## 7. Effect of a monetary policy shock on cross-border bank lending from eight OECD countries

## Source details

**Canonical URL:** [7. Effect of a monetary policy shock on cross-border bank lending from eight OECD countries](https://www.imf.org/-/media/files/publications/wp/2019/wpiea2019234-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2019/wpiea2019234-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2019/wpiea2019234-print-pdf.pdf.json)

---

### Research question and approach
- Primary question: Do domestic monetary policy actions affect cross-border bank lending, and through which transmission channels (bank lending, risk-taking, portfolio rebalancing)?
- Core claim: Identifying exogenous monetary policy surprises (orthogonal to systematic policy responses) is key to reconciling mixed empirical evidence.
- Identification strategies used:
  - Narrative shocks: Romer and Romer (2004) extended by Coibion (2012) for U.S. shocks (summed within quarter).
  - High-frequency external instruments / proxy-SVAR (Gertler and Karadi, 2015; Gürkaynak et al., 2005) as robustness.
  - Forecast-error-based shocks for eight other advanced economies following Furceri et al. (2018) with additional purification via projections on current and lagged GDP growth and inflation.
- Key sample windows:
  - U.S. analysis: 1990Q1 to 2012Q4.
  - Eight other advanced economies: 2001Q1 to 2012Q4.
- Scope of lender countries (eight): Canada, Germany, Italy, Japan, the Netherlands, Spain, Sweden, and the United Kingdom.

### Data and measurement
- Cross-border banking data source: BIS Locational Banking Statistics (LBS), residency principle; currency composition used to construct exchange-rate adjusted flows and cumulated stock.
- Exchange-rate adjustment method: convert U.S. dollar-equivalent amounts to original currency using end-of-period exchange rates, compute differences in original currency, convert differences to U.S. dollar-equivalents using period-average exchange rates.
- Data processing choices:
  - Drop offshore financial centers per IMF classification.
  - Drop observations with cross-border positions less than $5 million or negative total outstanding claims.
  - Exclude dependent variable observations in upper and lower one percentile of distribution.
- Coverage notes:
  - BIS LBS captures around 95 percent of all cross-border interbank business.
  - Final baseline estimation for U.S. covers cross-border lending to 45 recipient countries.
- Shock series properties:
  - Correlation between U.S. shocks from Furceri et al. (2018) and Coibion (2012): 0.62 for the overlapped sample.
  - Country-specific shocks typically have small correlation with U.S. shocks except for Canada.

### Empirical methodology
- Main estimator: Local projection method (Jordà, 2005) on bilateral panel data to estimate dynamic impulse responses for horizons h=0,1,2,...,7.
- Baseline regression (simplified):
  - y_{j,t+h} − y_{j,t−1} = α_{j,h} + β_{h} MPshock_{t} + Σ γ_{h} X_{j,t-p} + ε_{j,t+h}
  - y = log exchange-rate-adjusted cross-border claims from source to recipient j.
  - X includes four lags of dependent variable, monetary shocks, recipient GDP growth, short-term interest rate, inflation, exchange rate growth, and their lags.
- Fixed effects and standard errors:
  - Recipient country fixed effects in U.S. analysis.
  - For multi-source bilateral analysis: source-recipient fixed effects and recipient-time fixed effects.
  - Standard errors clustered by time; robustness checks include clustering by recipient and Driscoll-Kraay standard errors.
- Rationale:
  - Exogenous shocks obviate need for VAR identification restrictions.
  - Local projections enable flexible dynamics and state-dependent responses.

### Key baseline findings (U.S. monetary policy shocks)
- Sign and magnitudes:
  - Exogenous monetary policy tightening leads to an economically and statistically significant decline in cross-border bank lending, consistent with bank lending and risk-taking channels.
  - A 100 basis-point (bp) exogenous tightening leads to more than a 10 percent decline in cross-border bank lending after two quarters.
- Representative coefficients from baseline dynamic estimates (Table 1, selected):
  - Monetary policy shock (h=1): -10.953**.
  - Monetary policy shock (h=2): -12.077**.
  - Monetary policy shock (h=0): -2.429 (not significant).
  - Monetary policy shock (-1) (lagged): -5.659* at h=0.
  - Recipient exchange rate coefficients: e.g., -0.292** at h=0; -0.829*** at h=3.
- Model fit and sample size (Table 1):
  - Observations by horizon: Obs = 3,064 (h=0), 3,018 (h=1), 2,979 (h=2), 2,934 (h=3), 2,889 (h=4), 2,855 (h=5), 2,807 (h=6), 2,770 (h=7).
  - R-squared by horizon: 0.09 (h=0), 0.11 (h=1), 0.14 (h=2), 0.14 (h=3), 0.16 (h=4), 0.20 (h=5), 0.23 (h=6), 0.24 (h=7).
- Contrast with static/specification using policy-rate levels or changes:
  - Static regressions using lagged Federal funds rate or changes produce positive or mixed-signed associations (Table 2, Columns I–II).
  - Replacing policy-rate changes with exogenous monetary policy shocks flips the sign and reveals negative effects (Table 2, Column III and dynamic results).
- Proxy-SVAR robustness:
  - Proxy-SVAR structural shock confirms negative and significant decline in cross-border lending to a 100 bp structural tightening.
  - First-stage F-statistics by horizon: 330.42, 364.23, 358.67, 317.37, 332.79, 364.67, 333.84, 336.06.

