## _wp1309

## Source details

**Canonical URL:** [_wp1309](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2013/_wp1309.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2013/_wp1309.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2013/_wp1309.pdf.json)

---

### Introduction and objectives
- Objective: improve measurement of (i) foreign credit rollover risks (upstream exposure) facing borrowers from creditor foreign banking systems and (ii) creditor banks’ foreign credit exposure (downstream exposure) to crises in borrowing countries.
- Time coverage for evolution and drivers analysis: 2006-2012 (downstream/upstream analysis sample period: 2006Q2-2012Q1).
- Data and method highlights:
  - Uses BIS Consolidated Banking Statistics (BIS CBS) at ultimate risk basis combined with bank-level data (more than 700 identified banking subsidiaries/branches).
  - Includes on-balance sheet claims and gross off-balance sheet accounts (derivatives, credit commitments and guarantees) to proxy overall downstream exposure.
  - Novel procedure applied to account for coverage break-in-series and exchange rate variations; time series analysis focuses on on-balance sheet items due to data limitations in these adjustments.

### Measurement approach: downstream vs. foreign claims; upstream vs. foreign claims
- Downstream exposure (creditor banking system perspective):
  - Recognizes legal limits on parent-bank exposure to subsidiaries: legal exposure limited to capital incorporated in subsidiary plus non-capital debt owed by subsidiary to parent.
  - Downstream exposure measure components: A_ij (cross-border claims), B_ij (assetstotal_subs_ij + assetstotal_branch_ij - deposits_subs_ij), C_ij (claims local minus affiliates' assets), and D_ij (derivatives_ij + guarantees_credit_ij + commitment_credit_ij).
- Upstream exposure (borrower country perspective):
  - On-balance sheet upstream exposure = direct cross-border claims + affiliates’ claims not financed by local consumer deposits; conceptual formula:
    - Upstream_j = (sum over i of [cross-border claims_ij + claimsLocal_ij * (1 - min(1, loan/deposit ratio_ij))]) / Z_j
  - Deposit-to-loan ratio interpretation: higher deposit-to-loan ratio => lower share of affiliates’ claims financed by parent-bank resources/wholesale funding.

### Key empirical findings and exact figures
- Sample scope:
  - Combined BIS CBS and more than 700 identified banking subsidiaries/branches; upstream analysis covering 112 borrower countries; downstream analysis covering 26 of the 30 BIS CBS reporting countries (sample periods 2006Q2-2012Q1).
- Upstream exposure relative to BIS CBS foreign claims (as of March 2012):
  - Latin-America: on-balance sheet upstream exposure is, on average, only 45 percent of the foreign claims measure.
  - Emerging Asia: on-balance sheet upstream exposure is, on average, 55 percent of foreign bank claims.
  - Emerging Europe: on-balance sheet upstream exposure is, on average, 50 percent of foreign bank claims.
  - Advanced countries: on-balance sheet upstream exposure is, on average, 65 percent of foreign claims.
- Downstream exposure relative to BIS CBS foreign claims:
  - On average, downstream exposures are about 89 percent of foreign claims.
  - Heterogeneity:
    - Canadian and Dutch downstream exposures to developed country borrowers in terms of banks’ Tier I are about 180 and 120 percentage points lower than their foreign claims’ exposures in terms of banks’ Tier I, respectively.
    - Differences are small for Swiss and Luxemburg banks.
  - With off-balance sheet items included:
    - Overall levels of total downstream exposures (with off-balance sheet included) are an average of 925 percent of banks’ Tier I capital or 115 percent of GDP.
    - In Italy, Belgium, France, UK, Sweden, Germany, and Spain, off-balance sheet exposures represent about 40 to 30 percent of total downstream exposures.
    - For US banks, gross off-balance sheet exposures are larger than on-balance sheet exposures.
- Role of off-balance sheet credit commitments:
  - Credit commitments can represent a substantial share of upstream exposure; in some low income developing countries BIS reporting banks’ credit commitments represent more than ½ of on-balance sheet upstream exposures.
  - In Cape Verde and Seychelles, credit commitments represent about 24 percent and 18 percent of total credit, respectively.
- Country-level upstream extremes (as of March 2012, upstream as percentage of total credit):
  - Europe: Luxembourg (90 percent), Croatia (63 percent), Hungary (58 percent), Romania (55 percent), Serbia (54 percent), Finland (54 percent), Slovakia (50 percent).
  - Africa: Seychelles (69 percent), Cape Verde (56 percent).
  - Asia: Samoa (89 percent), New Zealand (57 percent).
  - Latin America: Panama (38 percent), Chile (36 percent).
- Noted extreme BIS foreign claims (percent of GDP) excluded from some plots:
  - Luxembourg (970 percent of GDP), Bahamas (324 percent), Ireland (210), Hong Kong (210 percent), New Zealand (170 percent), Cyprus (150 percent).

### Evolution, data adjustments, and heterogeneity across countries
- Coverage break-in-series and exchange rate issues:
  - BIS reports 84 series breaks during 2006-12 in BIS consolidated banking statistics at ultimate borrower risk basis; about 61 breaks due to mergers and acquisitions and about 23 due to expansion of banking sector coverage.
  - Coverage break-in-series example: USD 1.3 trillion in the first quarter of 2009 due to inclusion of former US investment banks’ claims (Box 2 cites the US 2009Q1 break as USD 1,334 billion; Annex B cites USD 1,333.7 billion).
  - Magnitude of coverage adjustments can be almost up to 15 percent of total foreign claims in 2006; exchange rate adjustments can be about five percent of total foreign claims in some quarters and often larger than coverage break adjustments.
- Adjustment methodology (bilateral creditor-borrower level):
  - Three exchange-rate corrections: (1) domestic-currency local affiliates’ claims corrected using bilateral USD–domestic exchange rate (proxied by share of BIS CBS immediate borrower basis); (2) identification of foreign-currency denominated local affiliates’ claims (assumed Euros in Europe, US dollars elsewhere); (3) bilateral cross-border claims adjusted using BIS locational banking statistics currency breakdown proxy.
- Resulting evolution after adjustments:
  - Adjusted downstream and upstream exposures show substantial heterogeneity across national banking systems and borrower countries.
  - Some countries experienced drastic reductions in exposures; others (examples) have current upstream exposure levels above pre-crisis levels: India, Indonesia, Turkey; downstream exposure increases in Spain and Sweden.
  - Downstream time-series: at the lowest point, US decrease in downstream exposures was about 15 percent and Hong Kong about 25 percent with respect to the December 2007 pre-crisis level.
  - Within Europe, Spanish, Swedish, Finnish, Turkish and Greek banking sectors saw increases in on-balance sheet downstream exposures from pre-crisis peaks as of March 2012; Belgian and Dutch banks reduced downstream exposure levels by about 80 and 70 percent, respectively, as of March 2012 relative to December 2007.

