## _wp1469 - 1. Variable Definitions and Sources

## Source details

**Canonical URL:** [_wp1469 - 1. Variable Definitions and Sources](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2014/_wp1469.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2014/_wp1469.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2014/_wp1469.pdf.json)

---

### Introduction: scope and motivation
- The financial cycle is becoming increasingly global, reflected in higher correlation of credit growth across countries since the mid 90s.
- Large amounts of funds flow from ‘financial center’ economies (G4: US, Euro Area, UK, Japan) to the rest of the world.
  - Example: in June 2013, cross-border bank claims by the G4 on the rest of the world exceeded claims by the rest of the world on G4 banks by 20 percent.
- International banks in the G4 intermediate much of cross-border credit between countries in the rest of the world; funding conditions within the G4 therefore affect funding conditions globally.
- Formal definition: Global liquidity enters the cross-border financing supply function Qs = Q(P, GL), where Qs is the quantity of cross-border financing provided, P is the “price” (expected return differentials), and GL is a vector of “non-price” supply factors.

### Data and empirical setup
- Sample: 77 countries over the period 1990-2012; series exchange-rate-adjusted; sectoral breakdown into claims on banks and non-banks.
- Main data source: BIS International Banking Statistics (IBS) — BIS Locational data (BIS IBS Table 6).
- Global liquidity drivers compiled separately for each G4 (US, UK, Euro Area, Japan):
  - Stock option market implied volatility (CBOE VIX and analogues).
  - US dealer bank leverage; non-US G4 commercial bank leverage.
  - TED spread (3 month Libor minus 3 month government bond yield).
  - Slope of yield curve (10 year government bond yield minus 3 month government bond yield).
  - Real policy rate (deflated with CPI).
  - Monetary aggregates (M2).
  - Real credit growth in G4 countries.
- Controls:
  - Lagged GDP growth and inflation in borrower countries.
  - Interest rate differential (ΔInterestSpread).
  - Borrower characteristics: exchange rate flexibility index, capital controls, institutional environment, bank regulation (capital adequacy, supervisory powers, limits on foreign bank presence).
- Base panel regression (country fixed effects; standard errors clustered at borrower country level):
  - ΔLjt = β0 + β1 DomesticFactorjt + β2 ΔInterestSpreadjt + β3 GlobalLiquidityt + γj + εjt
  - Dependent variable ΔLjt: quarterly difference in log of exchange-rate-adjusted stock of bank claims in borrower country j at time t.
  - Two dependent series used:
    - (i) change in log stock of BIS Locational cross-border claims on borrower country j banking sector.
    - (ii) change in log stock of BIS Locational cross-border claims on borrower country j non-bank sector.
- Extensions include interaction terms: BorrowerCharacteristicsjt and GlobalLiquidityt * BorrowerCharacteristicsjt.

### Key empirical findings: determinants of cross-border bank flows
- Robust global liquidity correlates:
  - Term premia in the US, UK and Euro Area have a robust negative association with flows (consistent with a “search for yield” mechanism).
  - Bank leverage in non-US G4 countries and real domestic credit growth in the G4 are positively associated with cross-border bank flows (except for Japan in some measures).
- Measures not robust across specifications:
  - Short-term interest rates and growth in money aggregates (e.g., M2) are not robust across all specifications.
- Role of non-US G4 countries:
  - UK and Euro Area supply factors are important globally, not just regionally.
  - For banking sector conditions (bank leverage and TED spreads), the importance of UK and Euro Area often exceeds that of US banking sector conditions.
  - For monetary policy variables, the US continues to play a dominant role.
- Time coverage:
  - Many relations appear strongly in the 2000s financial globalization period; main results driven by the 2001-2012 sub-period.
- Differential sensitivity of banks vs non-banks:
  - Changes in VIX and dealer bank leverage affect claims on banks more than on non-banks.
- Implied marginal effects (25th to 75th percentile changes):
  - Change in VIX from 25th to 75th percentile reduces cross-border claims on banks by 5¾ percent and by 3½ percent for non-banks.
  - Similar change in US dealer bank leverage increases cross-border claims on banks by 5½ percent and by 4½ percent for non-banks.
  - Increase in term premium from its 25th to 75th percentile decreases cross-border claims on banks by 1¾ percent and by ¼% for non-banks.

### Conceptual review: channels and indicators of global liquidity
- Global liquidity captures non-price supply factors faced by providers of cross-border funds (mainly major international banks).
- Key indicators and channels:
  - Uncertainty and risk aversion: proxied by the US VIX.
  - Funding conditions for global banks: proxied by TED spread and bank leverage (US dealer banks and non-US G4 commercial banks).
  - Monetary policy in the G4: level of interest rates and slope of yield curve; term premium influences search-for-yield incentives.
  - Money aggregates: changes in M2 may affect cross-border lending buoyancy though channels are less clear.
- Cross-border lending expected to be more volatile than domestic lending due to asymmetric information, reliance on hard information, and sovereign risk.