### State-dependency, uncertainty, and asymmetries
- Business cycle state (smooth transition; five-quarter moving average of real GDP growth; θ=1.5 calibration):
  - Effect is weaker and less precisely estimated during recessions; negative effect persists and is stronger during expansions for exogenous shocks.
  - Using changes in Federal funds rate shows different signs across expansions vs. recessions.
- Global uncertainty (VIX-based regimes):
  - Spillovers are stronger during low-uncertainty periods (low-VIX) and weaker during high-uncertainty periods (high-VIX).
- Tightening vs. easing:
  - Dummy analysis shows no strong symmetric/asymmetric differences in average responses; tightening during expansions may have more persistent effects than easing during recessions.

### Heterogeneity by borrower riskiness and currency-area patterns
- Borrower risk:
  - Grouping by ICRG Political Risk Index (above vs. below median) or by advanced vs. emerging status yields negative and statistically significant effects for both safe and risky borrowers.
  - Magnitudes are similar across groups; differences not statistically significant — no support for dominant international portfolio rebalancing at recipient-country aggregation.
- Euro-area patterns:
  - For euro-area source countries, spillovers tend to be stronger toward non-euro-area borrowers than toward euro-area borrowers (subsample evidence).

### International evidence across eight advanced economies (bilateral panel)
- Regression framework:
  - y_{i,j,t+h} − y_{i,j,t−1} = α_{i,j,h} + α_{j,t,h} + β_{h} MPshock_{i,t} + Σ γ_{h} X_{i,j,t-p} + ε
  - Includes source-recipient fixed effects and recipient-time fixed effects.
- Main result:
  - A one percent point (100 bp) exogenous tightening in lender countries leads to about a 5 percent decline in cross-border bank lending after one quarter and a decline of more than 10 percent after one year.
  - Using unpurged changes in the policy rate shows no short-run negative effect.
- Robustness:
  - Results robust to dropping euro-area countries other than Germany, controlling for lender domestic macro variables (real GDP growth, inflation, stock returns, exchange rate growth), alternative standard error clustering, more lags, and controlling for bilateral trade flows.
  - Effects stronger in low-uncertainty periods.

### Robustness checks and controls
- Controls tested without materially changing results:
  - Inclusion of additional U.S. controls (U.S. real GDP growth, inflation, stock returns).
  - Longer lag lengths (up to eight lags).
  - Alternative standard errors: recipient clustering, recipient-time clustering, Driscoll-Kraay.
  - Controlling for bilateral trade flows (IMF Directions of Trade Statistics).
  - Controlling for global financial risk (VIX) and liquidity risk (LIBOR-OIS spread).
- Data handling robustness:
  - Qualitative results robust to inclusion of observations < $5 million, alternative percentile trimming (top/bottom 2.5%), winsorizing, and including all observations.

### Interpretation and policy implications
- Mechanisms supported:
  - Evidence consistent with international bank lending channel and risk-taking channel: exogenous tightening reduces cross-border lending.
- Mechanisms not supported:
  - Little support for the international portfolio rebalancing channel at recipient-country aggregation.
- Policy relevance:
  - Exogenous conventional monetary tightening in systemically important advanced economies has substantial cross-border banking spillovers (declines in cross-border lending).
  - U.S. monetary policy acts as an independent source of the global financial cycle, even when controlling for VIX and LIBOR-OIS.
  - Monetary policy effectiveness in generating cross-border lending responses is state-dependent and weaker under high global uncertainty.
- Methodological contribution:
  - Demonstrates value of purged exogenous shock measures and local projections on bilateral datasets for tracing dynamic international spillovers.

### Conclusions and avenues for future research
- Main conclusion: Exogenous monetary policy tightening in systemically-important advanced economies leads to statistically and economically significant declines in cross-border bank lending; contrasts with previous studies using unpurged policy-rate measures.
- Future research directions:
  - Further disentangling bank lending vs. risk-taking channels in international transmission.
  - Investigating whether unconventional monetary policy effects differ from the conventional policy effects studied here.