### Econometric analysis — Drivers of exposure changes (2006-2012)
- Downstream (creditor perspective) — main findings:
  - Dependent variable: quarterly growth rate in the adjusted on-balance sheet downstream exposure ∆D_it.
  - Key explanatory variables: ∆GDP_ijt, Cred_Systemic_Crisis, Cred_DLR, Cred_CB_Share, VIX, TED spread.
  - Individual-variable results: higher global risk aversion (VIX) and presence of systemic bank crisis in creditor banking systems linked with reductions in downstream exposures.
  - Joint specifications:
    - Presence of a systemic banking crisis in the creditor banking system is associated with about a 3 percent decline in downstream exposures in a given quarter.
    - Interaction at sample peak global financial variables: systemic banking crisis associated with a decline in downstream exposures of about 8 percent (with risk aversion) or about 11 percent (with TED spreads).
    - Footnote: "A one standard deviation increase in global financial variables, together with the presence of a systemic banking crisis, would be associated with a decline in downstream exposures of about 4½ percent."
  - Table 1 metadata (reported):
    - Observations 622
    - Number of Creditor Banks 26
    - R2 values reported across specifications: 0.082, 0.092, 0.081, 0.074, 0.084, 0.074, 0.102, 0.164, 0.111, 0.108.
- Upstream (borrower perspective) — main findings:
  - Dependent variable: quarterly growth rate in the adjusted on-balance sheet upstream exposure ∆U_jt.
  - Key explanatory variables: ∆GDP_jt, Creditor_Crisis_ijt (weighted), Borrower_CB_share, Borrower_DLR, TED spread, VIX.
  - Core empirical results (selected exact coefficients from Table 2 and Annex D):
    - Cred_Syst_Crisis coefficients (selected): -9.58***; -12.54***; -5.46*; -6.09***; -18.74*** (standard errors: (1.172), (1.285), (3.159), (1.713), (3.306)).
    - TED * Cred_Syst_Crisis: -0.111***; -0.106*** (standard errors: (0.025), (0.026)).
    - Cred_Syst_Crisis * Borrower_CB_Share: -0.153*** (standard error: (0.035)).
    - Borrower_CB_Share coefficients (selected): -0.250***; -0.311***; -0.310***; -0.289***; -0.268***; -0.290*** (standard errors: (0.071), (0.066), (0.063), (0.066), (0.096), (0.065)).
    - TED Spread (selected coefficients): -0.0126**; -0.0315***; 0.3222; -0.0655***; -0.0030 (standard errors: (0.0055), (0.0060), (0.7527), (0.0168), (0.0091)).
    - TED * Borrower_CB_Share: -0.00029* (standard error: (0.00016)).
    - TED * Borrower_DLR: 0.0183*; 0.0368** (standard errors: (0.0097), (0.0174)).
    - ∆GDP_jt coefficients (selected): 0.470**; 0.284*; 0.149; 0.252*; 0.284*; 0.244* (standard errors: (0.188), (0.149), (0.139), (0.138), (0.149), (0.140)).
  - Interpretation highlights:
    - Borrowing countries whose creditor banking systems experienced systemic crisis saw large negative changes in upstream exposures (up to -12 percent if all creditor banking systems were through systemic banking crisis).
    - The larger the share of direct cross-border borrowing (Borrower_CB_Share), the larger the decline in upstream exposures during crises; interaction terms show higher deterioration during peaks in TED/VIX.
    - Global financial conditions: baseline specification indicates a one standard deviation increase in TED spreads reduced upstream exposures by 1¾ percentage points.
  - Table 2 / Annex D metadata:
    - Observations 2,458 (across specifications)
    - Number of borrower countries 112
    - R2 values across specifications (selected): 0.057; 0.078; 0.065; 0.055; 0.055; 0.107; 0.154; 0.116; 0.112; 0.118.
    - Robust standard errors clustered at borrower country level; significance denoted by ***, **, * at 1%, 5% and 10% levels.

### Expected loss frameworks
- Expected downstream losses:
  - D_i = (1 / Z_i) * sum_{j=1 to N} [ Pr_crisis_j * (A_ij + B_ij + C_ij + D_ij) * LGD_j ]
  - Z_i scaling factor: Tier I Capital or total banking sector assets in country i.
  - LGD application can confine to non-liquid affiliates’ assets using bank-level liquid asset data; formal decomposition described in Annex B.
- Expected upstream losses:
  - U_j = (1 / Z_j) * sum_{i=1 to N} [ Pr_crisis_i * (1 - min(loan_deposit_ratio_ij, 1)) * (claimsCrossBorder_ij + claimsLocal_ij) ]
  - Z_j scaling factor: GDP or total banking sector credit in country j.

### Data, limitations and methodological notes
- Data sources and coverage:
  - Core data: BIS Consolidated Banking Statistics (ultimate risk basis), BIS locational banking statistics for currency breakdown proxies, IFS, ECB, Bankscope, Central Banks’ data, and bank-level affiliate data (annual, interpolated to quarterly).
  - Time coverage for adjusted series: 2006Q2-2012Q1.
- Main limitations (selected exact points):
  - BIS CBS time-coverage: available since 2005 at ultimate risk borrower basis.
  - Country-reporting level can mask banking group heterogeneity; downstream/upstream exposures can be calculated only at the level of banking systems rather than individual banks.
  - BIS CBS overestimates credit exposures/rollover risks because affiliates’ customer deposits should not be counted as rollover risk to borrowing country; proxy corrections use bank-level subtraction of deposits from assets.
  - Off-balance sheet figures are reported at face value (maximum possible exposures) and are not netted; protection (e.g., CDS bought) not taken into account.
  - Exchange rate conversion: BIS CBS reported in US dollars at end-of-quarter exchange rates can induce large movements unrelated to position changes.
  - Coverage of creditor systems limited to 30 BIS CBS reporting countries (coverage still very high given size of reporting systems).
  - Mapping international banks’ group structures to BIS reporting is challenging; some affiliates/branches remain unidentified.
- Practical adjustments applied to time series:
  - Coverage breaks adjusted using BIS-provided “series without the break” and assumptions that pre-break adjustments evolve at same pace as unadjusted series.
  - Exchange-rate adjustments implemented via three bilateral corrections (domestic-currency local claims; foreign-currency local claims assumed Euros in Europe and USD elsewhere; cross-border claims currency breakdown from LBS).
  - Time-series analysis focused on on-balance sheet items because currency breakdowns for off-balance sheet were not available.

### Conclusions and policy recommendations (preserved wording and emphasis)
- Measurement improvements:
  - Combining BIS CBS aggregate data with bank-level data improves measurement of creditor banking systems’ foreign credit exposures (downstream exposure) and borrower countries’ reliance on foreign bank credit (upstream exposure), especially for emerging markets.
  - Taking into account international banks’ organizational and/or funding structure is essential to correct BIS overestimation of exposure levels and to uncover heterogeneity across countries.
- Key empirical conclusions:
  - Several creditor banking systems’ downstream exposures have not recovered to pre-crisis exposure levels; reductions were driven mostly by presence of systemic banking crisis and intensified when global risk aversion and bank funding costs rose.
  - During 2008-12 creditor banking systems were able in several cases to reduce foreign downstream exposure by selling subsidiaries; nature of claims (direct cross-border vs. affiliates’ claims) mattered little for downstream reductions.
  - In almost all advanced countries and many emerging markets, upstream exposures remain below pre-crisis levels due to international deleveraging; some large EMs (India, Indonesia, Turkey, Brazil) saw upstream exposures increase above pre-crisis levels.
- Policy recommendations for borrower-country policymakers:
  - Given limited control over global financial conditions, policymakers should avoid (to the extent possible) relying on unstable creditor banking systems and excessive reliance on direct cross-border borrowing.
  - Improve data, measurement, and analysis of cross-border bank interlinkages to inform policymakers/analysts in both core and peripheral countries of the global financial architecture.
- Methodological caveat reiterated:
  - Time series analysis of BIS data must account for coverage breaks and sharp exchange rate variations during the crisis to effectively analyze evolution of downstream/upstream exposures.

*Source: IMF working paper chapter and annexes extracted from _wp1309.*

### 1. Comparing On-Balance Sheet Downstream Exposures Measures ........................................9

### 1. Comparing On-Balance Sheet Downstream Exposures Measures ........................................9

### Introduction and objectives
- Objective: improve measurement of (i) foreign credit rollover risks (upstream exposure) facing borrowers from creditor foreign banking systems and (ii) creditor banks’ foreign credit exposure (downstream exposure) to crises in borrowing countries.
- Time coverage for evolution and drivers analysis: 2006-2012.
- Data and method highlights:
  - Uses BIS Consolidated Banking Statistics (BIS CBS) at ultimate risk basis combined with bank-level data (more than 700 identified banking subsidiaries/branches).
  - Includes on-balance sheet claims and gross off-balance sheet accounts (derivatives, credit commitments and guarantees) to proxy overall downstream exposure.
  - Novel procedure applied to account for coverage break-in-series and exchange rate variations; time series analysis focuses on on-balance sheet items due to data limitations in these adjustments.

