## _wp14180 - References

## Source details

**Canonical URL:** [_wp14180 - References](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2014/_wp14180.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2014/_wp14180.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2014/_wp14180.pdf.json)

---

### Introduction and objectives
- Context: Aggregate gross foreign banking claims as of end-2013 were some 20 percent below their pre-crisis peak in June 2008 of USD 30 trillion.
- Two main objectives:
  - Analyze the role of supply and lender–borrower factors in driving changes in international bank lending during the GFC, controlling for demand using bilateral data.
  - Identify motivations and constraints driving banks’ cross-border rebalancing between direct cross-border lending (headquarters → foreign borrower) and local foreign affiliates’ lending (subsidiaries and branches).
- Empirical observation motivating study: local affiliate lending declined by only 5 percent during the GFC compared to 23 percent for direct cross-border lending (aggregate differences noted in the introduction).

### Event, data, and methodology
- Event analyzed: the intense GFC period starting mid-2008 through June 2009.
- Main data source: BIS consolidated banking statistics (BIS CBS) on ultimate risk basis, adjusted for coverage break-in-series and exchange rate variations.
  - Adjusted totals: Total (adjusted) foreign claims about USD 25 trillion in mid-2012, down from above USD 30 trillion in mid-2008, for reporting banking systems in the sample.
  - Local affiliate lending share: about 50 percent of total foreign lending as of 2012, compared to a 40 percent share before the crisis.
- Identification strategy: difference-in-difference / cross-sectional approach with borrower country fixed effects (γj) to control for borrower demand; explanatory variables measured at end-2007 (ex ante).
- Main supply-side variable: SRISK measure (Acharya et al. (2010)) aggregated to creditor-country level and scaled by creditor country GDP, measured end-2007.
- Additional supply-side variables: banking-system accounting variables at end-2007 — return on assets (ROA), ratio of non-performing loans (NPL) to total gross loans, ratio of risk-weighted assets to total assets; Laeven and Valencia (2013) systemic crisis dummy (as of mid-2009).
- Bilateral lender–borrower variables (pre-episode): log distance between capital cities, geographical adjacency dummy, common language dummy, legal origin dummy, colonial past dummy, bilateral trade as proportion of lender country overall trade, pre-episode direct cross-border exposure share.

### Hypothesized scenarios for direct cross-border vs affiliates’ claims
- Scenario A (no intra-group frictions): Negative shock to home banking system i reduces both direct cross-border lending to borrower j and affiliates’ lending in j (funds flow from affiliates back to HQ), though responses need not be proportional.
- Scenario B (“ring-fencing”): Limitations on intra-group transfers cause a much larger reduction in direct cross-border lending than in affiliates’ lending because headquarters cannot tap affiliates’ liquidity/capital; authorities may require supported banks to “lend at home.”
- Scenario C (attempted bypass): Limits on internal movements lead banks to cut direct cross-border lending sharply but affiliates’ lending increases to partially compensate, an indirect bypass of host-country ring-fencing.

### Key empirical findings — supply-side determinants
- Aggregate bilateral medians for 2008Q2–2009Q2:
  - Median change in direct cross-border lending across lender–borrower pairs: -16 percent.
  - Median change in affiliates’ lending across lender–borrower pairs: 2 percent increase.
- SRISK effects on direct cross-border lending:
  - SRISK_GDP coefficients (selected from Table 2, Panel A): -0.0518*, -0.0746***, -0.0869*, -0.0834*, -0.0774***, -0.0735***, -0.0878**, -0.0842**.
  - Economic interpretation: coefficient in column 1 implies a one unit increase in SRISK translates into approximately a 0.05 percent decline in direct cross-border lending.
  - Given SRISK standard deviation is about 97.10, a one standard deviation increase in SRISK translates into a 5 percent decline in direct cross-border lending.
  - SRISK effects more pronounced for lender banking systems in North America and Asia; European banks adjust cross-border lending less in response to SRISK in some specifications.
- Accounting measures:
  - ROA, NPL ratio, and risk-weighted assets/total assets generally not statistically significant predictors of deleveraging in cross-border or affiliate lending across most specifications.
  - Combining accounting measures with market-based SRISK: SRISK remains significant; accounting measures remain insignificant.
- Affiliates’ lending:
  - SRISK_GDP coefficients in Table 2, Panel B are not consistently significant (examples: -0.00682, 0.0149, 0.0231).
  - SRISK_GDP* WH and SRISK_GDP* Asia often positive and significant for affiliates (examples: 0.436***, 0.284***).
  - Presence of a systemic crisis in the home country associated with declines in affiliates’ lending; some specifications show a net effect of a 17 percent decline in affiliate claims.
- Matched-sample evidence (creditor systems with both claim types to same borrower):
  - SRISK continues to affect direct cross-border lending more than affiliates’ lending (Table 3: SRISK_GDP -0.0646**, -0.103***, -0.100**).
  - Substitution effects between affiliate and cross-border lending exist but are smaller when creditor SRISK is higher.
    - Quantitative example: for banking systems with SRISK at the 75th percentile, a one standard deviation increase in affiliates’ lending would reduce direct cross-border lending by 1¼ percent; for systems at the 25th percentile, it would reduce direct cross-border lending by 4¾ percent.
  - Interpretation: evidence consistent with increased financial frictions (e.g., ring-fencing or regulatory limits) when lender banking systems are more vulnerable.