### Appendix B — key additional exercises and findings
- Domestic effects of U.S. monetary policy shocks (responses to a 100 bp exogenous shock):
  - Output: decline.
  - Investment: decline.
  - Nominal exchange rate (NEER): appreciates.
  - CPI: weak price puzzle on impact; increase not statistically significant.
  - Domestic bank lending (Bank credit to the private non-financial sector, real): significant decline after tightening.
- Econometric issues and IV robustness:
  - Generated regressor concern addressed: when residuals are used as shocks, OLS standard errors are consistent (Pagan, 1984).
  - Instrumenting changes in policy rates with Romer and Romer (2004) and Furceri et al. (2018) shocks yields similar findings (Figure B.2).
- Mundellian trilemma analysis (local projections with time-varying D_{j,t}):
  - Exchange rate regime (Panel A): floating exchange rates do not insulate a country from cross-border spillovers.
  - Capital account openness (Panel B): capital controls seem to moderate spillovers, but differences not statistically significant.
  - Monetary policy independence (Panel C): spillovers tend to be stronger when recipient country maintains monetary policy independence.
  - Two-by-two regimes (interaction of exchange rate stability and capital openness):
    - Strongest spillovers in open-peg (fixed exchange rate and open capital account).
    - Spillovers close to zero for closed-float (floating exchange rate and closed capital account).
  - Note: average correlation between exchange rate stability index and capital openness index: -0.54 (p-value of 0.005); large standard errors in regime subsamples; adoption of regimes is endogenous.
- Figures reported in Appendix B (descriptions):
  - Figure B.1: U.S. domestic responses to a 100 bp exogenous monetary policy shock (h=0 impact; units in percentage except Federal funds rate in basis points).
  - Figure B.2: Response of cross-border bank lending to a 100 bp increase in the Federal funds rate using Romer and Romer (2004) instrument (left) and to a 100 bp increase in policy rates in other advanced economies using Furceri et al. (2018) instrument (right).
  - Figure B.3: Response to a 100 bp U.S. shock with 68% and 90% confidence bands; panels by exchange rate regime, capital openness, and monetary policy independence.
  - Figure B.4: Responses under two-by-two regimes (open-peg, open-float, closed-peg, closed-float) with 68% and 90% confidence bands; horizon h=0 captures impact; units in percentage.

*Source: IMF Working Paper — chapter "7. Effect of a monetary policy shock on cross-border bank lending from eight OECD countries" and Appendices (extracted content provided).*

### References .........................................................................................................36

### References .........................................................................................................36

### Appendices
- A. Additional figures and tables .....................................................................................41  
- B. Estimation results from additional exercises .............................................................53  

### List of Tables
- 1. Baseline estimation results from a dynamic framework ....................................................29  
- 2. Results using a static framework .......................................................................................30  
- A.1 List of countries in the final sample ....................................................................................... 41  
- A.2 Total cross-border claims and liabilities as a share of GDP ............................................... 42  
- A.3 Summary of exogenous monetary policy shocks in 9 OECD countries: 2001Q1-2012Q4
 ............................................................................................................................................................ 42  

### List of Figures
- 1. Effect of a U.S. monetary policy shock on cross-border bank lending .............................31  
- 2. Effect of a change in the Federal funds rate on cross-border bank lending .......................31  
- 3. Effect of a U.S. monetary policy shock (proxy-SVAR structural shock) on cross-border 
bank lending ...........................................................................................................................32  
- 4. Effect of a U.S. monetary policy shock on cross-border bank lending: expansions vs. 
recessions ...............................................................................................................................33  
- 5. Effect of a U.S. monetary policy shock on cross-border bank lending: low uncertainty vs. 
high-uncertainty periods ........................................................................................................34  
- 6. Effect of a U.S. monetary policy shock on cross-border bank lending: safe vs. risky 
borrower countries .................................................................................................................34  

*wpiea2019234-print-pdf - References .........................................................................................................36*

### 7. Effect of a monetary policy shock on cross-border bank lending from eight OECD

### 7. Effect of a monetary policy shock on cross-border bank lending from eight OECD countries

### Research question and approach
- Primary question: Do domestic monetary policy actions affect cross-border bank lending, and through which transmission channels (bank lending, risk-taking, portfolio rebalancing)?
- Core claim: Identifying exogenous monetary policy surprises (orthogonal to systematic policy responses) is key to reconciling mixed empirical evidence.
- Identification strategies used:
  - Narrative shocks: Romer and Romer (2004) extended by Coibion (2012) for U.S. shocks (summed within quarter).
  - High-frequency external instruments / proxy-SVAR (Gertler and Karadi, 2015; Gürkaynak et al., 2005) as robustness.
  - Forecast-error-based shocks for eight other advanced economies following Furceri et al. (2018) with additional purification via projections on current and lagged GDP growth and inflation.
- Key sample windows:
  - U.S. analysis: 1990Q1 to 2012Q4 (data availability and well-identified U.S. shocks).
  - Eight other advanced economies: 2001Q1 to 2012Q4.
- Scope of lender countries (eight): Canada, Germany, Italy, Japan, the Netherlands, Spain, Sweden, and the United Kingdom.