### Measurement approach: downstream vs. foreign claims; upstream vs. foreign claims
- Downstream exposure concept:
  - Recognizes legal limits on parent-bank exposure to subsidiaries: legal exposure to a subsidiary is limited to the capital incorporated in that subsidiary plus non-capital debt owed by the subsidiary to the parent.
  - Downstream exposure measure includes: direct cross-border lending, lending by branches, and a proxy of foreign subsidiaries’ non-parent funded claims; also includes gross off-balance sheet accounts to proxy overall exposure.
- Upstream exposure concept:
  - Measures foreign rollover risks faced by borrowing countries from direct cross-border borrowing and the proportion of lending by foreign affiliates (branches and subsidiaries) not funded by local customer deposits.
  - Uses BIS CBS at ultimate risk basis plus bank-level data to estimate the share of affiliates’ lending not funded by local deposits and to include credit commitments.

### Key empirical findings and exact figures
- Sample scope: combines BIS CBS and more than 700 identified banking subsidiaries/branches operating in advanced and emerging markets.
- Upstream exposure relative to BIS CBS foreign claims (as of March 2012):
  - Latin-America: on-balance sheet upstream exposure is, on average, only 45 percent of the foreign claims measure.
  - Emerging Asia: on-balance sheet upstream exposure is, on average, 55 percent of foreign bank claims.
  - Emerging Europe: on-balance sheet upstream exposure is, on average, 50 percent of foreign bank claims.
  - Advanced countries: on-balance sheet upstream exposure is, on average, 65 percent of foreign claims.
- Role of off-balance sheet credit commitments:
  - Credit commitments can represent a substantial share of upstream exposure; in some low income developing countries BIS reporting banks’ credit commitments represent more than ½ of on-balance sheet upstream exposures.
- Downstream exposure relative to BIS CBS foreign claims:
  - On average, downstream exposures are about 89 percent of foreign claims.
  - Heterogeneity across creditor banking systems:
    - For exposure to developed country borrowers, Canadian and Dutch downstream exposures are about 180 and 120 percentage points of banks’ Tier I lower than what the use of only CBS foreign claims would indicate.
    - Differences are small for Swiss and Luxemburg banks.
  - Inclusion of off-balance sheet items materially increases downstream exposure for several countries; notably, for US banks gross off-balance sheet exposures are larger than on-balance sheet exposures.

### Evolution, data adjustments, and heterogeneity across countries
- Need for adjustments:
  - Coverage break-in-series and sharp exchange rate variations during the crisis materially affect the evolution of BIS CBS claims reported in US dollars.
  - Coverage break-in-series are numerous and large: about 21 break-in-series during 2006-12; an example is USD 1.3 trillion in the first quarter of 2009 due to inclusion of former US investment banks’ claims.
- Resulting evolution after adjustments:
  - Adjusted downstream and upstream exposures show substantial heterogeneity across national banking systems and borrower countries.
  - Some countries experienced drastic reductions in exposures, while others have current exposure levels above pre-crisis levels (examples: upstream exposure increases in India, Indonesia and Turkey; downstream exposure increases in Spain and Sweden).

### Drivers of exposure changes (2006-2012) — econometric analysis summary
- Main driving factors identified:
  - For downstream exposures (creditors’ perspective):
    - Presence of systemic banking crises in creditor countries played an important role in large reductions in downstream exposures.
    - Reductions intensified during the peak of the crisis when global risk aversion and funding pressures increased.
    - The nature of claims (direct cross-border vs. affiliates’ claims) mattered little from the creditor banks’ perspective.
  - For upstream exposures (borrowers’ perspective):
    - A larger set of factors explains variation in upstream exposures:
      - (i) From whom they borrowed: systemic banking crisis in creditor banking systems translated into declines in upstream exposures.
      - (ii) How they borrowed: rollover of direct cross-border lending was much more difficult than rollover of affiliates’ lending.
      - (iii) Global financial conditions: higher global risk aversion and/or bank funding costs increased rollover problems.
    - Domestic demand conditions show some evidence of influence but are not the primary drivers.
- Policy implication distilled from findings:
  - Policymakers should avoid reliance on unstable creditor banking systems and excessive concentration of direct cross-border borrowing; the origin and composition of cross-border borrowing affected rollover outcomes during the crisis.

### Data and methodological notes
- Data sources and adjustments:
  - Core data: BIS Consolidated Banking Statistics (BIS CBS) at ultimate risk basis, complemented with bank-level data for affiliates.
  - Adjustments performed: coverage break-in-series corrections and exchange rate variation adjustments (detailed procedures described elsewhere in the chapter).
  - Time series analysis limited to on-balance sheet items due to data limitations for exchange rate and break-in-series adjustments.
- Comparative advantages of approach:
  - Consolidated BIS data at ultimate risk basis allows inclusion of credit commitments and identification of ultimate creditor holdings, improving measurement of foreign rollover risks relative to locational BIS data which overestimates certain claims and cannot net intergroup lending.

*Source: IMF working paper chapter "1. Comparing On-Balance Sheet Downstream Exposures Measures" (excerpt).*

### BOX 1 – BIS Consolidated Banking Statistics: Characteristics and Limitations

### BOX 1 – BIS Consolidated Banking Statistics: Characteristics and Limitations

### Characteristics of BIS CBS
- The BIS consolidated banking statistics (BIS CBS) provide quarterly data reported by 30 national regulators on aggregate consolidated foreign claims (assets) of internationally active banks headquartered in their jurisdiction (country).
- Banks’ positions vis-à-vis residents of their home country are not included.
- Consolidation—netting out intergroup positions and consolidating positions across offices worldwide—produces an internationally comparable measure of banking systems’ risk exposures.
- BIS CBS on an ultimate risk basis:
  - Provides the residency of ultimate obligor.
  - Provides a clear breakdown into cross-border and affiliates’ extended claims.
  - Provides banks’ maximum off-balance sheet exposures.
- Off-balance sheet items include derivative contracts and contingent exposures:
  - Derivatives exposures include the positive market value of outstanding contracts covering foreign exchange, interest rate, equity, commodity and credit risks.
  - Contracts with negative market value are classified as liabilities, and are not reported and/or netted out.
  - Guarantees and credit commitments are reported at face value, i.e., at maximum possible exposures.
- BIS CBS collects internationally comparable measures of national banking systems’ exposure to country risk and reports gross off-balance sheet foreign exposures—derivatives, guarantees and credit commitments—on an ultimate borrower risk consolidated basis.

### Main limitations of BIS CBS (enumerated)
- Time-coverage: Data are only available since 2005 at ultimate risk borrower basis.
- Country Level vs. Bank level Analysis: The country reporting level could mask banking group heterogeneity and bias the systemic risk analysis.
- Overestimation of Credit Exposures and Rollover Risks:
  - A banking group exposure to its subsidiary in a borrowing country is limited to the capital incorporated in the subsidiary plus other non-equity lending from the parent bank to the subsidiary.
  - Funding risks faced by a borrowing country should not include the customer deposit liabilities on affiliates’ balance sheets.
  - These biases can be mitigated with bank level data.
- Shortcomings of Consolidated Balance sheets Analysis:
  - Consolidated balance sheet analysis implicitly assumes a frictionless flow of funds within a banking group.
  - Locational data does not necessarily resolve this problem due to duplication of exposure problems.
  - Access to granular bank-level data is essential to estimating potential ring fencing risks.
- Time Series Analysis: Numerous and important break-in-series, and exchange rate movements, affect time series analysis; claims reported in other currencies are converted into US dollars at end-of-quarter exchange rates, producing potentially significant changes even if actual positions are unchanged.
- Domestic Exposure Overestimation in Financial Centers:
  - Not all “foreign” banks are excluded from national supervisor submissions to BIS data.
  - In Financial Centers, some foreign-owned banks incorporated in that country may be reported as domestic (example discussed: Ireland).
- Breakdowns only available at the aggregate: maturity, sectoral, and currency breakdowns are only available for total foreign claims.
- Lack of liability data and non-bank exposures: BIS CBS does not include information on the structure of liabilities, and balance-sheet report of non-banks.
- Gross off-balance sheet figures: BIS CBS data are a proxy of the maximum potential off-balance sheet risks since they are not netted out (e.g., protection bought through CDS contracts are not taken into account).