### The role of U.S. versus other G4 drivers
- Motivation: evaluate relative explanatory power of US, UK, Euro Area, and Japan variables (many series are highly correlated; factors often compared individually).
- Main comparative findings (2001-2012, reduced sample excluding G4 borrowers):
  - VIX: similar explanatory power across G4; US VIX slightly higher R2.
  - Bank conditions:
    - UK TED spread has twice the explanatory power of US TED spread; US TED spread often not significant.
    - UK bank leverage has higher explanatory power than US bank leverage.
    - Euro Area real credit growth has higher explanatory power than US real credit growth.
  - Monetary policy:
    - US and UK real policy rates, and US/UK/Euro Area term premia have similar explanatory power.
    - Japan policy rate not significant; slope of yield curve significant with positive sign (likely reflecting Japan’s low, stable rates).
  - G4 M2:
    - UK and Euro Area M2 growth positively associated with cross-border credit.
    - US and Japan M2 growth show negative sign (interpreted as flight to safety / deleveraging).
    - Aggregate G4 M2 positive coefficient largely driven by UK and Euro Area M2.
- Regional effects:
  - UK and Euro Area drivers have explanatory power for borrowers in Asia and the Western Hemisphere.
  - UK and Euro Area TED spreads and UK bank leverage often have similar or higher explanatory power than US counterparts for these regions.
- Summary interpretation:
  - Global financial cycle largely driven by US monetary policy and UK/Euro Area bank conditions, reflecting significant roles of European banks in financial intermediation.

### Borrower-country characteristics and interaction effects
- Borrower policies and characteristics mitigate exposure to global liquidity shocks:
  - Exchange rate flexibility reduces cyclicality of inflows to banks and non-banks.
  - Capital controls reduce cyclical exposure to global liquidity.
  - More stringent bank supervision reduces cyclical exposure.
  - More stringent capital requirements:
    - Make cross-border flows less cyclical.
    - Reduce the level of cross-border inflows to banks.
  - Better institutional quality:
    - Increases the level of inflows.
    - Can increase cyclicality of inflows to banks for some global liquidity factors.
  - Fewer limits on foreign bank presence increase cyclicality of inflows to banks.
- Magnitude example (US dealer bank leverage, 25th to 75th percentile):
  - Country with capital controls at 25th percentile: growth in cross-border claims of about 19% (7% for flows to non-banks).
  - Country with capital controls at 75th percentile: growth in cross-border claims of about 10% (5% for flows to non-banks).
- Overall quantitative effect:
  - An increase from the 25th to 75th percentile in policy indexes (exchange rate flexibility, capital controls, bank supervision/regulation) reduces exposures by at least half, particularly for bank-directed cross-border flows.

### Regression highlights and summary statistics (selected exact figures)
- Sample period: 1990Q1–2012Q4.
- Summary statistics (Panel A, selected):
  - Log cross-border claims on banks: Obs 5,448; Mean 1.61; Median 1.30; Std. Dev. 10.43; P25 -3.08; P75 6.20; Min -42.62; Max 43.83.
  - Log cross-border claims on non-banks: Obs 5,420; Mean 1.44; Median 1.12; Std. Dev. 6.88; P25 -1.96; P75 4.55; Min -22.21; Max 27.15.
  - GDP Growth (lag): Obs 5,446; Mean 3.87; Median 3.79; Std. Dev. 4.78; P25 1.58; P75 6.28; Min -20.34; Max 24.50.
  - CBOE VIX: Obs 5,448; Mean 21.21; Median 20.18; Std. Dev. 9.00; P25 14.91; P75 24.97; Min 11.11; Max 68.51.
  - US Bank Leverage: Obs 5,448; Mean 19.11; Median 19.80; Std. Dev. 4.91; P25 14.60; P75 22.14; Min 8.91; Max 30.62.
  - G4 Countries M2: Obs 5,448; Mean 6.06; Median 5.96; Std. Dev. 5.80; P25 1.62; P75 10.83; Min -7.02; Max 18.63.
- Selected regression coefficients (Table 4, full sample, dependent = Log Changes in Cross-Border Claims on Banks):
  - GDP Growth (lag): coefficient 0.227***.
  - Inflation (lag): coefficient -0.0981***.
  - CBOE VIX (when included): coefficient -0.184***.
  - US Bank Leverage (when included): coefficient 0.279***.
  - Growth of Real US Credit (when included): coefficient 0.191***.
  - US Slope of Yield Curve (when included): coefficient -0.645***.
  - G4 Countries M2 (Annual growth rate): coefficient 0.105***.
  - Observations: 5,448; Number of countries: 77; R-squared (Model 1) 0.013.
- Selected regression coefficients (Table 4, full sample, dependent = Log Changes in Cross-Border Claims on Non-Banks):
  - GDP Growth (lag): coefficient 0.182***.
  - Inflation (lag): coefficient -0.0223 (not always significant).
  - CBOE VIX (when included): coefficient -0.118***.
  - US Bank Leverage (when included): coefficient 0.223***.
  - Growth of Real US Credit (when included): coefficient 0.195***.
  - US Slope of Yield Curve (when included): coefficient -0.669***.
  - G4 Countries M2 (Annual growth rate): coefficient 0.0382**.
  - Observations: 5,420; Number of countries: 77; R-squared (Model 1) 0.015.
- Significance convention in tables: *** 1 percent, ** 5 percent, * 10 percent.

