## 6.3 Monetary policy shocks with global monetary policy synchronization ## Source details **Canonical URL:** [6.3 Monetary policy shocks with global monetary policy synchronization](https://www.imf.org/-/media/files/publications/wp/2025/english/wpiea2025140-print-pdf.pdf) ## Other formats - [Markdown version](/-/media/files/publications/wp/2025/english/wpiea2025140-print-pdf.pdf.md) - [Structured JSON version](/-/media/files/publications/wp/2025/english/wpiea2025140-print-pdf.pdf.json) --- ### Motivation and puzzle - Euro area imports collapsed in 2023, falling by over 10 percent in real terms, while GDP growth remained resilient. - Traditional GDP-based import models and forecaster projections failed to predict the 2023 import collapse; IMF WEO and other forecasters over-predicted trade in 2023 and 2024Q1-Q3. - Hypothesis: monetary policy tightening alters the composition of domestic demand (across and within components) and, because demand components differ in import intensity, this composition change is an overlooked channel through which monetary policy affects imports. ### Data, sample, and measures - Sample: nine euro area countries (EA9); EA9 represent 91 percent of EA GDP. - Import-Adjusted Domestic Demand (IADct) constructed as: IADct = Cωc t + Gωg t + Iωi t + Xωx t, where ωc, ωg, ωi, ωx are import-content weights. - Baseline import-intensity weights: time-invariant averages from 2005 to 2020 (input-output tables up to 2021; Covid-19 excluded from weighting period). - Monetary policy proxies: - Country-specific lending rates proxied by rate on outstanding loans to non-financial corporations. - High lending rates dummy: equal to 1 when nominal lending rates are greater or equal than three (the cross-country average of the distribution). - Global monetary policy synchronization: share of countries tightening monetary policy (Caldara and others (2024) measure). ### Baseline import elasticity model and core estimates - Baseline regression (Equation 2): ∆lnMct = β∆IADct + δ∆lnPct + θXct + μc + τq + εct. - Key baseline coefficient estimates: - Baseline elasticity of imports to IAD: approximately 1.62 (reported values: 1.62***; variants 1.63***, 1.66*** across specifications). - Import prices coefficient: around -0.40 (reported values: -0.40***, -0.38***, -0.37***, -0.33*** across specifications). - Model fit improvements: - Including lending rates improves the fit: fit of the model including lending rates is around 13 percent higher than the model without monetary policy. - Residuals are around 40 percent smaller in 2023 when including lending rates. ### Non-linearity with lending rates (high-rate regime) - Interaction specification: IAD × High lending rates dummy (dummy = 1 when rates ≥ 3). - Estimated elasticities: - Elasticity of imports to IAD when interest rates are low: 1.12 (reported: IAD coefficient 1.12*** in Column 6 baseline interacted specification). - Marginal effect when high lending rates: additional 1.19*** on IAD (reported: Dummy high lending rates x IAD 1.19*** (0.22)), implying elasticity above 2 when rates are high. - Interpretation: when lending rates are high, imports become much more sensitive to changes in domestic demand. ### Cross-country heterogeneity - Country-by-country estimates of the IAD × lending rates interaction: - Most countries show positive interaction estimates in the range of 1–1.7. - Germany presents the largest coefficient, followed by France and the Netherlands. - Finland and Portugal display low and insignificant coefficients on the interaction term. - Statistical note: while qualitative heterogeneity exists, variation across most countries is not statistically significant except Portugal and Finland. ### Monetary policy synchronization and amplification - Interaction between domestic lending rates and the share of countries tightening: - Has a negative effect on import growth and shrinks the model residual by about two thirds. - Marginal effect of lending rates on imports becomes negative and significant when the share of countries tightening monetary policy is larger than 0.6. - When a country tightens while most other countries are easing (share close to zero), the marginal effect of lending rates on imports can be positive (due to larger currency appreciation making imports cheaper). - Mechanism hypothesis: synchronized global tightening reduces the appreciation of the domestic currency following a domestic tightening (or leads to depreciation), amplifying the negative demand-composition effect on imports. ### Differential sensitivities across goods categories - BEC classification results: - Capital