### On-balance sheet downstream exposure: methodology and key findings
- BIS CBS foreign claims (direct cross border plus affiliates claims) are designed to capture on-balance sheet foreign exposures but do not account for branch/subsidiary reporting structures, making them an upper-bound measure of a banking system’s foreign credit risk.
- Combining BIS CBS with bank-level data allows derivation of a lower-bound downstream exposure that accounts for variations in banks’ organizational structure.
- Proxy for a bank’s downstream exposure to its subsidiaries: subtract total customer deposits from total assets (to capture non-deposit fraction of subsidiaries’ exposures).
- Formal decomposition used (terms preserved as in source):
  - Aij = claimsborderCrossij, capturing the direct cross-border exposure from creditor country banking system i on debtor country j;
  - B__ij = branch_subs_ij = assetstotal__subs_ij + branch_ij - depositsassetstotal__subs_ij (capturing full exposure to branches and the non-deposit fraction of subsidiaries’ exposures);
  - C__ij = Σ(branch_subs_ij assetstotal - claimslocal_ij), capturing the non-identified exposures by bank level data with respect to BIS reported affiliates claims.
- Empirical comparisons (aggregated across advanced or emerging/developing borrower countries for each banking system) as of March 2012:
  - On average, the lower-bound downstream measure is about 50 and 25 percentage points of banks’ Tier I capital below the upper-bound foreign claim measure across developed and emerging borrower countries, respectively.
  - Specific examples:
    - Canadian and Dutch downstream exposures to developed countries’ borrowers in terms of banks’ Tier I are about 180 and 120 percentage points lower than their foreign claims’ exposures in terms of banks’ Tier I, respectively.
    - For Belgian, Austrian, and Spanish banks, downstream exposures to emerging countries are about 60 to 90 percentage points of banks’ Tier I capital below total foreign claim exposures.
  - For some creditor banking systems (e.g., Luxembourg), lower and upper-bound measures differ little despite very large absolute levels of on-balance sheet downstream exposures.

### On- and off-balance sheet downstream exposure: methodology and key findings
- Off-balance sheet foreign exposures D_ij are included with on-balance sheet exposures to depict a complete picture of vulnerabilities; D_ij = derivative_ij + guarantees_ij + credit_commitment_ij (as defined in source).
- Downstream exposure index (Downstream_i) formally aggregates A_ij, B_ij, C_ij, D_ij across borrower countries and scales by i_Z (domestic banks’ Tier I capital or GDP) as specified in the source.
- With off-balance sheet items included:
  - Banking systems’ downstream exposures are all above foreign claims’ exposures.
  - This is especially the case in the US where off-balance sheet exposures are larger than on-balance sheet exposures.
  - In Italy, Belgium, France, UK, Sweden, Germany, and Spain, off-balance sheet exposures represent about 40 to 30 percent of total downstream exposures.
  - Overall levels of total downstream exposures (with off-balance sheet included) are an average of 925 percent of banks’ Tier I capital or 115 percent of GDP.
- Further breakdown observations:
  - Swedish and Spanish banks are the two banking systems most exposed through their foreign affiliates’ lending, even after discounting their locally funded claims.
  - Dutch, Canadian, and Italian banks have important affiliate networks responsible for more than 50 percent of exposures to foreign affiliates.
  - Most BIS reporting banking systems’ exposures remain almost exclusively within advanced countries—only Austria, Greece, and Spain have a relatively more balanced regional exposure (Austria and Greece exposed to CEE; Spain mostly exposed to Latin America).
  - Almost all banking systems are more exposed to the private non-bank sector on their on-balance sheet exposures rather than to the public and/or banking sectors.

### Expected downstream exposure losses
- Expected downstream losses extend the downstream exposure measure using:
  - downstream exposure levels,
  - probability of crisis in borrowing countries (available from other modules in the VEA/VEE),
  - loss-given default (LGD) parameters for each borrower country.
- Bank-level data providing affiliates’ liquid assets allow confinement of LGD application to non-liquid affiliates’ assets only.
- Equations to estimate expected downstream losses are described in Annex B (as referenced in source).

### Limitations in the downstream exposure analysis (enumerated)
- The on-balance sheet downstream exposure does not correct all overestimation of BIS CBS foreign claims because it also includes non-parent bank wholesale funding.
- Downstream exposures can be calculated only at the level of banking systems rather than for individual banks due to BIS data characteristics.
- Off-balance sheet data refer to maximum possible exposures and are available only for banking systems reporting BIS CBS data at ultimate risk basis.
- Bank-level data on subsidiaries are (for the most part) available only annually and thus need to be interpolated to match quarterly BIS CBS data.
- Mapping international banks’ group structure and consistently matching it to the BIS CBS is difficult using commercial data (e.g., Bankscope, Bankers’ Almanac) and publicly available regulator data since not all affiliates can be identified, especially not all branches.

*Source: BOX 1 – BIS Consolidated Banking Statistics: Characteristics and Limitations (extracted content).*

### Annex A); and (vi) As well as with the use of BIS CBS, the interpretation of the downstream

### _wp1309 - Annex A); and (vi) As well as with the use of BIS CBS, the interpretation of the downstream

### III. MEASURING BORROWERS’ FOREIGN ROLLOVER RISKS: UPWARD EXPOSURE
- Perspective: borrowers (all sectors) of a country and the cross-border rollover risk from problems in creditor banking systems that could interrupt inflow of foreign credit.
- Purpose: provides a forward looking measure of rollover risks borrowers are exposed to with respect to crisis in a particular banking system or groups of banking systems.
- Data combination: upstream analysis combines BIS CBS and bank-level balance sheet data.
- BIS CBS foreign claims definition: worldwide consolidated direct cross-border claims on country j plus positions booked by banking system i’s affiliates (subsidiaries and branches) on country j.
- Caveat: BIS measure includes all affiliates’ assets and does not by itself capture borrower country rollover risks; adjustment using bank-level data is required to reflect the type of funding that financed affiliates’ claims.
- Upstream exposure components: on-balance sheet and off-balance sheet exposure measures.

### A. On-Balance Sheet Upstream Exposure Level
- Definition: on-balance sheet upstream exposure = direct cross-border claims + affiliates’ claims not financed by local consumer deposits.
- Formula (conceptual): Upstream_j = (sum over i of [cross-border claims_ij + claimsLocal_ij * (1 - min(1, loan/deposit ratio_ij))]) / Z_j
  - where:
    - cross-border claims_ij = direct cross-border claims from banking system i on borrower country j;
    - claimsLocal_ij = affiliates (subsidiaries and branches) claims of parent banks from country i on borrower country j;
    - )1,_(1 ij ratioloandepositMin is proxy of proportion of loans not financed by local consumer deposits;
    - N = total number of BIS CBS reporting countries;
    - Z_j = scaling factor (GDP or bank credit to the local non-bank private and public sectors in country j).
- Deposit-to-loan ratio calculation: uses subsidiaries and branches deposits and loans data summed over affiliates where available; based on developed and emerging markets bank-level data, or national country data where affiliates’ bank level data not available (e.g., borrower low income countries).
- Interpretation: higher deposit-to-loan ratio => lower share of local claims financed by parent bank resources and/or wholesale financing.
- Alternative proxy: using affiliates’ total assets minus deposits produces similar results but much lower country coverage.
- Coverage expansion: proposed definition enlarges sample from 26 to 53 countries (full sample of 193 countries in map Figure 4 referenced).
- Correlations:
  - Correlation between proposed on-balance sheet upstream exposures and alternative definition using affiliates’ total assets minus deposits is about 85 percent for all developed and emerging countries with bank-level data.
  - Correlation above 90 percent when sample restricted to only emerging countries.
- Country examples and comparisons:
  - On-balance sheet upstream exposure of borrowers in Latin-America (red circles) to creditor banks is, on average, only 45 percent of the foreign claims measure.
  - Upstream exposures for emerging Asia and Europe are on average only 55 and 50 percent of foreign bank claims, respectively.
  - Share for borrowers in advanced countries is around 65 percent.
- Geographic levels (as percentage of total—local and direct cross-border—non-bank credit):
  - Europe: Luxembourg (80 percent), Croatia (54 percent), Hungary (49 percent), Serbia (46 percent), Romania (46 percent).
  - Outside Europe: Bahamas (66 percent), New Zealand (45 percent).
  - Selected emerging markets: Chile (28 percent), Uruguay (25 percent), Turkey (20 percent).
- Right-hand panel of Figure 3: comparisons also expressed as percent of borrowing countries GDP; highlights relatively high exposure of developed European countries in GDP terms; several Latin America and Eastern Europe exposure levels lower in GDP terms due to lower financial deepening.
- Notes on extreme BIS foreign claims values (Figure 3 footnote):
  - Only foreign claims up to 120 percent of GDP displayed.
  - Luxembourg (970 percent of GDP), Bahamas (324 percent), Ireland (210), Honk Kong (210 percent), New Zealand (170 percent), and Cyprus (150 percent) are not plotted.
  - Four countries with BIS foreign claims as percent of non-bank credit larger than 100 percent (Czech Republic, New Zealand, Slovakia, Luxembourg) have specific drivers: inclusion of affiliates’ claims on central bank and other banks, large direct credit to domestic banks, or combinations thereof.