### Creditor–borrower determinants (bilateral factors)
- Pre-episode exposure:
  - Banks decreased direct cross-border lending more to countries where they had high pre-episode exposures (Table 4A, column 1); effect interpreted as rebalancing or easier deleveraging in high-exposure markets.
  - This effect not present for affiliate lending (Table 4B, column 1).
- Historical and cultural ties:
  - Direct cross-border lending: less reduction to borrower countries with recent (post-1945) colonial ties (col451 positive and significant in multiple columns; examples: 7.80***, 25.20***).
  - Affiliates’ lending: common language associated with less reduction (comlang_off1 mean 0.12; median 0.00).
- Geography and trade:
  - Contiguous borders reduced cross-border lending cutbacks in some specifications (contig1 mean 0.02; median 0.00).
  - Distance: larger distance generally associated with larger reductions in direct and affiliate lending (distlog mean 8.47; median 8.77), though not always significant.
  - Bilateral trade share (bitrade mean 0.15; median 0.01): effects ambiguous; trade intensity sometimes associated with fewer cutbacks; trade finance channel may be important.
- Overall: lender–borrower characteristics help explain deleveraging heterogeneity but add substantially less to R-squared than supply-side variables, particularly for direct cross-border lending.

### Robustness and additional checks
- Winsorized regressions (5th and 95th percentiles) and clustering choices (single vs double) yield robust results.
- Dollar liquidity / funding proxies (McGuire and Goetz (2009) short-term dollar funding need measure, change in market-to-book ratio of banks’ equity, average spread in overnight swap rate) not consistently significant; inclusion reduces sample sizes.
- Regional splits of borrower countries:
  - SRISK remains important for direct cross-border lending across regions.
  - For “other” borrowers (Middle East and Africa), pre-crisis NPLs, ROA, and relative portfolio risk further reduced direct cross-border lending.
  - Home-bias effect on affiliates’ claims driven by deleveraging to European borrowers in some specifications.
- Alternative specification with creditor-country fixed effects (instead of SRISK): similar results for creditor–borrower determinants.
- Additional checks reported: winsorized variables (Logdiff_Cross-border_wins and Logdiff_Affiliates_wins), matched-sample panels, creditor and borrower fixed effects.

### Key statistics and selected summary measures
- Summary statistics (Period 2008Q2-09Q2, Table 1):
  - Logdiff_Cross-border: Obs 1,858; Mean -0.18; Median -0.16; Std. Dev. 0.83; Min -5.35; Max 6.42.
  - Logdiff_Affiliates: Obs 846; Mean 0.00; Median 0.04; Std. Dev. 0.89; Min -5.47; Max 6.17.
  - SRISK_GDP: Obs 24,788; Mean 33.45; Median 8.24; Std. Dev. 97.10; Min 0.00; Max 416.28.
  - systemic_crisis: Obs 24,788; Mean 0.45; Median 0.00; Std. Dev. 0.50; Min 0.00; Max 1.00.
  - roa: Obs 24,788; Mean 0.80; Median 0.74; Std. Dev. 0.46; Min 0.10; Max 1.83.
  - npl: Obs 24,788; Mean 1.65; Median 1.10; Std. Dev. 1.52; Min 0.20; Max 5.30.
  - rwa_assets: Obs 24,788; Mean 0.50; Median 0.51; Std. Dev. 0.14; Min 0.19; Max 0.77.
  - share_cross-border: Obs 29,440; Mean 0.78; Median 0.02; Std. Dev. 2.72; Min 0.00; Max 43.87.
- Representative regression magnitudes (selected):
  - Table 2, Panel A (Direct Cross-Border Claims): SRISK_GDP examples -0.0518*, -0.0746***, -0.0869*; Observations 1,858; R-squared examples 0.120–0.139.
  - Table 2, Panel B (Local Affiliates Claims): SRISK_GDP examples -0.00682, 0.0149, 0.0231; Observations 846; R-squared examples 0.166–0.169.
  - Table 3 (Matching Samples): SRISK_GDP -0.0646**, -0.103***, -0.100**; system crisis -16.63***, -15.96***; Observations e.g., 831.
  - Table 4 (Creditor-Borrower Determinants): col451 examples 7.80***, 25.20***; share_cross-border examples -0.789*.
  - Table 8 (With Creditor and Borrower FE): col451 27.20***, 31.22***, 31.30***; distlog examples -11.47*; Observations examples 1,873 and 836.