### Data and measurement
- Cross-border banking data: BIS Locational Banking Statistics (LBS), residency principle, currency composition available to construct exchange-rate adjusted flows and cumulated stock.
- Exchange-rate adjustment method: convert U.S. dollar-equivalent amounts to original currency using end-of-period exchange rates, compute differences in original currency, convert differences to U.S. dollar-equivalents using period-average exchange rates.
- Data processing choices:
  - Drop offshore financial centers per IMF classification.
  - Drop observations with cross-border positions less than $5 million or negative total outstanding claims.
  - Exclude dependent variable observations in upper and lower one percentile of distribution to reduce outlier influence.
- Sample coverage notes:
  - BIS LBS captures around 95 percent of all cross-border interbank business.
  - Final baseline estimation for U.S. covers cross-border lending to 45 recipient countries.
- Shock series properties:
  - Correlation between U.S. shocks from Furceri et al. (2018) and Coibion (2012): 0.62 for the overlapped sample.
  - Country-specific shocks typically have small correlation with U.S. shocks except for Canada.

### Empirical methodology
- Main estimator: Local projection method (Jordà, 2005) applied to bilateral panel data to estimate dynamic impulse responses for horizons h=0,1,2,...,7 (two years).
- Baseline regression form (simplified):
  - y_{j,t+h} - y_{j,t-1} = α_{j,h} + β_{h} MPshock_{t} + Σ γ_{h} X_{j,t-p} + ε_{j,t+h}
  - y = log exchange-rate-adjusted cross-border claims from source to recipient j.
  - X includes four lags of dependent variable, monetary shocks, recipient GDP growth, short-term interest rate, inflation, exchange rate growth, and their lags.
- Fixed effects and standard errors:
  - Recipient country fixed effects in U.S. analysis.
  - For multi-source bilateral analysis: source-recipient fixed effects and recipient-time fixed effects.
  - Standard errors clustered by time; robustness checks include clustering by recipient and Driscoll-Kraay standard errors.
- Rationale for method:
  - Exogenous shocks used obviate need for VAR identification restrictions.
  - Local projections facilitate flexible dynamics and estimation of state-dependent responses (expansion vs. recession, low vs. high uncertainty).

### Key baseline findings (U.S. monetary policy shocks)
- Sign and economic magnitudes:
  - Exogenous monetary policy tightening leads to an economically and statistically significant decline in cross-border bank lending, consistent with bank lending and risk-taking channels.
  - A 100 basis-point (bp) exogenous tightening is found to lead to more than a 10 percent decline in cross-border bank lending after two quarters.
- Representative coefficients from baseline dynamic estimates (Table 1, selected):
  - Monetary policy shock (h=1): -10.953** (standard error reported in table).
  - Monetary policy shock (h=2): -12.077**.
  - Monetary policy shock (h=0): -2.429 (not significant at conventional levels).
  - Monetary policy shock (-1) (lagged): -5.659* at h=0 and positive lagged values at other horizons (see Table 1).
  - Recipient exchange rate coefficients: negative and often statistically significant (e.g., -0.292** at h=0; -0.829*** at h=3).
- Model fit and sample size (Table 1):
  - Observations by horizon: Obs = 3,064 (h=0), 3,018 (h=1), 2,979 (h=2), 2,934 (h=3), 2,889 (h=4), 2,855 (h=5), 2,807 (h=6), 2,770 (h=7).
  - R-squared by horizon: 0.09 (h=0), 0.11 (h=1), 0.14 (h=2), 0.14 (h=3), 0.16 (h=4), 0.20 (h=5), 0.23 (h=6), 0.24 (h=7).
- Contrast with static/specification using policy-rate levels or changes:
  - Static regressions using lagged Federal funds rate or changes in Federal funds rate produce positive or mixed-signed associations with cross-border lending (Table 2, Columns I–II), consistent with prior studies that used policy-rate levels.
  - Replacing policy-rate changes with exogenous monetary policy shocks flips the sign and reveals negative effects (Table 2, Column III and dynamic results).
- Proxy-SVAR (high-frequency instrument) robustness:
  - Proxy-SVAR structural shock (example using two-year treasury yield and three-month-ahead futures instrument) confirms negative and significant decline in cross-border lending to a 100 bp structural tightening.
  - First-stage F-statistics by horizon: 330.42, 364.23, 358.67, 317.37, 332.79, 364.67, 333.84, 336.06 (indicating strong instruments).