### B. On- and Off-Balance Sheet Upstream Exposure
- Extension: add credit commitments secured from BIS reporting banks to capture full upstream exposure (off-balance sheet items).
- Upstream exposure presented in Figure 4 represents total foreign rollover risks faced by a borrowing country with respect to problems in creditor countries’ banking systems.
- As of March 2012, most exposed countries by upstream exposure as percentage of total credit:
  - Europe: Luxembourg (90 percent of total credit), Croatia (63 percent), Hungary (58 percent), Romania (55 percent), Serbia (54 percent), Finland (54 percent), Slovakia (50 percent).
  - Africa: Seychelles (69 percent), Cape Verde (56 percent).
  - Asia: Samoa (89 percent), New Zealand (57 percent).
  - Latin America: Panama (38 percent), Chile (36 percent).
- Interpretation: upstream exposure values correspond to a worst case scenario — losing without possibility of domestically or externally replacing all BIS reporting banks’ financing.
- Role of off-balance sheet in some countries: credit commitments represent about 24 percent and 18 percent of total credit in Cape Verde and Seychelles, respectively.

- Expected upstream exposure credit losses:
  - Can be calculated as combination of upstream exposures and probability of crisis in creditor countries.
  - Often included in IMF Vulnerability exercises where crisis country probabilities are available from other modules (reference to IMF 2012).
  - Annex B described equations for estimating expected upstream losses (annex content not reproduced here).

### C. Limitations in the Upstream Exposure Analysis
- Data limitations from downstream analysis apply here as well.
- Additional limitations:
  - (i) Granular data on banks’ internal capital markets and wholesale sources of funds (e.g., interbank repo market borrowing, and other non-deposit funding) are generally not available.
  - (ii) Not all potential creditor banking systems are included—only claims from the 30 BIS CBS reporting countries are taken into account; coverage is nevertheless very high because BIS CBS covers the largest banking systems and consolidated data include lending by affiliates of BIS reporting countries.
    - Example: lending to a Lithuanian corporate by a subsidiary of a Spanish bank incorporated in Poland is included via consolidation.
  - (iii) Interpretation of upstream measures on a country level needs careful consideration due to country-specific factors (example: BIS reporting banks’ claims on Liberia largely driven by the fact the country has the second-largest maritime registry in the world).

### IV. DOWNWARD AND UPWARD EXPOSURES: A TIME SERIES PERSPECTIVE
- Objective: estimate evolution of on-balance sheet downstream and upstream exposures before and after the crisis.
- Challenges:
  - Need to adjust for exchange rate movements and selected BIS CBS break-in-series.
  - Adjustments to BIS CBS data for time series are not directly available from BIS CBS; can be estimated by combining Locational and Consolidated BIS data (see adjustments details in Box 2 and Annex C referenced).
  - High exchange rate volatility during the crisis caused BIS CBS claims (reported in US dollars) to display variation due to exchange rate movements rather than underlying positions.
  - Adjusted figures show exchange rate movements triggered negative and positive adjustments of up to 5 percent of foreign claims at different points in time.
  - Coverage break example: inclusion of claims of former investment banks (Goldman Sachs and Morgan Stanley) in US BIS CBS triggered a coverage break-in-series of about USD 1.3 trillion in the first quarter of 2009.
- Next steps indicated: presentation of evolution of upstream and downstream exposures, especially from the pre-crisis peak (figures and series adjustments referenced but not reproduced here).

*Source: Author’s estimations based on BIS, IFS, ECB, Bankscope and Central Banks’ data (content excerpt from provided PDF).*

### BOX 2 – Adjusting BIS CBS for Coverage Break-in-Series and Exchange Rate Movements

### BOX 2 – Adjusting BIS CBS for Coverage Break-in-Series and Exchange Rate Movements

### Coverage break-in-series: scope and implications
- BIS reports 84 series breaks during 2006-12 in BIS consolidated banking statistics at ultimate borrower risk basis.
- About 61 breaks are due to mergers and acquisitions among foreign banks.
- Other 23 coverage break-in-series are driven by an expansion of the banking sector coverage (e.g., increase in reporting population due to inclusion of former investment banks, merger of domestic banks that triggered a consolidation of foreign claims, etc.).
- Important example: the US 2009Q1 USD 1,334 billion break-in-series when former investment banks became banks (e.g. Goldman Sach’s foreign claims existed before 2009Q1).
- Magnitude of adjustments: almost up to 15 percent of total foreign claims in 2006.
- Consequence: Performing time series analysis without correcting the original BIS data can lead to misleading conclusions (example given: unadjusted BIS series would indicate increases in US cross-border and US local claims during the crisis, whereas adjusted series show decreases).
- BIS reports the value of the series without the break, which helps offset the break impact.

### Exchange rate adjustments: methodology and assumptions
- The impact of exchange rate movements was important during the crisis because of high exchange rate volatility.
- Three corrections performed at the bilateral creditor national banking system–borrower country level:
  - First: domestic-currency denominated local affiliates’ claims are corrected following the bilateral US dollar domestic currency exchange rate. Domestic-currency denominated local affiliates’ claims are proxied by using its share of total BIS CBS foreign claims at immediate borrower basis.
  - Second: this procedure identifies the amount of foreign-currency denominated local affiliates’ claims, which are assumed to be in Euros in Europe and US dollars in the rest of the countries.
  - Third: bilateral CBS cross-border claims positions are adjusted using, as proxy, the currency breakdown currency (among US. Dollar, Euro, British Pound, Japanese Yen, and Swiss Francs) available from the BIS locational banking statistics (LBS).

### Empirical context and data coverage for adjusted series
- Adjusted series used for downstream/upstream exposure analysis are corrected for both coverage break-in-series and exchange rate variations.
- Sample periods and coverage:
  - Downstream analysis: 2006Q2-2012Q1, covering 26 of the 30 countries reporting BIS CBS during the period.
  - Upstream analysis: 2006Q2-2012Q1, covering 112 borrower countries.
- Data sources referenced for broader analysis: BIS, IFS, ECB, Bankscope and Central Banks’ data.