### Conclusions and policy implications
- Confirmatory findings:
  - Multiple global liquidity factors drive cross-border bank flows alongside country-specific factors.
  - Cross-border bank flows decrease with higher VIX and steeper U.S. yield curve slope and increase with higher U.S. dealer bank leverage, higher real interest rates, and higher G4 money growth.
- New findings:
  - Bank conditions in U.K. and Euro Area (bank leverage, TED spreads) also drive cross-border bank flows and can be more important than equivalent U.S. conditions.
  - Global financial cycle largely driven by U.S. monetary policy and U.K./Euro Area bank conditions.
- Policy lessons:
  - Monitor term premia and banking-sector conditions in all G4 economies, not only the US.
  - Recipient-country toolkit to reduce exposure:
    - Exchange rate flexibility.
    - Capital flow management tools.
    - Stronger bank supervision and regulation.
  - Policies are particularly effective at reducing exposures of cross-border flows to banks; non-bank exposures are less affected by standard banking regulation.
- Research and surveillance needs:
  - Better understanding required of channels through which financial conditions affect global risk-taking, capital flows, and vulnerabilities, and of best measures of global liquidity.
  - Continuous review of empirically useful indicators warranted given evolving institutions, policies, innovations, and market structures.
- Broader assessment:
  - Cross-border flows bring benefits (support local activity during stress) and risks (procyclical amplification, asset price booms, financial fragility).
  - Domestic policy effectiveness can be weakened in volatile global conditions; favorable global conditions can contribute to vulnerability buildup.
  - Monitoring liquidity, funding, and credit conditions in systemic financial institutions is critically important.

*Source: _wp1469 - 1. Variable Definitions and Sources*

### 1. Variable Definitions and Sources ....................................................................................

### _wp1469 - 1. Variable Definitions and Sources

### Introduction: scope and motivation
- The financial cycle is becoming increasingly global, reflected in higher correlation of credit growth across countries since the mid 90s.
- A large amount of funds flows from ‘financial center’ economies (G4: US, Euro Area, UK, Japan) to the rest of the world. For example, in June 2013, cross-border bank claims by the G4 on the rest of the world exceeded claims by the rest of the world on G4 banks by 20 percent.
- International banks in the G4 intermediate much of cross-border credit between countries in the rest of the world; funding conditions within the G4 therefore affect funding conditions globally.
- Global liquidity is defined as factors in ‘financial center’ economies that affect the provision of cross-border credit; formally, Qs = Q(P, GL), where Qs is the quantity of cross-border financing provided, P is the “price” (expected return differentials), and GL is a vector of “non-price” supply factors.

### Data and empirical setup
- Dataset covers 77 countries over the period 1990-2012, with adjustments for exchange rate changes.
- Analysis focuses on cross-border bank flows, distinguishing claims on banks and non-banks.
- Empirical objectives:
  - Identify which measures of G4 financial conditions best capture global liquidity.
  - Assess whether global liquidity is primarily US-driven or whether other G4 countries play a significant role.
  - Evaluate how borrower-country policies and characteristics affect exposure to variations in global liquidity.

### Key empirical findings
- Measures robustly associated with cross-border bank flows:
  - Term premia in the US, UK and Euro Area have a robust negative association with flows, consistent with the “search for yield” mechanism (a flatter yield curve reduces domestic profit opportunities and can encourage cross-border lending).
  - Two proposed proxy measures—bank leverage in non-US G4 countries and real domestic credit growth in the G4—are positively associated with cross-border bank flows (except for Japan).
- Measures not robust across specifications:
  - Short-term interest rates and growth in money aggregates (e.g., M2) are not robust across specifications in explaining cross-border bank flows.
- Role of non-US G4 countries:
  - UK and Euro Area supply factors are important globally, not just regionally.
  - For banking sector conditions (bank leverage and TED spreads), the importance of UK and Euro Area often exceeds that of US banking sector conditions.
  - For monetary policy variables, the US continues to play a dominant role.
- Time and coverage:
  - Many of the documented relations appear strongly in the 2000s financial globalization period.
- Borrower-country heterogeneity and policy mitigants:
  - Borrower countries can reduce exposure to global liquidity variations by adopting:
    - More flexible exchange rate regimes.
    - Capital flow management tools.
    - More stringent bank supervision and regulation.
  - Quantified policy effect: an increase from the 25th to 75th percentile in the policy indexes for any of these dimensions reduces exposures by at least half.
  - The impact is higher for cross-border flows to banks versus non-bank borrowers.