### Figures and visual evidence (captions and sources)
- Figure 2: "Bilateral evolution of banks’ claims during 2008Q2-09Q2 — Cross-Border Claims / Affiliates’ claims." Source: BIS, IFS, and authors’ estimations.
- Figure 3: "Evolution of Cross-Border and Affiliates' Claims — Period 2008Q2-09Q2." Source: BIS, IFS, and authors’ estimations.
- Figure 4: "Supply Side Variables." Sources: NYU Stern, IFS, Central Banks, and authors’ estimations.
  - Example figure axis labels preserved in source: "SRISK as of Dec 2007  (as % of GDP)"; "roa (in %)"; "npl (in %)"; "rwa_assets (ratio, rhs)".
  - Example country ordering visible on axes: LU, PT, FI, AU, GR, CA, IT, AT, SP, US, JP, DN, IR, DE, SE, BE, ND, UK, FR, CH.

### Main conclusions and policy implications
- Major findings:
  - Deleveraging largely varied with ex ante, market-based measures of banking-system vulnerabilities (SRISK), while traditional accounting measures were not consistently significant.
  - Creditor–borrower characteristics (concentration, cultural and geographical proximity, trade links) played secondary roles relative to supply-side drivers.
  - Evidence of barriers to cross-border movement of resources within banking groups: supply-side factors explaining reductions in direct cross-border lending differ from those explaining reductions in affiliates’ lending.
  - Substitution from cross-border to affiliate lending occurred more readily for banking systems with lower vulnerabilities; for more vulnerable systems substitution was weaker, consistent with “ring-fencing” or intra-group flow constraints.
- Policy recommendations and implications:
  - Financial stability and assessment frameworks should incorporate forward-looking, market-based measures (such as SRISK) because backward-looking accounting measures can be poor guides to the risk of deleveraging.
  - Borrower countries should consider the type and origin of cross-border lending: direct cross-border lending is much more sensitive to market supply factors than lending by foreign affiliates.
  - More ex ante coordination among bank regulators could reduce ring-fencing and unilateral measures that impede intra-group flows and thereby limit sharp contractions in direct cross-border lending during periods of financial turmoil.

*Source: IMF Working Paper chapter “4. Supply Side Variables” (GFC deleveraging analysis).*

### References .............................................................................................................

### _wp14180 - References

### Tables
- 1. Summary Statistics (Period 2008Q2-09Q2) ........................................................................26
- 2. Deleveraging during 2008Q2-09Q2: Supply Side Determinants ........................................27
- 3. Deleveraging during 2008Q2-09Q2: Supply Side Determinants with Matching Samples .28
- 4. Deleveraging during 2008Q2-09Q2: Creditor-Borrower Determinants .............................29
- 5. Deleveraging during 2008Q2-09Q2: Creditor-Borrower Determinants with 
 Matching Samples ...........................................................................................................30
- 6. Robustness: Supply Side Determinants and Regional Breakdowns ...................................31
- 7. Robustness: Creditor-Borrower Determinants and Regional Breakdowns .........................32
- 8. Deleveraging during 2008Q2-09Q2: Creditor-Borrower Determinants with 
  Creditor and Borrower FE ..............................................................................................33

### Figures
- 1. BIS Reporting Banks’ Adjusted Foreign Claims ...............................................................34
- 2. Bilateral Evolution of Banks’ Claims During 2008Q2-09Q2............................................35
- 3. Evolution of Cross-Border and Affiliates’ Claims ............................................................36

*Source: _wp14180 - References*

### 4. Supply Side Variables ...............................................................................................

### 4. Supply Side Variables

### Introduction and objectives
- Context: Aggregate gross foreign banking claims as of end-2013 were some 20 percent below their pre-crisis peak in June 2008 of USD 30 trillion.
- Two main objectives:
  - Analyze the role of supply and lender–borrower factors in driving changes in international bank lending during the GFC, controlling for demand using bilateral data.
  - Identify motivations and constraints driving banks’ cross-border rebalancing between direct cross-border lending (headquarters → foreign borrower) and local foreign affiliates’ lending (subsidiaries and branches).
- Key empirical observation motivating study: local affiliate lending declined by only 5 percent during the GFC compared to 23 percent for direct cross-border lending (aggregate differences noted in the introduction).