### Key empirical findings — Downstream exposures (creditor banking system perspective)
- Main empirical specification: quarterly growth rate in the adjusted stock of on-balance sheet downstream exposure of creditor banking system i in quarter t (∆D_it).
- Explanatory variables include: ∆GDP_ijt (creditor-weighted borrower GDP growth), Creditor_Bank measures (Cred_Systemic_Crisis, Cred_DLR, Cred_CB_Share), global measures (TED spread, VIX), creditor fixed effects, and quarter/time dummies.
- Individual-variable results (columns 1–6 in Table 1): most variables have expected sign and are statistically significant when considered individually:
  - Higher global risk aversion (VIX) and presence of systemic bank crisis in creditor banking systems are linked with reductions in downstream exposures.
  - An increase in borrower countries’ GDP growth (∆GDP_ijt) and in the deposit-to-loan ratio of creditor banking systems (Cred_DLR) show positive significant relationships with downstream exposures.
  - TED spread and Cred_CB_Share do not display statistically significant correlations when considered individually.
- Joint specifications (columns 7–10 in Table 1):
  - Presence of a systemic banking crisis in the creditor banking system is associated with about a 3 percent decline in downstream exposures in a given quarter (reported: "A systemic banking crisis would trigger about a 3 percent decline in downstream exposures in a given quarter.").
  - Interaction of global financial variables with creditor systemic crisis:
    - At sample peak global financial variables, the presence of a systemic banking crisis is associated with a decline in downstream exposures of about 8 percent (specification with risk aversion) or about 11 percent (specification with TED spreads).
  - Footnote: "A one standard deviation increase in global financial variables, together with the presence of a systemic banking crisis, would be associated with a decline in downstream exposures of about 4½ percent."
- Table 1 sample statistics and regression metadata (as reported):
  - Observations 622
  - Number of Creditor Banks 26
  - R2 values reported for columns: 0.082, 0.092, 0.081, 0.074, 0.084, 0.074, 0.102, 0.164, 0.111, 0.108 (presented in table)
  - Quarter dummies: Yes in most specifications; Time Fixed Effects: used in some specifications.
- Aggregate and country-level results highlighted:
  - Within Europe, only the Spanish, Swedish, Finnish, Turkish and Greek banking sectors have seen an increase of their on-balance sheet downstream exposures from the pre-crisis peaks as of March 2012.
  - Belgian and Dutch banks reduced downstream exposure levels by about 80 and 70 percent, respectively, as of March 2012 with relation to December 2007.
  - Outside Europe, the US and Hong Kong banking systems were particularly affected during 2008-09 and remained below the December 2007 pre-crisis level: at the lowest point, the US decrease in downstream exposures was about 15 percent and Hong Kong about 25 percent with respect to the December 2007 pre-crisis level.

### Key empirical findings — Upstream exposures (borrower country perspective)
- Baseline specification: quarterly growth rate in the adjusted stock of on-balance sheet upstream exposure of borrower country j in quarter t (∆U_jt).
- Explanatory variables include: ∆GDP_jt (borrower GDP growth), Creditor_Crisis_ijt (weighted proportion of creditor banking systems experiencing systemic crisis), Borrower characteristics (Borrower_CB_share, Borrower_DLR), global measures (TED spread, VIX), borrower fixed effects, and quarter/time dummies.
- Expected signs:
  - ∆GDP_jt: positive (higher borrower GDP growth associated with higher lending demand).
  - Creditor_Crisis (weighted): negative (creditor systems in crisis reduce foreign lending).
  - Borrower_CB_share: negative (higher share of direct cross-border in borrower foreign borrowing implies larger fall in upstream exposures during crisis).
  - Borrower_DLR: positive (more reliance on domestic deposits could insulate foreign affiliates).
  - Global variables (TED, VIX): negative (higher funding pressures/risk aversion reduce foreign borrowing supply).
- Empirical results summary:
  - A larger set of factors explains upstream exposures than downstream exposures (text summary).
  - Table 2 focuses on TED spreads and interaction terms; VIX results are presented in appendix tables (not reproduced here).

### Analytical and methodological considerations
- Distinction emphasized between downstream and upstream exposure measures: differences in treatment of branches vs affiliates and aggregation across creditors vs borrowers can produce different empirical implications (example: an acquisition by creditor A of a subsidiary from creditor B changes downstream exposures of A and B but not upstream exposure of borrower J).
- Fixed effects and time/quarter dummies are used to control for unobserved time-invariant factors and seasonal/annual reporting frequency issues.
- Data limitations noted: limited coverage for Norway, Singapore, Taiwan, and Mexico constrained some specifications; inclusion of additional bank characteristics (e.g., Tier I ratios) reduced sample size but suggested solvency may also matter (Tier I positive and statistically significant at 10 percent in reduced sample).

*Source: BOX 2 – Adjusting BIS CBS for Coverage Break-in-Series and Exchange Rate Movements (extracted from the provided content unit).*

### Annex D, and results are very similar. The reading of the results highlights that:

### Annex D, and results are very similar. The reading of the results highlights that:

### Impact of creditor banking system crises on upstream exposures
- Borrowing countries operating with creditor banking systems that were experiencing systemic banking crisis suffered a negative change in upstream exposures (up to -12 percent if all creditor banking systems were through systemic banking crisis).
- Table 2 coefficients and related results (selected):
  - Cred_Syst_Crisis: -9.58***; -12.54***; -5.46*; -6.09***; -18.74*** (standard errors in parentheses: (1.172), (1.285), (3.159), (1.713), (3.306)).
  - TED * Cred_Syst_Crisis: -0.111***; -0.106*** (standard errors: (0.025), (0.026)).
  - Cred_Syst_Crisis * Borrower_CB_Share: -0.153*** (standard error: (0.035)).
- Interpretations highlighted in the text:
  - The negative impact of systemic banking crisis in creditor banking systems was lower the higher the borrower deposit to loan ratio (interaction coefficient column 9).
  - Countries with domestic banking systems with lower exposure to non-deposit funding were able to insulate themselves better during the crisis.
  - The explanatory power of the proportion of systemic banking crisis in creditor banking systems is not only based on timing in the sample; it aligns with evidence that deterioration of creditor banking systems’ health—proxied by creditors’ banks’ CDS spreads—explains variation in emerging markets’ cross-border bank borrowing.

### How a country borrows (composition of exposures)
- The larger the share of cross-border on total borrower foreign claims, the further the decline in upstream exposures.
- Borrower_CB_Share coefficients (Table 2): -0.250***; -0.311***; -0.310***; -0.289***; -0.268***; -0.290*** (standard errors: (0.071), (0.066), (0.063), (0.066), (0.096), (0.065)).
- Interaction findings:
  - The interaction term with the share of cross-border and TED spread was statistically significant, indicating deterioration in upstream exposures during the peak of the crisis was higher in the presence of larger direct cross-border.
  - The interaction term with Borrower_CB_Share and Cred_Syst_Crisis (column 10) was statistically significant and negative, highlighting that systemic bank crisis in creditor countries increased the negative effect of large direct cross-border share in upstream exposures.
- Mechanism and divergence with downstream analysis:
  - Sale/acquisition of foreign affiliates can affect downstream and upstream differently: acquisition may change the name of creditor banking system without much change in affiliate lending from the borrowing country perspective, whereas reductions in creditor banking system exposures via cross-border channels did not necessarily imply substitution by other creditor banking systems.

### International financial conditions
- Global risk measures were key drivers during the period.
- Baseline specification (column 6 in Table 2): a one standard deviation increase in TED spreads reduced upstream exposures by 1¾ percentage points.
- TED Spread coefficients and interactions from Table 2 (selected):
  - TED Spread: -0.0126**; -0.0315***; 0.3222; -0.0655***; -0.0030 (standard errors: (0.0055), (0.0060), (0.7527), (0.0168), (0.0091)).
  - TED * Borrower_CB_Share: -0.00029* (standard error: (0.00016)).
  - TED * Borrower_DLR: 0.0183*; 0.0368** (standard errors: (0.0097), (0.0174)).
- Robustness:
  - Results are similar when using global risk aversion (VIX) instead of TED spreads (see Annex D reference).

### Domestic (borrower country) factors
- GDP growth:
  - ∆GDPijt coefficients in Table 2: 0.470**; 0.284*; 0.149; 0.252*; 0.284*; 0.244* (standard errors: (0.188), (0.149), (0.139), (0.138), (0.149), (0.140)).
  - The text: a one percent increase in GDP gowth increase upstream exposures by up to 1/3 of a percent (noting significance only at 10 percent in some specifications).
  - Note: the estimation procedure was not designed to break down supply and demand factors; a difference-in-difference procedure using bilateral creditor-borrower data would be necessary for that decomposition.