### Conceptual review: determinants and channels of global liquidity
- Global liquidity (supply-side ease of funding) reflects non-price supply factors faced by providers of cross-border funds (mainly major international banks).
- Key indicators and channels identified in the literature:
  - Uncertainty and risk aversion: commonly proxied by the US VIX (stock option implied volatility).
  - Funding conditions for global banks: proxied by the TED spread and bank leverage (often measured for US dealer banks; authors add non-US G4 bank leverage).
  - Monetary policy in the G4: includes level of interest rates and slope of the yield curve; the term premium affects banks’ incentives to search for yield.
  - Money aggregates: changes in M2 may affect cross-border lending buoyancy, though channels are less clear (components like wholesale deposits may complement leverage measures).
- Cross-border lending is expected to be more volatile than domestic lending due to higher asymmetric information, reliance on hard information, and sovereign risk.

### Relation to existing literature and contributions
- Complements Bruno and Shin (2014) by using additional measures (bank leverage and real credit growth in the G4) and longer time series.
- Clarifies the roles of individual G4 countries (US, UK, Euro Area, Japan) and shows stronger roles for UK and Euro Area in banking sector supply factors.
- Confirms Rey (2013) on capital controls’ effectiveness, while identifying additional effective tools: stricter bank regulation and supervision.
- Builds on “push” and “pull” factor literature explaining cross-border capital flows by adding detailed treatment of borrower-country characteristics and policy responses.

### Policy implications and practical guidance
- Source-country policy focus:
  - Monitor and understand term premia and banking-sector conditions in all G4 economies, not only the US, because these affect global cross-border credit supply.
- Borrower-country policy toolkit to limit exposure:
  - Exchange rate flexibility: promotes insulation from cross-border liquidity swings.
  - Capital flow management tools: can reduce capital flow sensitivity to global liquidity variations.
  - Stronger bank supervision and regulation: substantially reduces bank-related cross-border exposures.
- Targeting:
  - Policies are particularly effective at reducing exposures of cross-border flows to banks; non-bank exposures are less affected by standard banking regulation.
  - Countries with better institutions or larger foreign bank presence may be more exposed to global liquidity and thus stand to gain more from these policy measures.

*Source: _wp1469 - 1. Variable Definitions and Sources*

### 2012. In addition, we investigate which borrower countries’ policies and characteristics (e.g.,

### _wp1469 - 2012. In addition, we investigate which borrower countries’ policies and characteristics (e.g.,

### A. Data
- Data source: BIS International Banking Statistics (IBS) — BIS Locational data (BIS IBS Table 6) used for analysis.
- Rationale for dataset choice:
  - Locational data conform closer to notion that conditions in specific ‘financial center’ countries affect flows.
  - Locational data provide a long time span; Consolidated data often consistently available only from mid-2000s.
  - Locational data provide exchange rate adjusted series and sectoral breakdown of lending to banks and non-banks.
- Sample coverage:
  - 77 countries over the period 1990-2012.
- Global liquidity drivers used (compiled separately for each of the G4: US, UK, Euro Area, Japan):
  - Stock option market implied volatility (CBOE VIX).
  - US dealer bank leverage.
  - TED spread (3 month Libor minus 3 month government bond yield).
  - Slope of yield curve (10 year government bond yield minus 3 month government bond yield).
  - Real policy rate (deflated with CPI).
  - Monetary aggregates (M2).
  - Two additional credit condition measures: bank leverage and credit growth in G4 countries.
- Controls:
  - Credit demand: lagged GDP growth rate and inflation in borrower countries.
  - Price determinants: differential between local and international interest rates (ΔInterestSpread).
  - Additional borrower characteristics: exchange rate flexibility index, capital controls, institutional environment, bank regulation (capital adequacy, supervisory powers, limits on foreign bank presence).
- Supporting tables referenced: Table 1 (definitions and sources), Tables 2 and 3 (summary statistics and correlations), Table 3 Panel B (correlations of global liquidity factors across G4).

### B. Empirical Specification
- Base panel regression (country fixed effects; standard errors clustered at borrower country level):
  - ΔLjt = β0 + β1 DomesticFactorjt + β2 ΔInterestSpreadjt + β3 GlobalLiquidityt + γj + εjt
  - Dependent variable ΔLjt: quarterly difference in log of exchange-rate-adjusted stock of bank claims in borrower country j at time t.
  - Two dependent series used:
    - (i) change in log stock of BIS Locational cross-border claims on borrower country j banking sector.
    - (ii) change in log stock of BIS Locational cross-border claims on borrower country j non-bank sector.
- Extensions:
  - Introduce BorrowerCharacteristicsjt and interaction terms:
    - ΔLjt = β0 + β1 DomesticFactorjt + β2 ΔInterestSpreadjt + β3 GlobalLiquidityt + β4 BorrowerCharacteristicsjt + β5 (GlobalLiquidityt * BorrowerCharacteristicsjt) + γj + εjt
  - BorrowerCharacteristicsjt includes: (i) exchange rate regime type; (ii) use of capital controls; (iii) institutional development; (iv) bank regulatory variables.