### Event studied, data, and methodology
- Event analyzed: the intense GFC period starting mid-2008 through June 2009.
- Main data source: BIS consolidated banking statistics (BIS CBS) on ultimate risk basis, adjusted for coverage break-in-series and exchange rate variations.
  - Adjusted totals: Total (adjusted) foreign claims were about USD 25 trillion in mid-2012, down from above USD 30 trillion in mid-2008, for the reporting banking systems in the sample.
  - Local affiliate lending share: about 50 percent of total foreign lending as of 2012, compared to a 40 percent share before the crisis.
- Identification strategy: difference-in-difference / cross-sectional approach with borrower country fixed effects (γj) to control for borrower demand; explanatory variables measured at end-2007 (ex ante).
- Main supply-side variable: SRISK measure (Acharya et al. (2010)) aggregated to creditor-country level and scaled by creditor country GDP, measured end-2007.
- Additional supply-side variables: banking-system accounting variables at end-2007 — return on assets (ROA), ratio of non-performing loans (NPL) to total gross loans, ratio of risk-weighted assets to total assets; Laeven and Valencia (2013) systemic crisis dummy (as of mid-2009).
- Bilateral lender–borrower variables used (measured pre-crisis): log distance between capital cities, geographical adjacency dummy, common language dummy, legal origin dummy, colonial past dummy, bilateral trade as proportion of lender country overall trade, pre-episode direct cross-border exposure share.

### Hypothesized scenarios for direct cross-border vs affiliates’ claims
- Scenario A (no intra-group frictions): Negative shock to home banking system i reduces both direct cross-border lending to borrower j and affiliates’ lending in j (funds flow from affiliates back to HQ), though responses need not be proportional.
- Scenario B (“ring-fencing”): Limitations on intra-group transfers cause a much larger reduction in direct cross-border lending than in affiliates’ lending because headquarters cannot tap affiliates’ liquidity/capital; authorities may require supported banks to “lend at home.”
- Scenario C (attempted bypass): Limits on internal movements lead banks to cut direct cross-border lending sharply but affiliates’ lending increases to partially compensate, an indirect bypass of host-country ring-fencing.

### Key empirical findings — supply-side determinants
- Aggregate bilateral medians for 2008Q2–2009Q2:
  - Median change in direct cross-border lending across lender–borrower pairs: -16 percent.
  - Median change in affiliates’ lending across lender–borrower pairs: 2 percent increase.
- SRISK results (Table 2, Panel A; cross-border lending):
  - SRISK is statistically significant and negative in base regressions and many specifications.
  - Economic interpretation: the coefficient in column 1 implies a one unit increase in SRISK translates into approximately a 0.05 percent decline in direct cross-border lending.
  - Given SRISK standard deviation is about 97, a one standard deviation increase in SRISK translates into a 5 percent decline in direct cross-border lending.
  - SRISK effects more pronounced for lender banking systems in North America and Asia (2008–09 deleveraging), with European banks (base case) adjusting cross-border lending less in response to SRISK in some specifications.
- Accounting measures (ROA, NPL ratio, risk-weighted assets/total assets):
  - Generally not statistically significant predictors of deleveraging in cross-border or affiliate lending across most specifications.
  - Combining accounting measures with market-based SRISK: SRISK remains significant; accounting measures remain insignificant.
- Affiliated lending (Table 2, Panel B):
  - Supply factors generally less important for affiliates’ lending; SRISK not always statistically significant and sometimes positive.
  - For lender banking systems from North America and Asia, affiliates’ lending adjusted somewhat upwards in response to perceived capital shortfalls — evidence of segmentation and substitution (local operations acting as substitutes for cross-border lending).
  - Presence of a systemic crisis in the home country: associated with declines in affiliates’ lending; net effect for affiliates in some specifications is a 17 percent decline in affiliate claims (noted as similar to direct cross-border lending decline when government support induced home bias).
- Matched-sample results (creditor systems with both cross-border and affiliate lending to same borrower — Table 3):
  - Qualitatively similar results: SRISK affects direct cross-border lending more than affiliates’ lending.
  - Interaction tests: substitution effects exist between affiliate and cross-border lending, but substitution is smaller when creditor banking system SRISK is higher.
    - Quantitative example: for banking systems with SRISK at the high 75th percentile, a one standard deviation increase in affiliates’ lending would reduce direct cross-border lending by 1¼ percent; for banking systems with SRISK at the low 25th percentile, it would reduce direct cross-border lending by 4¾ percent.
  - Interpretation: evidence consistent with increased financial frictions (e.g., ring-fencing or regulatory limits) when lender banking systems are more vulnerable.