goods: highest interaction effect (capital goods sample; IAD coefficient 1.01***; Dummy high lending rates x IAD 1.12***). - Consumption goods (aggregate): interaction term insignificant. - Durables/Semi-durables: positive and significant interaction (IAD 1.09***; Dummy high lending rates x IAD 0.34**). - Non-durables: no varying elasticity with respect to financial conditions. - Interpretation: capital goods imports (investment-related) show the largest increase in import elasticity in high-rate regimes; durables and semi-durables also display increased elasticity but smaller. ### Robustness checks and key numeric robustness outputs - Robustness exercises include: - Lags of IAD and inventories. - Inclusion of the federal funds rate (FFR) to proxy global financial conditions. - Controls for trade policy uncertainty and the real effective exchange rate (REER). - A restrictive specification including all covariates simultaneously. - Robustness results: - Dummy high lending rates x IAD coefficients remain significant and around 1.21–1.29 in robustness columns. - REER coefficient reported in one robustness column: -0.68*** and another -0.46*** (when included). ### High-frequency monetary policy shocks and local projections - Identification: Jarociński and Karadi (2020) high-frequency monetary policy shock series used; standardized shock plotted in Figure 17. - Estimation: panel local projections (Equation 3) for GDP and demand components with controls and four lags; sample 2005q1–2024q2 excluding 2019q1–2022q1 (pandemic and recovery excluded). - Aggregate impulse responses to an exogenous monetary policy tightening: - Private consumption: declines. - Private investment: declines. - Exports: significant negative impact (controlling for exchange rate). - Government consumption: largely unaffected. - GDP: declines less than exports and investment; shares of exports and investment in GDP fall. - Implication: monetary policy tightening reduces GDP’s import intensity because the shares of the most import-intensive components (exports and investment) fall while government consumption (least import-intensive) is unaffected or rises in share. ### Demand sub-component dynamics (local projections and compositional analysis) - Exports components: - Intermediate goods exports and capital goods exports fall significantly more than total exports. - Consumption goods exports fall less. - Import intensity by BEC (Table 4): Capital goods 33.8; Intermediate goods 33.3; Consumption goods 24.4 (import intensity in %). - Investment components: - Machinery investment falls more than overall investment. - Construction falls by less than overall investment. - Machinery has higher import intensity than construction (e.g., foreign value added around 20 percent for machinery vs 1.3 percent for construction in TiVa 2021/2020). - Private consumption components: - Durables consumption falls slightly less; non-durables slightly more relative to total private consumption. - Monetary policy has a limited effect on consumption composition. - Compositional-vector (Aitchison/ilr) results (Appendix B): - Aggregate GDP composition: - Export-to-GDP ratio declines by over 2 percent at impact. - Government consumption rises by nearly 3 percent at impact. - After 2 years: shares of exports, private consumption, and government consumption broadly return to initial values; share of investment remains about 2 percent below the pre-shock level. - Exports composition: capital and intermediates-to-export ratios decline at impact; consumption goods-to-export ratio increases at impact. - Investment composition: machinery-to-investment share declines; construction share rises; intellectual property share largely unchanged; takes nearly two years to revert. - Private consumption composition: no clear evidence that private consumption becomes less import-intensive; changes are much smaller than for exports and investment. ### Exchange rate channel under synchronization - Local projection interaction (shock × I{Synct > Synct̄}) results: - Imports: negative response to monetary policy shock; amplified (more negative) during periods of high global tightening synchronization (red line = sum of coefficients βh + θh). - Real effective exchange rate (REER): - When global monetary policy is not synchronized, REER appreciates after three quarters. - When global tightening is high, REER initially does not move and eventually depreciates after domestic monetary policy tightening. - Interpretation: synchronized global tightening alters exchange rate dynamics in ways that can amplify the negative import response to