### State-dependency, uncertainty, and asymmetries
- Business cycle state (smooth transition using five-quarter moving average of real GDP growth; θ=1.5 calibration):
  - Effect is weaker and less precisely estimated during recessions; negative effect persists and is stronger during expansions for exogenous shocks.
  - Using changes in Federal funds rate shows different signs across expansions vs. recessions, highlighting identification issues with raw policy-rate measures.
- Global uncertainty (VIX-based regimes):
  - Spillovers are stronger during low-uncertainty periods (low-VIX) and weaker during high-uncertainty periods (high-VIX), consistent with evidence of monetary policy ineffectiveness under high uncertainty.
- Tightening vs. easing:
  - Estimates using a dummy for tightening vs. easing do not show strong symmetric/asymmetric differences in average responses, though tightening during expansions may have more persistent effects than easing during recessions.

### Heterogeneity by borrower riskiness and currency-area patterns
- Borrower risk:
  - Grouping recipients by ICRG Political Risk Index (above vs. below median) or by advanced vs. emerging status yields negative and statistically significant effects for both safe and risky borrowers.
  - Magnitudes are similar across groups; difference not statistically significant, providing no support for a dominant international portfolio rebalancing channel (i.e., no evidence that tightening reallocates lending away from domestic borrowers toward safer foreign borrowers).
- Euro-area patterns:
  - For euro-area source countries, spillovers tend to be stronger toward non-euro-area borrowers than toward euro-area borrowers (subsample evidence).

### International evidence across eight advanced economies (bilateral panel)
- Regression framework: y_{i,j,t+h} - y_{i,j,t-1} = α_{i,j,h} + α_{j,t,h} + β_{h} MPshock_{i,t} + Σ γ_{h} X_{i,j,t-p} + ε
  - Source-recipient fixed effects and recipient-time fixed effects included to control time-invariant bilateral factors and recipient-specific shocks.
- Main result:
  - A one percent point (100 bp) exogenous tightening in monetary policy in the eight advanced-economy lender countries leads to an about 5 percent decline in cross-border bank lending after one quarter and a decline of more than 10 percent after one year.
  - Using changes in the policy rate (non-purged measure) shows no short-run negative effect, underscoring the importance of exogenous shock identification.
- Robustness:
  - Results robust to dropping euro-area countries other than Germany, controlling for domestic macro variables of lender countries (real GDP growth, inflation, stock returns, exchange rate growth), alternative standard error clustering, more lags, and controlling for bilateral trade flows.
  - Effects tend to be stronger in low-uncertainty periods.

### Robustness checks and controls
- Controls tested and found not to materially change results:
  - Inclusion of additional U.S. control variables (U.S. real GDP growth, inflation, stock returns).
  - Longer lag lengths (up to eight lags).
  - Alternative standard error methods: recipient clustering, recipient-time clustering, Driscoll-Kraay.
  - Controlling for bilateral trade flows (IMF Directions of Trade Statistics).
  - Controlling for global financial risk (VIX) and liquidity risk (LIBOR-OIS spread) — monetary policy shock effect remains statistically significant, implying U.S. monetary policy is an independent driver of global financial cycles.
- Data handling robustness:
  - Qualitative results robust to inclusion of observations < $5 million, alternative percentile trimming (top/bottom 2.5%), winsorizing, and including all observations.

### Interpretation and policy implications
- Mechanisms supported:
  - Evidence consistent with international bank lending channel and risk-taking channel: exogenous tightening reduces cross-border lending.
- Mechanisms not supported:
  - Little support for the international portfolio rebalancing channel at the recipient-country aggregation level (no larger lending increases toward safe borrowers).
- Policy relevance:
  - Exogenous conventional monetary tightening in systemically important advanced economies has substantial cross-border banking spillovers (declines in cross-border lending).
  - U.S. monetary policy acts as an independent source of the global financial cycle, even when controlling for VIX and LIBOR-OIS.
  - Monetary policy effectiveness in generating cross-border lending responses is state-dependent and weaker under high global uncertainty.
- Methodological contribution:
  - Demonstrates value of purged exogenous shock measures and local projections on bilateral datasets for tracing dynamic international spillovers.

### Conclusions and avenues for future research
- Main conclusion: Exogenous monetary policy tightening in systemically-important advanced economies leads to statistically and economically significant declines in cross-border bank lending; this contrasts with previous studies that used unpurged policy-rate measures.
- Future research directions proposed by authors:
  - Further disentangling bank lending vs. risk-taking channels in the international transmission.
  - Investigating whether unconventional monetary policy effects differ from the conventional policy effects studied here.

*Source: IMF Working Paper — chapter "7. Effect of a monetary policy shock on cross-border bank lending from eight OECD countries" (extracted content provided).*

### References

### References

### Major thematic areas covered by the references
- Monetary policy transmission, identification, and surprise measurement (narrative, proxy-SVAR, Fed funds futures).
- Cross-border banking, global banks, and international shock/liquidity transmission.
- Capital flows, global financial cycle, and international lending retrenchment.
- Risk, uncertainty, and the interaction of monetary policy with risk-taking, credit spreads, and financial stability.
- Fiscal multipliers and output spillovers; government spending and tax-change dynamics.
- Econometric methods relevant for the analysis (local projections, Driscoll-Kraay standard errors, weak instruments, non-linear modeling).