### Creditor–borrower determinants (bilateral factors)
- Pre-episode exposure (share of cross-border claims to a borrower):
  - Banks decreased direct cross-border lending more to countries where they had high pre-episode exposures (Table 4A, column 1) — interpreted as rebalancing or easier deleveraging in high-exposure markets.
  - This effect is not present for affiliate lending (Table 4B, column 1).
- Historical and cultural ties:
  - Direct cross-border lending: less reduction to borrower countries with recent (post-1945) colonial ties (Table 4A, column 5).
  - Affiliates’ lending: common language associated with less reduction (Table 4B, column 2).
- Geography and trade:
  - Contiguous borders: presence lowered reduction in direct cross-border lending in some specifications.
  - Distance: larger distance generally associated with larger reductions in direct and affiliate lending, though not always statistically significant.
  - Bilateral trade share: ambiguous effects; trade intensity sometimes associated with fewer cutbacks, but when distance is added results vary; trade finance channel may be important during crisis (banks cutting trade finance).
- Overall: lender–borrower characteristics help explain deleveraging heterogeneity but explain substantially less of the variation (R2 contribution) than supply-side variables, particularly for direct cross-border lending.

### Robustness and additional checks
- Winsorized regressions (5th and 95th percentiles), single clustering vs double clustering: results robust.
- Dollar liquidity / funding proxies: McGuire and Goetz (2009) short-term dollar funding need measure, change in market-to-book ratio of banks’ equity, average spread in overnight swap rate — not consistently statistically significant, and sample sizes drop substantially when included.
- Regional splits of borrower countries (Asian, European, Western Hemisphere, other): SRISK remains important for direct cross-border lending across regions.
  - For “other” borrowers (Middle East and Africa), pre-crisis NPLs, ROA, and relative portfolio risk further reduced direct cross-border lending.
  - Home-bias effect on affiliates’ claims driven by deleveraging to European borrowers in some specifications.
- Alternative specification with creditor-country fixed effects (instead of SRISK): similar results for creditor–borrower determinants.

### Main conclusions and policy implications
- Major findings:
  - Deleveraging largely varied with ex ante, market-based measures of banking-system vulnerabilities (SRISK), while traditional accounting measures were not consistently significant.
  - Creditor–borrower characteristics (concentration, cultural and geographical proximity, trade links) played secondary roles relative to supply-side drivers.
  - Suggestive evidence of barriers to cross-border movement of resources within banking groups: supply-side factors explaining reductions in direct cross-border lending differ from those explaining reductions in affiliates’ lending.
  - Substitution from cross-border to affiliate lending occurred more readily for banking systems with lower vulnerabilities; for more vulnerable systems substitution was weaker, consistent with “ring-fencing” or intra-group flow constraints.
- Policy recommendations and implications:
  - Financial stability and assessment frameworks should incorporate forward-looking, market-based measures (such as SRISK) because backward-looking accounting measures can be poor guides to the risk of deleveraging.
  - Borrower countries should consider the type and origin of cross-border lending: direct cross-border lending is much more sensitive to market supply factors than lending by foreign affiliates.
  - More ex ante coordination among bank regulators could reduce ring-fencing and unilateral measures that impede intra-group flows and thereby limit sharp contractions in direct cross-border lending during periods of financial turmoil.

*Source: IMF Working Paper chapter “4. Supply Side Variables” (GFC deleveraging analysis).*