domestic tightening. ### Overall conclusions and policy implications - Monetary policy is a significant driver of import dynamics through a demand-composition channel: - Tightening raises lending rates, which disproportionately depress import-intensive components (exports, machinery investment, capital & intermediate exports), reducing GDP’s import intensity. - The import elasticity of domestic demand increases in high-rate regimes, particularly for capital goods and (to a lesser extent) durables/semi-durables. - Global monetary policy synchronization amplifies the negative import impact of domestic tightening by affecting exchange rate dynamics and global demand co-movement. - Practical considerations for policymakers and modelers: - For central banks, assessing heterogeneous impacts of policy on GDP components is important for calibrating policy and anticipating effects on trade, exchange rates, and distributional outcomes. - Models that rely solely on GDP to forecast imports may miss important composition-driven channels, especially in high-rate and synchronized-tightening environments. *Source: wpiea2025140-print-pdf (sections and page numbers as provided).* ### 5.1 Baseline results. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 ### 5.1 Baseline results ### Major themes and subsections - 5.1 Baseline results. . . . . . . . . . . . . . . . . . . . . . . . . 15 - 5.2 The role of monetary policy synchronization. . . . . . . . . . . . . . . . . . . . . 18 - 5.3 Differential sensitivities across types of goods. . . . . . . . . . . . . . . . . . . . 19 - 5.4 Robustness. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 ### Key statistics (document structure) - Section 5.1 begins on page 15. - Section 5.2 begins on page 18. - Section 5.3 begins on page 19. - Section 5.4 begins on page 20. ### 6 The effect of monetary policy shocks on demand composition ### Major themes and subsections - 6 The effect of monetary policy shocks on demand composition. . . . . . . . . . . . . . . 22 - 6.1 Baseline results. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 - 6.2 The effect of monetary policy on demand sub-components. . . . . . . . . . . . . . . . 26 ### Key statistics (document structure) - Section 6 begins on page 22. - Section 6.1 begins on page 23. - Section 6.2 begins on page 26. *Source: wpiea2025140-print-pdf (sections and page numbers as provided).* ### 6.3 Monetary policy shocks with global monetary policy synchronization. . . . . . . . . 29 ### 6.3 Monetary policy shocks with global monetary policy synchronization. ### Motivation and puzzle - Euro area imports collapsed in 2023, falling by over 10 percent in real terms, while GDP growth remained resilient. - Traditional GDP-based import models and forecaster projections failed to predict the 2023 import collapse; IMF WEO and other forecasters over-predicted trade in 2023 and 2024Q1-Q3. - Hypothesis: monetary policy tightening alters the composition of domestic demand (across and within components) and, because demand components differ in import intensity, this composition change is an overlooked channel through which monetary policy affects imports. ### Key data and measures - Sample: nine euro area countries (EA9); EA9 represent 91 percent of EA GDP. - Import-Adjusted Domestic Demand (IADct) constructed following Bussière and others (2013): - IADct = Cωc t + Gωg t + Iωi t + Xωx t, where ωc, ωg, ωi, ωx are import-content weights. - Baseline import-intensity weights: time-invariant averages from 2005 to 2020 (input-output tables up to 2021; Covid-19 excluded from weighting period). - Monetary policy proxies: - Country-specific lending rates proxied by rate on outstanding loans to non-financial corporations. - High lending rates dummy: equal to 1 when nominal lending rates are greater or equal than three (the cross-country average of the distribution). - Global monetary policy synchronization: share of countries tightening monetary policy (Caldara and others (2024) measure). ### Baseline import elasticity model and main coefficients - Baseline regression (Equation 2): ∆lnMct = β∆IADct + δ∆lnPct + θXct + μc + τq + εct. - Key baseline coefficient estimates (Table 1 and text): - Baseline elasticity of imports to IAD: approximately 1.62 (reported values: 1.62***; variants 1.63***, 1.66*** across specifications). - Import prices coefficient: around -0.40 (reported values: -0.40***, -0.38***, -0.37***, -0.33*** across specifications). - Model fit: - Including lending rates improves the fit: fit of the model including lending rates is around 13 percent higher than the model without monetary