### C. Base Results: Drivers of Global Liquidity
- Country demand factors:
  - Lagged GDP growth and inflation are statistically significant in explaining cross-border flows to banks (Table 4, Panel A) for 1990-2012 and several sub-periods.
  - Lagged inflation is somewhat less significant for flows to non-bank borrowers (Panel B).
- Interest rate differentials:
  - Coefficients of changes in interest rate differentials are not statistically different from zero in full sample (potentially due to some developing countries where interest rates are not market-based).
- Individual US global liquidity factors (when considered individually) — statistical significance and signs:
  - VIX and TED spreads: negative coefficients (cross-border flows decrease during times of uncertainty).
  - US dealer bank leverage: positive coefficient (banks expand more cross-border when funding conditions are accommodative).
  - Real credit growth: positive sign (captures leverage and financial cycle).
  - US real interest rate: positive sign (during less favorable conditions—when interest rates are lower—global banks lend less cross-border).
  - US term premium: negative coefficient (search-for-yield incentives; when US investment opportunities are more attractive, cross-border flows decline).
  - G4 M2 growth: positively associated with cross-border flows (aggregate).
- Multivariate results (include most drivers simultaneously):
  - VIX, US dealer bank leverage, and the term premium remain statistically significant determinants of changes in cross-border claims on banks and non-banks (columns 12-13, Table 4).
  - M2 growth affects claims on banks but not on non-banks.
  - Changes in VIX and dealer bank leverage affect claims on banks more than on non-banks (banks more sensitive to financial conditions).
- Sub-period findings:
  - Results for full 1990-2012 sample largely driven by 2001-2012 sub-period.
  - Results for pre-global financial crisis period (2001-2006) are similar to 2001-2012, suggesting crisis and aftermath do not drive main results.
- Implied economic effects (marginal effects; 25th to 75th percentile changes):
  - Change in VIX from 25th to 75th percentile reduces cross-border claims on banks by 5¾ percent and by 3½ percent for non-banks.
  - Similar change in US dealer bank leverage increases cross-border claims on banks by 5½ percent and by 4½ percent for non-banks.
  - Increase in term premium from its 25th to 75th percentile decreases cross-border claims on banks by 1¾ percent and by ¼% for non-banks.
- Interpretation:
  - US monetary policy stance is a less important driver of global liquidity than market uncertainty or bank funding conditions.
  - Market conditions are more important than direct government actions in driving global liquidity.

### D. The Role of U.S. versus Other G4 Drivers
- Motivation:
  - Assess whether US variables are most relevant or whether UK, Euro Area, and Japan also play important roles.
  - Non-US G4 leverage measured using commercial bank leverage (instead of dealer bank leverage).
  - Many non-US G4 series are highly correlated with US series; therefore factors compared individually.
  - Reduced sample excludes US, UK, Japan, and Euro Area borrowers to capture cross-border impacts.
- Main findings (Table 5; estimations for 2001-2012):
  - VIX: similar explanatory power across G4; US VIX has slightly higher R2.
  - Bank conditions:
    - UK and Euro Area variables often have the same or higher explanatory power than US variables.
    - UK TED spread has twice the explanatory power of US TED spread; US TED spread not significant and lowest R2.
    - UK bank leverage has higher explanatory power than US bank leverage.
    - Euro Area credit growth has higher explanatory power than US credit growth.
  - Monetary policy:
    - US and UK real policy rates, and US/UK/Euro Area term premia have similar explanatory power.
    - Japan: policy rate not significant; slope of yield curve significant with positive sign (likely due to stable, low interest rates in Japan).
  - G4 M2 measures:
    - UK and Euro Area M2 growth positively associated with cross-border credit (consistent with bank intermediation role).
    - US and Japan M2 growth show negative sign (interpreted as reflecting flight to safety / deleveraging).
    - Aggregate G4 M2 positive coefficient driven largely by UK and Euro Area M2.
- Regional effects and robustness (Table 6):
  - UK and Euro Area global liquidity drivers have explanatory power for borrowers beyond their own region (Asia and Western Hemisphere).
  - UK and Euro Area TED spreads have higher explanatory power for lending to Asia and Western Hemisphere than US TED spread.
  - UK bank leverage has similar or higher explanatory power than US bank leverage for these regions.
- Dominant role findings:
  - U.S. monetary policy factors remain dominant in some cases:
    - US term premium is the only variable with explanatory power for cross-border lending to both banks and non-banks in Asia.
    - US, UK, and Euro Area term premia have similar significance for the Western Hemisphere.
  - Suggestion: global financial cycle driven in large part by US monetary policy and UK/Euro Area bank conditions (consistent with prominent role of European banks in intermediating funds).

### E. Borrower Country Characteristics
- Analysis: impact of borrower country policies and characteristics and interactions with global liquidity drivers (Table 7).
- Main results:
  - Flexible exchange rate regime reduces borrower exposures to variation in key global liquidity drivers (dealer bank leverage, term premium, M2 growth) — inflows become less cyclical.
  - Capital controls reduce cyclical exposure to global liquidity.
  - More stringent bank supervision reduces cyclical exposure.
  - More stringent capital requirements:
    - Make cross-border flows less cyclical.
    - Reduce the level of cross-border inflows to banks.
  - Better institutional quality:
    - Increases the level of inflows.
    - Increases cyclicality of inflows to banks for some global liquidity factors.
  - Fewer limits on foreign bank presence increase cyclicality of inflows to banks.
- Magnitude example:
  - When US dealer bank leverage increases from its 25th to 75th percentile:
    - Country with capital controls at 25th percentile: growth in cross-border claims of about 19% (7% for flows to non-banks).
    - Country with capital controls at 75th percentile: growth in cross-border claims of about 10% (5% for flows to non-banks).
  - Effects of similar magnitude present for exchange rate flexibility and stringency of bank capital regulation and supervision.
  - Overall: changes along these policy dimensions reduce borrower country banks’ exposures to cyclical variations in global liquidity roughly by half.
- Policy implication: flexible FX regimes, capital flow management tools, and more stringent bank regulation/supervision can reduce a country’s exposure to global liquidity swings, particularly for more open countries with better institutions.