### References

### _wp14180 - References

### References (selected from source)
- Aiyar, S., C. Calomiris, J. Hooley, Y. Korniyenko, and T. Wieladek, 2014, “The International Transmission of Bank Capital Requirements: Evidence from the UK,” Journal of Financial Economics, 113, 368–382.
- Aiyar, S. C. Calomiris, and T. Wieladek, 2014, “Identifying Channels of Credit Substitution When Bank Capital Requirements Are Varied,” Bank of England Working Paper 485.
- Acharya, V., C. Brownlees, R. Engle, F. Farazmand, and M. Richardson, 2010, “Measuring Systemic Risk” in Acharya, Viral, Thomas Cooley, and Mathew Richardson (eds.), Regulating Wall Street: The Dodd-Frank Act and the New Architecture of Global Finance, John Wiley and Sons.
- Cerutti, E., 2014, “Drivers of Cross-Border Banking Exposures During the Crisis,” Journal of Banking and Finance, forthcoming.
- Cerutti, E., G. Hale, and C. Minoiu, 2014, “Financial Crisis and the Composition of Cross-border Lending,” Presented at USC-JIMF Conference “Financial adjustment in the aftermath of the global crisis 2008–09: New global order?” Los Angeles, April.
- Cerutti, E., and C. Schmieder, 2014, “Ring Fencing and Consolidated Banks’ Stress Tests,” Journal of Financial Stability, 11, 1-12.
- Cetorelli, N., and L. Goldberg, 2011, “Global Banks and International Shock Transmission: Evidence from the Crisis,” IMF Economic Review, 59, 41-76.
- Cetorelli and Goldberg, 2012a, “Liquidity Management of US Global Banks: Internal Capital Markets in the Great Recession,” Journal of International Economics, 88, 299-311.
- Cetorelli and Goldberg, 2012b, “Banking Globalization and Monetary Transmission,” Journal of Finance, 67(5), 1811-1843.
- Chor, D., and K. Manova, 2012, “Off the Cliff and Back? Credit Conditions and International Trade during the Global Financial Crisis,” Journal of International Economics, 87, 117-33.
- Claessens, S., and N. van Horen, 2013, “Impact of Foreign Banks,” Journal of Financial Perspectives, 1 (1), 1-14.
- Claessens, S., and N. van Horen, 2014a, “Foreign Banks: Trends and Impact” Journal of Money Credit and Banking, 46(1), 295-326.
- Claessens, S., and N. van Horen, 2014b, “The Impact of the Global Financial Crisis on Banking Globalization,” IMF Working Paper 14/.
- Cull, R., and M. S. Martinez Peria, 2012, “Bank Ownership and Lending Patterns During the 2008-2009 Financial Crisis: Evidence from Latin America and Eastern Europe,” World Bank Policy Research Working Paper, No. 6195.
- De Haas, R., and I. Van Lelyveld, 2010, “Internal capital markets and lending by multinational bank subsidiaries,” Journal of Financial Intermediation, 19, 689-721.
- De Haas, R., and I. Van Lelyveld, 2014, “Multinational banks and the global financial crisis: weathering the perfect storm,” Journal of Money, Credit and Banking, 46(1), 333-364.
- De Haas, R., and N. van Horen, 2012, “International Shock Transmission after the Lehman Brothers Collapse: Evidence from Syndicated Lending,” American Economic Review, 102, 231-37.
- De Haas, R., and N. van Horen, 2013, “Running for the Exit? International Bank Lending During a Financial Crisis,” Review of Financial Studies, 26(1), pp. 244-85.
- D’Hulster, K., 2014, “Ring-fencing Cross Border Banks: How is it Done and It is Important?” World Bank, mimeo.
- Giannetti, M., and L. Laeven, 2012, “The Flight Home Effect: Evidence from Syndicated Loan Market during Financial Crises,” Journal of Financial Economics, 104(1), 23-43.
- Hale, G., T. Kapan, and C. Minoiu, 2014, “Crisis Transmission in the Global Banking Network,” paper presented at the Concluding Conference of the Macro-prudential research (MaRs) Network of the European System of Central Banks (June 23–24, Frankfurt).
- Idier, J, G. Lamé, and J-S. Mésonnier, 2013, “How Useful is the Marginal Expected Shortfall for the Measurement of Systemic Exposure? A Practical Assessment,” ECB Working Paper 1546.
- Ivashina, V., and D. Scharfstein, 2010, “Bank Lending during the Financial Crisis,” Journal of Financial Economics, 97(3), 319-38.
- Kamil, Herman and Kulwant Rai, 2010, The Global Credit Crunch and Foreign Banks‘ Lending to Emerging Markets: Why Did Latin America Fare Better?, IMF Working Paper 10/02.
- Kapan, T., and C. Miniou, 2013, “Balance Sheet Strength and Bank Lending During the Global Financial Crisis,” IMF Working Paper 13/102.
- Kerl, C. and F. Niepmann, 2014, “What determines the Composition of International Bank Flows,” mimeo, Deutsche Bundesbank and Federal Reserve Bank of New York.
- Khwaja, A., and A. Mian, 2008, “Tracing the Effect of Bank Liquidity Shocks: Evidence from Emerging Markets,” American Economic Review, 98(4), 1413-42.
- Laeven, L., and F. Valencia, 2013 “Systemic Banking Crises Database," IMF Economic Review, 61(2), 225-270.
- López-Espinos, G., A. Moreno, A. Rubia, and L. Valderrama, 2012, “Short-term Wholesale Funding and Systemic Risk: A Global CoVaR,” Journal of Banking and Finance, 36(12), 3150-162.
- McGuire, P., and G. von Peter, 2009, “The US Dollar Shortage in Global Banking,” BIS Quarterly Review, March.
- Ongena, S., J. L. Peydro, and N. van Horen, 2013, “Shocks Abroad, Pain at Home? Bank-firm Level Evidence on Financial Contagion During the 2007-2009 Crisis,” De Nederlandsche Bank, Working Paper 385.
- World Bank 2009. “Global Development Finance. Charting a Global Recovery,” Washington D.C.