### Empirical specification and sample statistics (Table 2, selected)
- Dependent variable: Change in Adjusted Upstream Exposure (in percent).
- Sample period: 2006Q2-2012Q1.
- Estimation: Panel OLS with Fixed Effects; baseline panel fixed effect described in equation (4).
- Fixed effects and controls:
  - Quarterly Dummies: Yes in most columns.
  - Time Fixed Effects: included in column 7; otherwise No in reported columns.
  - Borrower Fixed Effect: Yes across reported specifications.
- Observations: 2,458 in most columns; one column shows 2,458 or 2,458 with an 8th column and one with 2,458 / 2,458 / 2,458 / 2,458 / 2,458 / 2,458 / 2,458 / 2,458 / 2,458 / 2,458 (table shows observations 2,458 repeated).
- Number of borrower countries: 112.
- R2 values (selected): 0.057; 0.078; 0.065; 0.055; 0.055; 0.107; 0.154; 0.116; 0.112; 0.118.
- Note: Robust standard errors are clustered at the borrower country level. Asterisks denote significance with ***, **, * indicating significance at 1%, 5% and 10% levels, respectively.

### Conclusions and policy recommendations
- Measurement improvements:
  - Combining BIS CBS aggregate data with bank-level data improves measurement of creditor banking systems’ foreign credit exposures (downstream exposure) and borrower countries’ reliance on foreign bank credit (upstream exposure), especially for emerging markets.
  - Taking into account international banks’ organizational and/or funding structure is essential to correct BIS overestimation of exposure levels and to uncover heterogeneity across countries.
- Key empirical conclusions:
  - Several creditor banking systems’ downstream exposures have not recovered to pre-crisis exposure levels; in some cases they experienced substantial reductions driven mostly by the presence of systemic banking crisis and intensified during the crisis peak when risk aversion increased and bank funding costs rose.
  - During 2008-12, creditor banking systems were able in several cases to reduce foreign downstream exposure by selling subsidiaries; the nature of claims (direct cross-border vs. affiliates’ claims) seems not to have mattered much for downstream reductions.
  - In almost all advanced countries and many emerging markets, upstream exposures remain below pre-crisis levels due to international deleveraging; some large EMs (India, Indonesia, Turkey, Brazil) saw upstream exposures increase to above pre-crisis levels.
- Policy recommendations for borrower-country policymakers:
  - Given limited control over global financial conditions, policymakers should avoid (to the extent possible) relying on unstable creditor banking systems and excessive reliance on direct cross-border borrowing.
  - Improve data, measurement, and analysis of cross-border bank interlinkages to inform policymakers/analysts in both core and peripheral countries of the global financial architecture.
- Additional methodological notes:
  - Time series analysis of BIS data must account for coverage breaks and sharp exchange rate variations during the crisis to effectively analyze evolution of downstream/upstream exposures.

*Source: Annex D and related text from the PDF content unit provided.*

### ANNEX B – CALCULATION OF DOWNSTREAM AND UPSTREAM EXPOSURES

### ANNEX B – CALCULATION OF DOWNSTREAM AND UPSTREAM EXPOSURES

### B.1 - Downstream Analysis
- Purpose: Quantifies a creditor country’s vulnerability to crises in countries that borrow from its banks by capturing exposures on:
  - direct cross-border lending;
  - off-balance sheet positions (derivatives, credit guarantees, and credit commitments);
  - affiliates’ claims on borrowing countries (subsidiaries and/or branches).
- Components and formula:
  - Downstream exposure for creditor country i vis-à-vis borrower country j:
    - ExposureDownstream_ij = A_ij + B_ij + C_ij + D_ij
  - Definitions:
    - A_ij = Cross-border claims_ij (direct cross-border exposure from creditor country i on debtor country j)
    - B_ij = assetstotal_subs_ij + assetstotal_branch_ij - deposits_subs_ij (captures exposure to subsidiaries and branches, accounting for legal differences between them)
    - C_ij = claimslocal_ij - (assetstotal_branch_ij + assetstotal_subs_ij) (represents non-identified exposure by bank-level data with respect to BIS reported affiliates claims)
    - D_ij = derivatives_ij + guarantees_credit_ij + commitment_credit_ij (captures (gross) off-balance sheet exposure from country i banks on country j based on BIS data)
- Practical measurement notes:
  - BIS data allow quantification of cross-border lending and off-balance sheet positions.
  - BIS affiliates’ claims often overstate parent banks’ legal exposure to subsidiaries; parent legal exposure limited to capital in the subsidiary plus non-equity lending from parent to subsidiary.
  - Proxy for parent non-equity lending to subsidiaries (when detailed data unavailable): subtract total customer deposits from total assets at bank level.
  - Downstream exposure to branches legally equals entire branch assets.

### Expected Downstream Exposure Losses
- Expected downstream index (D_i) combines downstream exposures with borrowing-country crisis probabilities and loss-given-default (LGD) estimates:
  - D_i = (1 / Z_i) * sum_{j=1 to N} [ Pr_crisis_j * (A_ij + B_ij + C_ij + D_ij) * LGD_j ]
    - (Source expression in text: D_i estimated as the summation of crisis probability times exposures and LGDs scaled by Z_i.)
- Definitions:
  - Z_i is a scaling factor (Tier I Capital or total banking sector assets in country i).
  - Pr_crisis_j is the probability of crisis in borrower country j.
  - LGD_j is the loss-given default in borrower country j.
- Distinguishing liquid vs. non-liquid assets in LGD application:
  - B_ij * LGD_j can be decomposed to reflect parent banks’ legal ceiling exposure to subsidiaries and losses on non-liquid assets:
    - B_ij_LGD_j = min( B_ij, LGD_j * (assets_other_subs_j + (deposits_subs_j - assetstotal_subs_ij) ) )
    - (Text formulation: B_ijLGD_j written as: LGD_sub_j * min( assets_other_subs_j + deposits_subs_ij - assetstotal_subs_ij , B_ij ) — preserving the interaction of legal ceiling and losses on non-liquid assets.)

### B.2 - Upstream Analysis
- Purpose: Focuses on a borrowing country’s funding risk to crises in its creditor banking systems; summarizes potential rollover risks from:
  - direct cross-border borrowing;
  - borrowing from foreign affiliates funded by their creditor countries’ parent banks.
- Upstream exposure for borrower country j relative to creditor country i:
  - ExposureUpstream_ij = (1 - min(loan_deposit_ratio_ij, 1)) * (claimsCrossBorder_ij + claimsLocal_ij)
    - Text formula: (1,_(1) * Min loan deposit ratio expression multiplied by cross-border claims and local claims)
- Definitions:
  - claimsCrossBorder_ij = direct cross-border claims from country i on country j.
  - claimsLocal_ij = affiliates (subsidiaries and branches) claims of parent banks from country i on country j.
  - (1,_(1) Min loan_deposit_ratio_ij) is a proxy for the proportion of loans not financed by local consumer deposits; higher deposit-to-loan ratio implies lower share of local claims financed by parent bank resources and/or wholesale financing.

### Expected Upstream Exposure Losses
- Expected upstream index (U_j) combines upstream exposures with creditor-country crisis probabilities:
  - U_j = (1 / Z_j) * sum_{i=1 to N} [ Pr_crisis_i * (1 - min(loan_deposit_ratio_ij, 1)) * (claimsCrossBorder_ij + claimsLocal_ij) ]
- Definitions:
  - Z_j is a scaling factor (e.g., GDP or total banking sector credit in country j).
  - Pr_crisis_i is the probability of a banking crisis in creditor country i.