### IV. Conclusions and Policy Implications
- Confirmatory findings:
  - Multiple global liquidity factors drive cross-border bank flows alongside country-specific factors.
  - Cross-border bank flows:
    - Decrease with higher VIX and steeper U.S. yield curve slope.
    - Increase with higher U.S. dealer bank leverage, higher real interest rates, and higher G4 money growth.
- New findings:
  - Bank conditions in other financial center countries (U.K. and Euro Area), notably bank leverage and TED spreads, also drive cross-border bank flows and can be more important than equivalent U.S. conditions.
  - Global financial cycle largely driven by U.S. monetary policy and U.K./Euro Area bank conditions.
- Policy lessons:
  - Domestic financial conditions in all financial center economies can affect the rest of the world via cross-border bank flows.
  - Major economies should consider effects of their policies on other countries (feedback effects), though internalizing externalities is difficult in practice.
  - Recipient countries have policy options to reduce exposure to global liquidity (macroeconomic management, regulatory environment adjustments).
  - Monitoring liquidity, funding, and credit conditions in systemic financial institutions is critically important.
- Research and surveillance needs:
  - State of the art limited on channels through which financial conditions affect global risk-taking, capital flows, vulnerabilities, and on best measures of global liquidity.
  - Need better understanding of drivers of liquidity in advanced economies and mechanisms of international propagation/amplification of financial shocks.
  - Continuous review of empirically useful indicators is warranted given evolving institutions, policies, innovations, and market structures.
- Broader assessment:
  - Cross-border flows bring both benefits (support local activity during stress) and risks (procyclical amplification, asset price booms, financial fragility).
  - In volatile global conditions, domestic monetary and fiscal policies can become less effective; favorable global conditions can contribute to build-up of vulnerabilities.
  - Policy responses may need adaptation both domestically and internationally because systemic financial institutions’ distress can propagate widely.

*Source: Excerpt from provided IMF working paper content.*

### References

### _wp1469 - References

### Key References (selected from the bibliography)
- Adrian, T., and H. S. Shin, 2010, “Liquidity and leverage,” Journal of Financial Intermediation, Vol. 19, No. 3, pp. 418-37.
- Altunbas, Y., L. Gambacorta, and D. Marqués-Ibáñez., 2014, “Does Monetary Policy Affect Bank Risk-Taking?” International Journal of Central Banking, March, 10:1, 95-135.
- Borio, C.E., and P. Lowe, 2004, “Securing sustainable price stability: should credit come back from the wilderness?” BIS Working Papers No. 157.
- Brunnermeier, M. K., and L.H. Pedersen, 2009, “Market liquidity and funding liquidity. Review of Financial studies”, 22(6), 2201-2238.
- Committee on the Global Financial System (CGFS), 2011, “Global Liquidity—Concept, Measurement and Policy Implications,” CGFS Papers no 45, December.
- International Monetary Fund, 2010, Global Financial Stability Report, “Global Liquidity Expansion: Effects on `Receiving’ Economies and Policy Response Options,” April (Washington).
- Shin, H.S., 2012, “Global Banking Glut and Loan Risk Premium” Mundell‐Fleming Lecture, IMF Economic Review 60 (2), 155‐92.
- Turner, P., 2014, “The global long-term interest rate, financial risks and policy choices in EMEs,” BIS Working Papers no. 441, February.
(References list continues in source.)

### Variable definitions and sources (Table 1 — selected variables)
- Dependent variables:
  - Log cross-border claims on banks: Log Changes in BIS Locational Cross-Border Claims on Banks (exchange rate adjusted) — BIS Locational statistics (Table 6).
  - Log cross-border claims on non-banks: Log Changes in BIS Locational Cross-Border Claims on Banks (exchange rate adjusted) — BIS Locational statistics (Table 6).
- Global drivers (selected):
  - Real GDP Growth: Growth rate of real GDP — WEO.
  - Inflation: Inflation — IFTSTSUB and GDS.
  - Interest rate Differential: Difference between domestic rate and Fed funds rate — IFTSTSUB.
  - Exchange rate flexibility: Ranges from 1-4, with higher values indicating more flexibility — Ilzetzki, Reinhart and Rogoff (2008).
  - Capital controls: Higher values of the index represent more restrictions — Quinn (2011).
  - Institution quality: The average of bureaucracy quality; law and order; corruption; investment profile. Higher values indicate lower quality — International Country Risk Guide.
  - G4 VIX indicators: US VIX — CBOE S&P500 Volatility VIX; UK VIX — FTSE 100 Volatility Index; EA VIX — VDAX Volatility Index (new); JP VIX — NIKKEI Stock Average Volatility Index — Datastream.
  - Note: 1/ Data on Euro Government AAA 3-month bill is available since 2007, so the period 2001-2006 is based on the 3 month French treasury bill rate.