### Summary Statistics (Table 1: Period 2008Q2-09Q2)
- Logdiff_Cross-border: Obs 1,858; Mean -0.18; Median -0.16; Std. Dev. 0.83; Min -5.35; Max 6.42
- Logdiff_Cross-border_wins: Obs 1,858; Mean -0.19; Median -0.16; Std. Dev. 0.76; Min -2.63; Max 2.39
- Logdiff_Affiliates: Obs 846; Mean 0.00; Median 0.04; Std. Dev. 0.89; Min -5.47; Max 6.17
- Logdiff_Affiliates_wins: Obs 846; Mean -0.01; Median 0.04; Std. Dev. 0.78; Min -3.25; Max 2.72
- SRISK_GDP: Obs 24,788; Mean 33.45; Median 8.24; Std. Dev. 97.10; Min 0.00; Max 416.28
- SRISK_GDP* Asia: Obs 24,788; Mean 2.64; Median 0.00; Std. Dev. 10.55; Min 0.00; Max 48.24
- SRISK_GDP* WH: Obs 24,788; Mean 2.78; Median 0.00; Std. Dev. 8.90; Min 0.00; Max 37.71
- systemic_crisis: Obs 24,788; Mean 0.45; Median 0.00; Std. Dev. 0.50; Min 0.00; Max 1.00
- roa: Obs 24,788; Mean 0.80; Median 0.74; Std. Dev. 0.46; Min 0.10; Max 1.83
- npl: Obs 24,788; Mean 1.65; Median 1.10; Std. Dev. 1.52; Min 0.20; Max 5.30
- rwa_assets: Obs 24,788; Mean 0.50; Median 0.51; Std. Dev. 0.14; Min 0.19; Max 0.77
- comlang_off1: Obs 24,788; Mean 0.12; Median 0.00; Std. Dev. 0.33; Min 0.00; Max 1.00
- comlegal1: Obs 24,788; Mean 0.31; Median 0.00; Std. Dev. 0.46; Min 0.00; Max 1.00
- contig1: Obs 24,788; Mean 0.02; Median 0.00; Std. Dev. 0.15; Min 0.00; Max 1.00
- col451: Obs 24,788; Mean 0.02; Median 0.00; Std. Dev. 0.15; Min 0.00; Max 1.00
- bitrade: Obs 24,440; Mean 0.15; Median 0.01; Std. Dev. 0.59; Min 0.00; Max 10.19
- distlog: Obs 24,788; Mean 8.47; Median 8.77; Std. Dev. 0.93; Min 4.09; Max 9.88
- share_cross-border: Obs 29,440; Mean 0.78; Median 0.02; Std. Dev. 2.72; Min 0.00; Max 43.87

### Key Regression Results (Selected coefficients from Tables 2–8)
Notes common to regression tables: Cross-sectional regressions estimated; dependent variable changes are log differences between end of deleveraging episode and its start (coefficients reported in percentage). Standard errors double clustered by creditor bank and borrower country. Robust standard errors in parentheses. Significance: *** p<0.01, ** p<0.05, * p<0.1.

- Table 2 — Panel A (Dependent variable: Log Changes in Direct Cross-Border Claims):
  - SRISK_GDP coefficients across columns: -0.0518*, -0.0746***, -0.0869*, -0.0834*, -0.0774***, -0.0735***, -0.0878**, -0.0842** (standard errors in parentheses shown in table)
  - SRISK_GDP* WH: consistently negative and significant (e.g., -0.973***, -1.044***, -0.944***)
  - SRISK_GDP* Asia: consistently negative and significant (e.g., -0.307***, -0.427***, -0.383***)
  - Borrower FE: Yes in all columns
  - Observations: 1,858; R-squared (varies) e.g., R-squared 0.120, 0.133, 0.118, 0.135, 0.137, 0.135, 0.137, 0.137, 0.139

- Table 2 — Panel B (Dependent variable: Log Changes in Local Affiliates Claims):
  - SRISK_GDP coefficients not consistently significant (examples: -0.00682, 0.0149, 0.0231)
  - SRISK_GDP* WH: positive and significant in several columns (e.g., 0.436***)
  - SRISK_GDP* Asia: positive and significant in several columns (e.g., 0.284***)
  - system crisis negative and significant in some specifications (e.g., -14.45*, -15.32**, -15.53**, -16.22**)
  - Observations: 846; R-squared ranges (e.g., R-squared 0.166, 0.169, 0.167, 0.169)

- Table 3 (Supply Side Determinants with Matching Samples):
  - SRISK_GDP: negative and significant across columns (e.g., -0.0646**, -0.103***, -0.100**)
  - SRISK_GDP* WH: negative and significant (e.g., -0.973***)
  - SRISK_GDP* Asia: coefficients range; some negative and significant (e.g., -0.131***)
  - system crisis: large negative and significant (e.g., -16.63***, -15.96***)
  - Inclusion of Change in Affiliate Claims: coefficient -0.0283 (not significant) and -0.0868** (significant) in different columns
  - Observations: e.g., Panel A Observations 831; Panel B Observations 831