policy. - Residuals are around 40 percent smaller in 2023 when including lending rates. ### Non-linearity with lending rates (high-rate regime) - Interaction of IAD with high lending rates (dummy = 1 when rates ≥ 3): - Estimated elasticity of imports to IAD when interest rates are low: 1.12 (reported: IAD coefficient 1.12*** in Column 6 baseline interacted specification). - Estimated marginal effect when high lending rates: additional 1.19*** on IAD (reported: Dummy high lending rates x IAD 1.19*** (0.22)), implying elasticity above 2 when rates are high. - Interpretation: when lending rates are high, imports become much more sensitive to changes in domestic demand. ### Cross-country heterogeneity - Country-by-country estimates of the IAD × lending rates interaction (from Table 1 – Column 6 run separately): - Most countries show positive interaction estimates in the range of 1–1.7. - Germany presents the largest coefficient, followed by France and the Netherlands. - Finland and Portugal display low and insignificant coefficients on the interaction term. - Statistical note: while qualitative heterogeneity exists, variation across most countries is not statistically significant except Portugal and Finland. ### Monetary policy synchronization and amplification - Empirical finding (section 5.2 and Figures 15–16): - Interaction between domestic lending rates and the share of countries tightening monetary policy has a negative effect on import growth and shrinks the model residual by about two thirds. - Marginal effect of lending rates on imports becomes negative and significant when the share of countries tightening monetary policy is larger than 0.6. - When a country tightens while most other countries are easing (share close to zero), the marginal effect of lending rates on imports can be positive (due to larger currency appreciation making imports cheaper). - Mechanism hypothesis: synchronized global tightening reduces the appreciation of the domestic currency following a domestic tightening (or leads to depreciation), amplifying the negative demand-composition effect on imports. ### Differential sensitivities across goods categories - Classification via BEC and results (Table 2): - Capital goods: highest interaction effect (capital goods sample; IAD coefficient 1.01***; Dummy high lending rates x IAD 1.12***). - Consumption goods (aggregate): interaction term insignificant. - Durables/Semi-durables: positive and significant interaction (IAD 1.09***; Dummy high lending rates x IAD 0.34**). - Non-durables: no varying elasticity with respect to financial conditions. - Interpretation: capital goods imports (investment-related) show the largest increase in import elasticity in high-rate regimes; durables and semi-durables also display increased elasticity but smaller. ### Robustness checks - Robustness exercises (Table 3) include: - Lags of IAD and inventories. - Inclusion of the federal funds rate (FFR) to proxy global financial conditions. - Controls for trade policy uncertainty and the real effective exchange rate (REER). - A restrictive specification including all covariates simultaneously. - Result: the baseline finding that import elasticity to domestic demand increases in high-rate regimes is robust across these specifications. - Selected numeric robustness outputs (Table 3 highlights): - Dummy high lending rates x IAD coefficients remain significant and around 1.21–1.29 in robustness columns. - REER coefficient reported in one robustness column: -0.68*** and another -0.46*** (when included). ### High-frequency monetary policy shocks and local projections - Identification: Jarociński and Karadi (2020) high-frequency monetary policy shock series used; standardized shock plotted in Figure 17. - Estimation: panel local projections (Equation 3) estimated for GDP and demand components with controls and four lags; sample 2005q1–2024q2 excluding 2019q1–2022q1 (pandemic and recovery excluded). - Aggregate responses to an exogenous monetary policy tightening (Figure 18): - Private consumption: declines. - Private investment: declines. - Exports: significant negative impact (controlling for exchange rate). - Government consumption: largely unaffected. - GDP: declines less than exports and investment; shares of exports and investment in GDP fall. - Implication: monetary policy tightening reduces GDP’s import intensity because the shares of the most import-intensive components (exports and investment) fall while government consumption (least import-intensive) is unaffected or rises in share. ### Demand sub-components (local projections at greater granularity) - Exports