### Summary statistics and sample composition (Table 2 — selected figures)
- Sample period: 1990Q1–2012Q4.
- Panel A — Summary statistics (Obs, Mean, Median, Std. Dev., P25, P75, Min, Max):
  - Log cross-border claims on banks: 5448, Mean 1.61, Median 1.30, Std. Dev. 10.43, P25 -3.08, P75 6.20, Min -42.62, Max 43.83.
  - Log cross-border claims on non-banks: 5420, Mean 1.44, Median 1.12, Std. Dev. 6.88, P25 -1.96, P75 4.55, Min -22.21, Max 27.15.
  - GDP Growth (lag): 5446, Mean 3.87, Median 3.79, Std. Dev. 4.78, P25 1.58, P75 6.28, Min -20.34, Max 24.50.
  - Inflation (lag): 5447, Mean 5.06, Median 3.29, Std. Dev. 6.18, P25 1.83, P75 6.32, Min -2.80, Max 70.59.
  - CBOE VIX: 5448, Mean 21.21, Median 20.18, Std. Dev. 9.00, P25 14.91, P75 24.97, Min 11.11, Max 68.51.
  - US Bank Leverage: 5448, Mean 19.11, Median 19.80, Std. Dev. 4.91, P25 14.60, P75 22.14, Min 8.91, Max 30.62.
  - G4 Countries M2: 5448, Mean 6.06, Median 5.96, Std. Dev. 5.80, P25 1.62, P75 10.83, Min -7.02, Max 18.63.
- Panel B — Selected correlations (full sample):
  - Correlation of GDP Growth (lag) with Inflation (lag): 0.01.
  - Correlation of CBOE VIX with US TED Spread: 0.40.
  - Correlation of US Bank Leverage with Growth of Real US Credit: 0.68.
  - Correlation of Real US Federal Fund Rate with US Slope of Yield Curve: -0.63.

- Panel C — Regional distribution: sample includes 77 countries; G4 members flagged where relevant (United States 1/, Japan 1/, United Kingdom 1/, France 1/, etc.). (Full country listing in source.)

### Individual G4 variables — summary (Table 3 — 2001Q1–2012Q4, selected)
- US VIX: Obs 2503, Mean 22.22, Std. Dev. 10.38, Min 11.24, Max 68.51.
- UK VIX: Obs 2503, Mean 21.13, Std. Dev. 8.40, Min 10.12, Max 49.57.
- EA VIX: Obs 2503, Mean 25.36, Std. Dev. 9.75, Min 12.70, Max 57.94.
- JP VIX: Obs 2503, Mean 26.26, Std. Dev. 9.07, Min 15.48, Max 65.49.
- US real policy rate: Obs 2503, Mean -0.47, Std. Dev. 1.71, Min -3.63, Max 3.32.
- US bank leverage: Obs 2503, Mean 20.00, Std. Dev. 5.50, Min 12.43, Max 30.62.
- US growth rate of M2: Obs 2503, Mean 6.35, Std. Dev. 2.26, Min 1.27, Max 10.54.

### Regression highlights — cross-border claims (Table 4 — selected coefficients, significance)
- Panel A — Dependent variable: Log Changes in BIS Locational Cross-Border Claims on Banks (in %):
  - GDP Growth (lag): coefficient 0.227*** (Model 1, 1990-2012).
  - Inflation (lag): coefficient -0.0981*** (Model 1).
  - CBOE VIX (when included): coefficient -0.184***.
  - US Bank Leverage (when included): coefficient 0.279***.
  - Growth of Real US Credit (when included): coefficient 0.191***.
  - US Slope of Yield Curve (when included): coefficient -0.645***.
  - G4 Countries M2 (Annual growth rate): coefficient 0.105***.
  - Observations: 5,448 (full-sample), Number of countries: 77, R-squared (Model 1) 0.013.

- Panel B — Dependent variable: Log Changes in BIS Locational Cross-Border Claims on Non-Banks (in %):
  - GDP Growth (lag): coefficient 0.182*** (Model 1).
  - Inflation (lag): coefficient -0.0223 (Model 1; not always significant).
  - CBOE VIX (when included): coefficient -0.118***.
  - US Bank Leverage (when included): coefficient 0.223***.
  - Growth of Real US Credit (when included): coefficient 0.195***.
  - US Slope of Yield Curve (when included): coefficient -0.669***.
  - G4 Countries M2 (Annual growth rate): coefficient 0.0382**.
  - Observations: 5,420 (full-sample), Number of countries: 77, R-squared (Model 1) 0.015.