### Representative cited works (authors and exact titles as listed)
- Aastveit, Knut Are, Natvik, Gisle James and Sola, Sergio. “Economic uncertainty and the influence of monetary policy,” Journal of International Money and Finance, 76(C), (2017): 50-67.
- Aizenman, Joshua, Menzie David Chinn, and Hiro Ito. “The “impossible trinity” hypothesis in an era of global imbalances: Measurement and testing.” Review of International Economics 21.3 (2013): 447-458.
- Argimon, Isabel, Clemens Bonner, Ricardo Correa, Patty Duijm, Jon Frost, Jakob de Haan, Leo de Haan, and Viktors Stebunovs. “Financial institutions’ business models and the global transmission of monetary policy.” Journal of International Money and Finance 90 (2019): 99-117.
- Auerbach, Alan J., and Yuriy Gorodnichenko. “Fiscal multipliers in recession and expansion.” Fiscal policy after the financial crisis. University of Chicago Press, (2012): 63-98.
- Avdjiev, Stefan, and Galina Hale. “US monetary policy and fluctuations of international bank lending.” Journal of International Money and Finance 95 (2019): 251-268.
- Baker, Scott R., Nicholas Bloom, and Steven J. Davis. “Measuring economic policy uncertainty.” Quarterly Journal of Economics 31.4 (2016): 1593-1636.
- Bernanke, Ben S., and Mark Gertler. “Inside the black box: the credit channel of monetary policy transmission.” Journal of Economic Perspectives 9.4 (1995): 27-48.
- Borio, Claudio, and Haibin Zhu. “Capital regulation, risk-taking and monetary policy: a missing link in the transmission mechanism?” Journal of Financial Stability 8.4 (2012): 236-251.
- Bruno, Valentina, and Hyun Song Shin. “Cross-border banking and global liquidity.” Review of Economic Studies 82.2 (2015a): 535-564.
- Bruno, Valentina, and Hyun Song Shin. “Capital flows and the risk-taking channel of monetary policy.” Journal of Monetary Economics 71 (2015b): 119-132.
- Cerutti, Eugenio, Stijn Claessens, and Lev Ratnovski. “Global liquidity and cross-border bank flows.” Economic Policy 32.89 (2017): 81-125.
- Cetorelli, Nicola, and Linda S. Goldberg. “Global banks and international shock transmission: Evidence from the crisis.” IMF Economic Review 59.1 (2011): 41-76.
- Choi, Sangyup, Davide Furceri, Prakash Loungani, Saurabh Mishra, and Marcos Poplawski-Ribeiro. “Oil prices and inflation dynamics: Evidence from advanced and developing economies.” Journal of International Money and Finance 82 (2018): 71-96.
- Coibion, Olivier. “Are the effects of monetary policy shocks big or small?” American Economic Journal: Macroeconomics 4.2 (2012): 1-32.
- Correa, Ricardo, Linda S. Goldberg, and Tara Rice. “International banking and liquidity risk transmission: evidence from the United States.” IMF Economic Review 63.3 (2015): 626-643.
- Cover, James Peery. “Asymmetric effects of positive and negative money-supply shocks.” The Quarterly Journal of Economics 107.4 (1992): 1261-1282.
- Dedola, Luca, Giulia Rivolta, and Livio Stracca. “If the Fed sneezes, who catches a cold?” Journal of International Economics 108 (2017): S23-S41.
- Dell’Ariccia, Giovanni, Luc Laeven, and Gustavo A. Suarez. “Bank leverage and monetary policy’s risk‐taking channel: evidence from the United States.” Journal of Finance 72.2 (2017): 613-654.
- Furceri, Davide, Prakash Loungani, and Aleksandra Zdzienicka. “The effects of monetary policy shocks on inequality.” Journal of International Money and Finance 85 (2018): 168-186.
- Gertler, M. and P. Karadi (2015). “Monetary policy surprises, credit costs, and economic activity.” American Economic Journal: Macroeconomics 7(1), 44-76.
- Ivashina, Victoria, David S. Scharfstein, and Jeremy C. Stein. “Dollar funding and the lending behavior of global banks.” Quarterly Journal of Economics 130.3 (2015): 1241-1281.
- Jordà, Òscar. “Estimation and inference of impulse responses by local projections.” American Economic Review 95.1 (2005): 161-182.
- Kuttner, Kenneth N. “Monetary policy surprises and interest rates: Evidence from the Fed funds futures market.” Journal of Monetary Economics 47 (3) (2001): 523-44.
- Miranda-Agrippino, Silvia, and Hélène Rey (2018). US Monetary Policy and the Global Financial Cycle. NBER Working Paper No. 21722, Issued in November 2015, Revised in March 2019.
- Ramey, Valerie A. “Macroeconomic shocks and their propagation.” Handbook of Macroeconomics. Vol. 2. Elsevier, (2016). 71-162.
- Romer, Christina D., and David H. Romer. “A new measure of monetary shocks: Derivation and implications.” American Economic Review 94.4 (2004): 1055-1084.
- Tenreyro, Silvana, and Gregory Thwaites. “Pushing on a string: US monetary policy is less powerful in recessions.” American Economic Journal: Macroeconomics 8.4 (2016): 43-74.
- Wu, Jing Cynthia, and Fan Dora Xia. “Measuring the macroeconomic impact of monetary policy at the zero lower bound.” Journal of Money, Credit and Banking 48.2-3 (2016): 253-291.
- (Additional references are listed in the source; entries above are representative and quoted verbatim as in the source.)