components (Figure 19): - Intermediate goods exports and capital goods exports fall significantly more than total exports. - Consumption goods exports fall less. - Import intensity by BEC (Table 4): Capital goods 33.8; Intermediate goods 33.3; Consumption goods 24.4 (import intensity in %). - Investment components (Figure 20): - Machinery investment falls more than overall investment. - Construction falls by less than overall investment. - Machinery has higher import intensity than construction (e.g., foreign value added around 20 percent for machinery vs 1.3 percent for construction in TiVa 2021/2020). - Consumption components (Figure 21): - Durables consumption falls slightly less; non-durables slightly more relative to total private consumption. - Monetary policy has a limited effect on consumption composition; tendency to slightly increase consumption import intensity is ambiguous. - Net effect: within-demand composition shifts (less machinery, less intermediate/capital exports) lower import content of GDP following monetary tightening. ### Exchange rate channel under synchronization - Local projection interaction (Equation 5) with shock × I{Synct > Synct̄}: - Imports: negative response to monetary policy shock; amplified (more negative) during periods of high global tightening synchronization (red line = sum of coefficients βh + θh). - Real effective exchange rate (REER): when global monetary policy is not synchronized, REER appreciates after three quarters; when global tightening is high, REER initially does not move and eventually depreciates after domestic monetary policy tightening. - Interpretation: synchronized global tightening alters exchange rate dynamics in ways that can amplify the negative import response to domestic tightening. ### Overall conclusions and policy implications - Monetary policy is a significant driver of import dynamics through a demand-composition channel: - Tightening raises lending rates, which disproportionately depress import-intensive components (exports, machinery investment, capital & intermediate exports), reducing GDP’s import intensity. - The import elasticity of domestic demand increases in high-rate regimes, particularly for capital goods and (to a lesser extent) durables/semi-durables. - Global monetary policy synchronization amplifies the negative import impact of domestic tightening by affecting exchange rate dynamics and global demand co-movement. - Practical considerations: - For central banks (e.g., the ECB), assessing the heterogeneous impact of policy on GDP components is important for calibrating policy and anticipating effects on trade, exchange rates, and distributional outcomes. - Models that rely solely on GDP to forecast imports may miss important composition-driven channels, especially in high-rate and synchronized-tightening environments. *Source: https://www.imf.org/-/media/files/publications/wp/2025/english/wpiea2025140-print-pdf.pdf* ### References ### References (wpiea2025140-print-pdf) ### Data and variables (Appendix A) - Data sources: - Import, GDP, inventories, import prices, and the share of services in total private consumption: Eurostat. - IAD: constructed using Eurostat and OECD Input-Output tables (see Section 4). - Lending rates: ECB (downloaded through Haver), interest rate on outstanding loans to non-financial corporations. - Share of countries tightening (monetary policy synchronization): calculated following Caldara and others (2024) using IMF BIS data. - Federal fund rate: Federal Reserve Board (obtained through Haver). - Trade policy uncertainty: Caldara and others (2020). - Real effective exchange rate: BIS. - Table 5: Summary Statistics - Import Elasticity Regressions (all variables presented in year-over-year percentage changes) - Real import: mean 0.00, sd 0.11, min -0.42, max 0.30, count 666 - Import prices: mean 0.02, sd 0.08, min -0.24, max 0.26, count 666 - GDP: mean 0.01, sd 0.04, min -0.24, max 0.18, count 666 - IAD: mean 0.01, sd 0.05, min -0.24, max 0.26, count 666 - Inventories: mean 0.01, sd 0.01, min -0.04, max 0.07, count 666 - Share of services in consumption: mean -0.00, sd 0.04, min -0.27, max 0.22, count 518 - Lending rates: mean 0.00, sd 0.24, min -0.80, max 1.14, count 666 - Share of countries tightening: mean 0.35, sd 0.27, min 0.00, max 0.95, count 702 - Federal fund rate: mean 0.03, sd 1.24, min -3.24, max 3.39, count 666 - Trade policy uncertainty: mean 0.05, sd 0.41, min -1.10, max 1.29, count 666 - Real effective exchange rate: mean -0.00, sd 0.02, min -0.06, max 0.05, count 592 - Broad Economic Classification (BEC) mapping (Eurostat): - Capital