Notes: Table 4 reports panel regressions with country fixed effects and clustered standard errors at the borrower country level. Significance: *** 1 percent, ** 5 percent, * 10 percent.

### G4 individual variable effects (Table 5 — coefficients introduced individually)
- Panel A — Claims on Banks (coefficients, significance):
  - US VIX: -0.251*** (US); -0.258*** (UK); -0.243*** (EA); -0.271*** (JP).
  - US Bank Leverage: 0.364*** (US).
  - US Real Credit Growth: 0.284*** (US).
  - US Real Policy Rate: 0.446*** (US).
  - US Slope of yield curve: -1.309*** (US).
  - US M2 growth: -0.879*** (US).

- Panel B — Claims on Non-Banks (coefficients, significance):
  - US VIX: -0.163*** (US); -0.163*** (UK); -0.162*** (EA); -0.157*** (JP).
  - US Bank Leverage: 0.264*** (US).
  - US Real Credit Growth: 0.288*** (US).
  - US Real Policy Rate: 0.636*** (US).
  - US Slope of yield curve: -1.234*** (US).
  - US M2 growth: -0.523*** (US).

Notes: Only non-G4 countries included in these estimations, sample reduced to 58 countries (2,503 observations). Variables in the table were introduced individually; regressions also include lag GDP growth, lag CPI inflation, and change in interest rate differentials (not reported).

### Regional G4 factor effects (Table 6 — selected coefficients by region)
- Claims on Banks — Asia vs. Western Hemisphere (selected coefficients):
  - US TED spreads: Asia -2.817**; Western Hemisphere -0.908.
  - UK TED spreads: Asia -5.640***; Western Hemisphere -5.006***.
  - EA TED spreads: Asia -5.091***; Western Hemisphere -1.698**.
  - UK bank leverage: Asia 0.409*; Western Hemisphere 0.667**.
  - US growth of M2: Asia -0.841**; Western Hemisphere -0.744*.
- Claims on Non-banks — Asia vs. Western Hemisphere (selected coefficients):
  - UK TED spreads: Asia -3.845***; Western Hemisphere -2.142**.
  - EA real credit growth: Western Hemisphere 0.190***.
  - US slope of yield curve: Asia -1.161***; Western Hemisphere -1.027**.
  - EA slope of yield curve: Western Hemisphere -0.556***.

Notes: Each region estimated separately; only non-G4 countries included. Variables introduced individually; regressions include lag GDP growth, lag CPI inflation, and change in interest rate differentials (not reported). Significance levels: *** 1 percent, ** 5 percent, * 10 percent.

### Interaction effects with country characteristics (Table 7 — selected interaction coefficients)
- Panel A — Dependent: Log Changes in Claims on Banks (in %):
  - Exchange rate flexibility baseline coefficients (selected): UK real policy rate 4.180***; UK slope of yield curve 1.969**.
  - Exchange rate flexibility * X interactions (selected): with US Dealer Bank Leverage -0.132***; with UK real policy rate -0.270***; with UK slope of yield curve 0.802***; with G4 Countries M2 -0.0689***.
  - Capital controls * X interactions (selected): with US Dealer Bank Leverage -0.00518**; with UK real policy rate -0.0139***; with UK slope of yield curve 0.0301***; with G4 Countries M2 -0.00346**.
  - Institution quality (high values indicate lower institutional quality): baseline -3.834***; interaction with UK slope of yield curve 0.606***; interaction with G4 Countries M2 -0.0484**.
  - Limits on foreign banks baseline: UK real policy rate 5.303**; interaction with UK real policy rate -0.404**; interaction with UK slope of yield curve 1.091***.

- Panel B — Dependent: Log Changes in Claims on Non-Banks (in %):
  - Exchange rate flexibility * X interactions (selected): with US Dealer Bank Leverage -0.0988***; with UK real policy rate -0.171***; with UK slope of yield curve 0.450***; with G4 Countries M2 -0.0517***.
  - Capital controls * X interactions (selected): with US Dealer Bank Leverage -0.00395***; with UK real policy rate -0.0107***; with UK slope of yield curve 0.0218***; with G4 Countries M2 -0.00166*.
  - Supervisory power * X interactions (selected): with US Dealer Bank Leverage -0.0185**; with UK slope of yield curve 0.0695*; with G4 Countries M2 -0.00769*.
  - Institution quality baseline: -3.197***; institution quality * X (selected): with US Dealer Bank Leverage -0.0491**; with UK real policy rate -0.110**; with UK slope of yield curve 0.249**.

Notes: Interaction regressions include lag GDP growth, lag CPI inflation, change in interest rate differentials, and the respective interacted variable. High values of institution quality indicate lower institutional quality. Significance: *** 1 percent, ** 5 percent, * 10 percent.

*Source: _wp1469 - References (IMF Working Paper content excerpt — references, tables, and regression results).*

### Annex A. Time series charts of the drivers of global liquidity

### Annex A. Time series charts of the drivers of global liquidity

### Overview
- Annex A presents time series charts of the drivers of global liquidity.

*Source: Annex A. Time series charts of the drivers of global liquidity*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2014/_wp1469.pdf_