### Appendix A items (tables and figures explicitly listed)
- Table A.1. List of countries in the final sample (includes columns: =1 if advanced economy; =1 if euro area; =1 if fully pegged; =1 if capital account is fully open; =1 if monetary policy is fully independent). Note: time-series averages computed; country with “*” denotes also a source country.
- Table A.2. Total cross-border claims and liabilities as a share of GDP. Note: values reported (example entries): Canada 88.99 claims, 66.26 liabilities; Germany 289.92 claims, 130.79 liabilities; Netherlands 524.19 claims, 469.70 liabilities; U.K. 643.95 claims, 379.29 liabilities; U.S. 63.55 claims, 49.65 liabilities. Note: data refer to 2010Q4 under locational banking statistics with the residency principle.
- Table A.3. Summary of exogenous monetary policy shocks in 9 OECD countries: 2001Q1-2012Q4. Note: reported columns include Standard deviation and correlations with U.S. monetary policy shocks (Furceri et al., 2018) and (Coibion, 2012). Example entries: Canada 0.215 sd, 0.592 correlation (Furceri), 0.441 correlation (Coibion); U.S. 0.341 sd, 1.000 (Furceri), 0.619 (Coibion).
- Figures A.1–A.15. A sequence of figures documenting: exchange-rate adjustments to U.S. cross-border bank claims; bilateral claims series; distribution of monetary policy shocks (2001Q1-2012Q4); effects of a 100 bp U.S. monetary policy shock on cross-border bank lending across multiple specifications (narrative vs proxy-SVAR; robustness checks; controls for global financial and liquidity risks; non-linear effects by tightening/easing and by recession/expansion regimes; advanced vs emerging market recipients; euro vs non-euro recipients; country group and uncertainty-split analyses). Notes attached to figures specify sample periods (for example, many series from 1990Q1 to 2012Q4) and units (responses measured as percentage; horizon h=0 captures impact).

### Empirical and methodological emphasis reflected in references and appendix
- Use of narrative shocks (Romer and Romer, Coibion), proxy-SVAR identification, and Fed funds futures as instruments.
- Local projections for impulse response estimation (Jordà).
- Robustness methods: Driscoll-Kraay standard errors; alternative lag lengths; controls for domestic variables, bilateral trade flows, global financial risks, liquidity risks.
- Non-linearity and regime dependence: recession vs expansion weighting (NBER recessions used), tightening vs easing asymmetries, and uncertainty-dependent effects.

*Source: References and Appendix A (Additional figures and tables) of wpiea2019234-print-pdf.*

### Appendix B. Estimation results from additional exercises

### Appendix B. Estimation results from additional exercises

### Domestic effect of U.S. monetary policy shocks
- Identification check: responses of domestic variables to an exogenous U.S. monetary policy tightening are generally consistent with Romer and Romer (2004) and Coibion (2012) despite some sample differences.
- Directional effects (responses to a 100 bp exogenous monetary policy shock):
  - Output: decline.
  - Investment: decline.
  - Nominal exchange rate (NEER): appreciates.
  - CPI: weak price puzzle on impact, but the increase in CPI is not statistically significant.
- Bank lending channel:
  - Domestic bank lending series used: “Bank credit to the private non-financial sector” from BIS, deflated by U.S. CPI to obtain a real measure.
  - Finding: significant decline in domestic bank lending following monetary policy tightening, consistent with the bank lending channel prediction.
  - Note on comparability: magnitude of domestic lending response cannot be directly compared to cross-border bank lending estimates in the main text because (i) counterparty entities for domestic claims are the non-financial private sector (narrower than cross-border counterparties), and (ii) main-text estimates control for a variety of fixed effects and covariates in recipient countries.