goods: category 410. - Consumption goods: categories 111, 121, 210, 222, 310, 322, 420, 530. - Durables and semi-durables: categories 610 and 620. - Non-durables: category 630. ### Compositional methodology (Appendix B) - Purpose: - Model changes in demand composition simultaneously so that shares of one demand component affect the others. - Definitions and constructions: - Demand-component shares: - y_{w,c,t} = w_{c,t} / GDP_{c,t}, where w_{c,t} ∈ {Private consumption, Government Consumption, Investment, Exports}. - y_{k,c,t} = k_{c,t} / w_{c,t}, where k_{c,t} are sub-components (e.g., Durable consumption, Services Consumption, Consumption goods other than durables; Machinery, Construction, Intellectual property; Capital goods, Intermediate goods, Consumption goods for exports). - Vectors Y_{w,c,t} and Y_{k,c,t} belong to the simplex S_D: - S_D = {⟨x_1, ..., x_D⟩^⊤ : x_i > 0, i = 1, ..., D; Σ_{i=1}^D x_i = κ }. - Aitchison geometry: - Inner product ⟨x,y⟩_a = (1/(2D)) Σ_{i=1}^D Σ_{j=1}^D ln(x_i/x_j) ln(y_i/y_j). - Norm ∥x∥_a^2 = ⟨x,x⟩_a and distance d_a(x,y) = ∥xΘy∥_a. - Perturbation and power operations defined; closure operator C(·) used to map to the simplex. - Log-ratio transformations: - Isometric log-ratio (ilr) transformation applied (Egozcue and Pawlowsky-Glahn (2005)) to map compositions in S_D to orthonormal coordinates in R^{D−1}: - ilr(x) = (z_1, ..., z_{D−1})^⊤, z_j = √[(D−j)/(D−j+1)] ln( x_j^{√} / (∏_{i=j+1}^D x_i)^{1/√} ), j = 1, ..., D−1 (formal expression retained as in source). - Inverse ilr used to map coordinates back to the simplex. - Justification: - Compositional time series (CTS) are multivariate time series with each observation on the simplex; standard multivariate time series methods require real coordinates and thus log-ratio transformations are necessary. - The estimation of full compositional VARIMA/VAR models is independent of the chosen log-ratio transformation (Kynčlová, Filzmoser, and Hron (2015)). - Regression framework (Aitchison-linear structure): - Compositional vector model on the simplex: - Y_{w,c,t} = b ⊕ (A^{(1)} ⊙ Y_{w,c,t−1}) ⊕ (A^{(2)} ⊙ Y_{w,c,t−2}) ⊕ · · · ⊕ (β ⊙ MP_t) ⊕ w_t. - After ilr transformation, becomes a standard local projection (LP): - z_{c,t+h} = Σ_{p=0}^P c_{h,p} z_{t−p} + Σ_{p=0}^P A_{h,p} shock_{t−p} + ε_{c,t+h}. ### Main empirical findings from compositional analysis (Appendix B results) - Aggregate GDP composition (Figure 23): - Demand becomes less import-intensive following a monetary policy tightening. - Export-to-GDP ratio declines by over 2 percent. - Government consumption rises by nearly 3 percent. - Slight initial decrease in investment-to-GDP ratio and corresponding increase in private consumption-to-GDP ratio. - After 2 years: - Shares of exports, private consumption, and government consumption broadly return to initial values. - Share of investment remains about 2 percent below the pre-shock level. - Conclusion: monetary policy changes composition of demand in a way that lowers import intensity; result is robust to simultaneous compositional modeling. - Exports composition (Figure 24): - Exports become less import-intensive following a monetary policy tightening. - Capital and intermediates-to-export ratios decline at impact. - Consumption goods-to-export ratio increases at impact. - Share of intermediates keeps falling for 3 quarters before recovering. - Consistent with Section 6 analysis on export composition. - Investment composition (Figure 25): - Machinery investment is more interest-sensitive than construction: - Decline in machinery-to-investment share. - Rise in construction share. - Intellectual property investment share remains largely unchanged. - It takes nearly two years for these shares to return closer to initial levels. - Implication: investment becomes less import-intensive following a monetary policy tightening. - Private consumption composition (Figure 26): - No clear evidence that private consumption becomes less import-intensive following a monetary policy tightening. - Dynamics: - Semi-durables share initially increases while non-durables share decreases; durable consumption remains largely unchanged. - After three quarters: - Semi-durables share falls below its initial level. - Non-durables return to their original proportion. - Overall: compositional changes in consumption are much smaller than for exports and investment, implying minimal shift in consumption import intensity. *Source: wpiea2025140-print-pdf - References (appendices A and B).* --- _Source: https://www.imf.org/-/media/files/publications/wp/2025/english/wpiea2025140-print-pdf.pdf_