- Table 4 (Creditor-Borrower Determinants):
  - Panel A (Direct Cross-Border Claims) SRISK_GDP: -0.0744*** to -0.0797*** across columns
  - SRISK_GDP* WH: -0.948*** to -0.829*** across columns
  - SRISK_GDP* Asia: -0.306*** to -0.266***
  - share_cross-border: estimated coefficients sometimes negative (example -0.789*), varying significance by column
  - col451 (colonial relationship after 1945): positive and significant in multiple columns (e.g., 7.80***, 25.20***)
  - Panel B (Local Affiliates Claims) SRISK_GDP near zero or positive (e.g., 0.0148, 0.0161)
  - SRISK_GDP* WH positive and significant (e.g., 0.464***)
  - SRISK_GDP* Asia positive and significant (e.g., 0.287***)
  - Observations: Panel A 1,858; Panel B 846; R-squared examples: Panel A R-squared 0.133, Panel B R-squared 0.169

- Table 5 (Creditor-Borrower Determinants with Matching Samples):
  - Panel A SRISK_GDP coefficients: -0.103*** to -0.108*** across columns
  - SRISK_GDP* WH: -0.957*** to -0.888*** across columns
  - SRISK_GDP* Asia: small positive coefficients in some columns (e.g., 0.0117) or negative in others
  - Panel B SRISK_GDP: positive small (e.g., 0.0147)
  - SRISK_GDP* WH: positive and significant (e.g., 0.483***)
  - SRISK_GDP* Asia: positive and significant (e.g., 0.289***)
  - Observations: Panel A 831; Panel B 831; R-squared examples: Panel A R-squared 0.264, Panel B R-squared 0.176

- Table 6 (Robustness: Supply Side Determinants and Regional Breakdowns):
  - SRISK_GDP: negative and significant in many subsamples (e.g., -0.0746***, -0.0842**)
  - SRISK_GDP* WH and SRISK_GDP* Asia show varying signs and significance across regional subsamples
  - system crisis, roa, npl, RWA over assets reported in various subsamples with large coefficients and significance in some panels (examples: system crisis -20.56*** in one cell)
  - Observations vary by subsample (e.g., 1,858; 307; 692; 368; 491)

- Table 7 (Robustness: Creditor-Borrower Determinants and Regional Breakdowns):
  - SRISK_GDP: negative and significant in many specifications (e.g., -0.0744***, -0.0797***, -0.107***)
  - SRISK_GDP* WH and SRISK_GDP* Asia: coefficients vary by regional breakdowns with instances of large negative and positive values (examples include -1.426***, 1.002***)
  - share_cross-border and other bilateral controls (comlang_off, comlegal, contig, col451, distlog, bitrade) show variable coefficients and significance across region-specific regressions
  - Observations and R-squared values reported per subsample (examples: Observations 1,858; 1,830; 307; 692; etc.)

- Table 8 (Creditor-Borrower Determinants with Creditor and Borrower FE):
  - share_cross-border coefficients: e.g., -0.744 to -0.827 (not uniformly significant)
  - col451 positive and significant across columns (e.g., 27.20***, 31.22***, 31.30***)
  - distlog and bitrade coefficients reported; distlog sometimes negative and significant (e.g., -11.47*), bitrade sometimes negative and significant (e.g., -5.812)
  - Both Borrower FE and Creditor FE included in many specifications
  - Observations examples: 1,873; 836; R-squared with both FE up to 0.169–0.203 in reported panels

### Figures, Sources, and Notes
- Figure 2 caption: "Bilateral evolution of banks’ claims during 2008Q2-09Q2 — Cross-Border Claims / Affiliates’ claims." Source: BIS, IFS, and authors’ estimations.
  - Notes: Each cell depicts bilateral percentage changes in cross-border claims (lender in columns, borrowers in rows). First column of each panel displays banks’ total lending to each borrower country. Columns sorted by overall degree of deleveraging of creditor country; rows sorted by overall degree of deleveraging of borrowing country.
- Figure 3 caption: "Evolution of Cross-Border and Affiliates' Claims — Period 2008Q2-09Q2." Source: BIS, IFS, and authors’ estimations.
- Figure 4 caption: "Supply Side Variables." Sources: NYU Stern, IFS, Central Banks, and authors’ estimations.
  - Chart labels and data points shown in source (example axis labels preserved): "SRISK as of Dec 2007  (as % of GDP)"; "Bank Fundamentals as of Dec 2007"; "roa (in %)"; "npl (in %)"; "rwa_assets (ratio, rhs)".
  - Example country ordering visible in figures: LU, PT, FI, AU, GR, CA, IT, AT, SP, US, JP, DN, IR, DE, SE, BE, ND, UK, FR, CH (as shown on axes in figure).

*Source: _wp14180 - References (PDF content provided).*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2014/_wp14180.pdf_