### Econometric issues: generated regressors and IV robustness
- Generated regressor concern: the shocks used are residuals (not first-stage predictors). When residuals are used as shocks, OLS standard error estimates are consistent (Pagan, 1984).
- Robustness exercise: instrumenting changes in the federal funds rate using Romer and Romer (2004)’s exogenous shocks (residuals) for the U.S.; instrumenting policy rates in other advanced economies using Furceri et al. (2018)’s exogenous shocks (residuals).
- Result: Figure B.2 confirms that findings hardly change when using the IV approach, alleviating concerns about generated regressors and allowing direct comparison with IV-based literature.

### International transmission through the Mundellian trilemma
- Motivation: debate whether floating exchange rates insulate countries from global financial cycle spillovers (Rey 2013; Miranda-Agrippino and Rey 2019) versus trilemma arrangements still matter (Aizenman et al. 2016). Han and Wei (2018) indicate role of exchange rate regime may depend on sign of center-country shocks.
- Expanded local-projection specification (equation as in source):
  y_{j,t+h} − y_{j,t−1} = α_{j}^h + β_{1}^h D_{j,t} MPshock_{t} + β_{2}^h (1−D_{j,t}) MPshock_{t}
  + ∑_{p=1}^n γ_{p}^h X_{j,t−p} + ε_{j,t+h}  (B.1)
  - D_{j,t} is an indicator of recipient country j’s trilemma status at time t.
  - X_{j,t} includes the four lags of D_{j,t} plus previous control variables.
- Advantage of local projections: allow time-varying trilemma characteristics at the country level, reducing measurement error relative to VAR studies that use time-invariant groupings.
- Trilemma measurement:
  - Use Aizenman et al. (2013) trilemma index that combines:
    - Exchange rate stability: annual standard deviation of monthly exchange rate vs. base country, normalized between zero and one.
    - Monetary policy independence: reciprocal of the annual correlation of monthly money market rates with the base country, normalized between zero and one.
    - Financial openness: updated Chinn-Ito index (KAOPEN), July 2017 update; KAOPEN is a de jure index based on four binary dummy variables codifying reported restrictions from the IMF’s Annual Report on Exchange Arrangements and Exchange Restrictions.
  - Identification note: recipient-country fixed effects absorb time-invariant characteristics; estimates identify within variation in the time-varying trilemma index.
- Empirical findings on cross-border bank lending spillovers (responses to a 100 bp U.S. monetary policy shock; units in percentage; horizons h=0 captures impact):
  - Panel A (exchange rate regime): floating exchange rate regime does not insulate a country from cross-border spillovers (consistent with Rey 2013 and Miranda-Agrippino and Rey 2019).
  - Panel B (capital account openness): capital controls seem to moderate spillovers, but differences across regimes are not statistically significant.
  - Panel C (monetary policy independence): spillovers tend to be stronger when the recipient country maintains monetary policy independence (i.e., does not increase its interest rate in response to U.S. tightening).
- Interactions and two-by-two regimes:
  - Average correlation between exchange rate stability index and capital openness index: -0.54 (p-value of 0.005), consistent with the “binding” trilemma and indicating endogeneity/collinearity concerns when ignoring joint dependence.
  - Two-by-two regime constructed from interaction of exchange rate stability index and capital account openness index:
    - Strongest monetary policy spillovers are observed in countries with a fixed exchange rate regime and open capital accounts (open-peg).
    - Spillovers are close to zero across horizons for recipient countries with floating exchange rates and closed capital accounts (closed-float).
    - Large standard errors noted due to decreased effective sample sizes in regime subsamples.
  - Interpretation: results somewhat reconcile the dilemma vs. trilemma debate — floating regimes do not universally insulate, but trilemma arrangements (exchange rate flexibility combined with capital account stance) matter for spillover magnitude.
- Caution: adoption of exchange rate regimes and capital controls is endogenous and correlated with other structural characteristics; evidence is suggestive and calls for more careful future research.

### Figures and inference details
- Figures reported:
  - Figure B.1: U.S. domestic responses to a 100 bp exogenous monetary policy shock (h=0 impact; units in percentage except Federal funds rate in basis points).
  - Figure B.2: Response of cross-border bank lending to a 100 bp increase in the Federal funds rate using Romer and Romer (2004) instrument (left panel) and to a 100 bp increase in policy rates in other advanced economies using Furceri et al. (2018) instrument (right panel). Horizon h=0 captures impact; units in percentage.
  - Figure B.3: Response of cross-border bank lending to a 100 bp U.S. monetary policy shock, showing 68% and 90% confidence bands; panels for exchange rate regime (peg vs. float), capital account openness (open vs. closed), and monetary policy independence (independent vs. dependent).
  - Figure B.4: Responses under two-by-two regimes (open-peg, open-float, closed-peg, closed-float) with 68% and 90% confidence bands; horizon h=0 captures impact; units in percentage.

*Source: Appendix B. Estimation results from additional exercises, wpiea2019234-print-pdf*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2019/wpiea2019234-print-pdf.pdf_