### Adjustment for Coverage Break-in-Series and Exchange Rate Variations (annex material relevant to exposure calculations)
- Coverage break-in-series:
  - BIS reports 84 series breaks during 2006-12 in BIS consolidated banking statistics at ultimate borrower basis.
  - About one quarter (21) of those breaks are classified as "coverage" break-in-series (expansion of banking sector coverage).
  - Coverage breaks can create apparent level changes not reflecting true changes in exposures (examples given: US inclusion of former investment banks in 2009Q1 produced a USD 1,333.7 billion accounting increase; France merger in 2009Q3 triggered a USD -234.0 billion decrease).
  - Adjustment approach:
    - BIS reports what the series would have been without the break; cross-border and affiliate claims series are adjusted at the bilateral creditor-borrower level using pre-break series.
    - Adjustments before the coverage breaks are performed so that growth rates in adjusted and unadjusted series are the same in periods without coverage breaks; assumption: size of adjustment before break evolved at the same pace as unadjusted series.
  - Table C.1 excerpts (coverage break examples with Change in Foreign claims (Billions USD)):
    - 2011 4 Ireland: coverage, Change in Foreign claims = -125.6
    - 2009 1 United States: coverage, Change in Foreign claims = 1333.7
    - 2009 3 France: coverage, Change in Foreign claims = -234.0
    - 2007 4 France: coverage, Change in Foreign claims = 312.0
    - 2007 1 Italy: coverage, Change in Foreign claims = 621.8
    - 2006 1 Canada: coverage, Change in Foreign claims = 61.1
    - 2006 1 Netherlands: coverage, Change in Foreign claims = 55.7
    - (Table C.1 contains additional entries and notes; classification based on criteria developed in the paper.)
- Exchange rate variations:
  - BIS CBS are reported in US dollars; movements in series can reflect exchange rate changes rather than underlying position changes.
  - Three corrections at the bilateral level:
    1. Domestic-currency denominated local affiliates’ claims corrected following the bilateral US dollar–domestic currency exchange rate; domestic-currency local affiliates’ claims proxied by their share of total BIS CBS foreign claims at immediate borrower basis.
    2. Identification of foreign-currency denominated local affiliates’ claims, assumed to be in Euros in Europe and US dollars elsewhere.
    3. Bilateral CBS cross-border claims adjusted using currency breakdown proxy from BIS locational banking statistics (US Dollar, Euro, British Pound, Japanese Yen, Swiss Francs).
  - Magnitude: exchange rate movement adjustments can be about five percent of total foreign claims in some quarters and often larger than coverage break-in-series adjustments; US/EURO exchange rate evolution had large influence.

### Adjustment of Downstream and Upstream Exposures (practical application)
- Time-series analysis of upstream and downstream exposures focuses on on-balance sheet items due to limitations adjusting off-balance sheet items (currency breakdown of off-balance sheet items not available).
- Adjusted on-balance sheet downstream exposures (A_ij + B_ij + C_ij):
  - Step 1: Replace unadjusted cross-border exposures with adjusted cross-border exposures (A_ij_adjusted).
  - Step 2: Adjust downstream exposures to subsidiaries and branches (B_ij + C_ij) using an assumption that currency breakdown of assets is similar to liabilities (local deposits):
    - claimslocal_adjusted_ij + claimslocal_ij = CB_adjusted_ij + CB_ij  (text expression: (adjusted claimslocal + claimslocal) CBCB = ...)
- Adjusted on-balance sheet upstream exposures:
  - ExposureUpstream_ij_adjusted = (1 - min(loan_deposit_ratio_ij_adjusted, 1)) * (claimsLocal_ij_adjusted + claimsCrossBorder_ij_adjusted)

### Annex D – Statistical Appendix (summary of variables and key statistics used in econometric analysis)
- Notes on sample and outliers:
  - Aggregation of bilateral creditor-borrower data reduces number of outliers.
  - After coverage break-in-series adjustments, outliers remaining: 2 cases in downstream analysis out of 624 observations; 43 cases in upstream analysis out of 2,591 observations. These observations were excluded; they reflect relatively small nominal variations over very small bases.
- Table D1 — Variables used in Downstream Estimations (selected statistics)
  - ∆D_it (Growth rate (in percent) of the adjusted on-balance sheet downstream exposure of creditor banking system i in quarter t; dependent variable)
    - Mean = 0.63
    - Median = 0.63
    - Std Dev = 10.44
    - Min = -49.85
    - Max = 55.12
  - ∆GDP_ijt (Creditor banking system i weighted average of the GDP growth in its borrower countries j in quarter t (in percent))
    - Mean = 0.40; Median = 0.55; Std Dev = 0.74; Min = -2.6; Max = 2.8
  - Cred_Systemic_Crisis (dummy reflecting presence of systemic banking crisis in banking system i in quarter t)
    - Mean = 0.22; Median = 0.00; Std Dev = 0.41; Min = 0.0; Max = 1.0
  - Cred_DLR (Ratio of bank deposit to loans in the creditor banking system i)
    - Mean = 0.78; Median = 0.71; Std Dev = 0.35; Min = 0.22; Max = 2.6
  - Cred_CB_Share (share of direct cross-border lending in total foreign claims (in percent))
    - Mean = 63.90; Median = 62.60; Std Dev = 24.30; Min = 16.9; Max = 100.0
  - VIX (normalized quarterly volatility of the S&P 500 financial index)
    - Mean = 2.00; Median = -0.03; Std Dev = 10.38; Min = -10.83; Max = 6.8
  - TED Spread (three-month US dollar Libor minus three-month US treasury bill rate)
    - Mean = 64.01; Median = 40.63; Std Dev = 56.25; Min = 15.5; Max = 245.2
- Table D2 — Variables used in Upstream Analysis (selected statistics)
  - ∆U_jt (Growth rate (in percent) of the adjusted on-balance sheet upstream exposure of borrower country j in quarter t; dependent variable)
    - Mean = 2.08; Median = 1.59; Std Dev = 16.87; Min = -85.65; Max = 96.51
  - ∆GDP_jt (Borrower country j GDP growth in quarter t (in percent))
    - Mean = 0.99; Median = 1.14; Std Dev = 2.95; Min = -26.39; Max = 15.56
  - Creditor_Crisis (proportion of creditor banking systems of borrower country j that are experiencing systemic banking crisis in quarter t)
    - Mean = 0.22; Median = 0.00; Std Dev = 0.29; Min = 0.000; Max = 0.93
  - Borrower_DLR (ratio of bank deposit to loans in borrower country j)
    - Mean = 0.88; Median = 0.84; Std Dev = 0.49; Min = 0.32; Max = 4.08
  - Borrower_CB_Share (share of direct cross-border in borrower country foreign borrowing (lag one period; in percent))
    - Mean = 46.74; Median = 44.0; Std Dev = 18.99; Min = 4.4; Max = 100.0
  - VIX
    - Mean = 2.40; Median = 0.21; Std Dev = 11.4; Min = -10.83; Max = 6.8
  - TED Spread
    - Mean = 74.94; Median = 44.96; Std Dev = 90.0; Min = 15.5; Max = 245.2
- Table D.3 (Selected econometric results summary, determinants of change in upstream exposures 2006Q2-2012Q1, Panel OLS with Fixed Effects; dependent variable: Change in Upstream Exposure (in percent))
  - Sample size and structure:
    - Observations = 2,458 (across specifications)
    - Number of borrower countries = 112
  - Reported R2 values across specifications: 0.057, 0.078, 0.065, 0.055, 0.090, 0.119, 0.154, 0.124, 0.122, 0.125
  - Selected coefficient signs and significance (from table excerpts):
    - ∆GDP_ijt: coefficient example = 0.470** (standard errors shown in parentheses across columns)
    - Cred_Syst_Crisis: coefficients include -9.581***, -7.576***, -5.462*, -7.604***, -11.01*** (robust standard errors reported)
    - Borrower_CB_Share: coefficients include -0.250***, -0.322***, -0.310***, -0.318***, -0.306***, -0.296*** (robust standard errors reported)
    - VIX: coefficients include -0.311***, -0.258***, 0.396, -0.348***, -0.137** (robust standard errors reported)
    - Interaction terms reported with significance in some specifications:
      - VIX * Cred_Syst_Crisis: -0.470***, -0.378** (standard errors reported)
      - VIX * Borrower_CB_Share: -0.003*** (standard error reported)
      - Cred_Syst_Crisis * Borrower_CB_Share: -0.156*** (standard error reported)
  - Model features:
    - Quarterly dummy included in many specifications.
    - Borrower fixed effects included.
    - Time fixed effects included in at least one specification.
  - Note: Robust standard errors are in parentheses and clustered at the borrower country level. Asterisks denote significance levels: ***, **, * indicating significance at 1%, 5% and 10% levels, respectively.

*Source: _wp1309 - ANNEX B – CALCULATION OF DOWNSTREAM AND UPSTREAM EXPOSURES*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2013/_wp1309.pdf_