## Cross-Border Portfolio Claims (_060111)

## Source details

**Canonical URL:** [Cross-Border Portfolio Claims (_060111)](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/np/pp/eng/2011/_060111.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/np/pp/eng/2011/_060111.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/np/pp/eng/2011/_060111.pdf.json)

---

### Executive Summary
- Objectives:
  - Map cross-border financial linkages and identify factors that drive them to inform the design of a global financial safety net (GFSN).
  - Build on previous staff work and complement the companion paper on the Analytics of Systemic Crises and the Role of Global Financial Safety Nets.
  - Emphasize the value of a GFSN that forestalls localized liquidity shocks propagating through the global financial network into large-scale systemic crises.
- Key findings:
  - Cross-border financial linkages have increased dramatically and become more complex.
  - A few “core” advanced economies (AEs), including some financial centers, continue to dominate linkages across asset classes and regions, both as sources and recipients.
  - Emerging markets’ (EMs) strongest linkages remain with AEs, although cross-EM linkages rose very rapidly during the last decade (from a low base).
  - Increased cross-border linkages:
    - Promote risk diversification at the individual country level.
    - Generate a network externality that makes the global financial network more prone to systemic risk—risk that shocks to a “core” node lead to a breakdown of the entire network.
    - Reduce investor information about specific exposures as complexity grows, amplifying systemic risks from panic responses.
- Implications:
  - Countries with shallow domestic financial markets and concentrated exposures to a few lenders are more prone to synchronized shifts in cross-border flows.
  - Common factors (such as global risk aversion) increasingly drive global financial markets and tend to intensify abruptly during stress, amplifying shock transmission.
  - Potentially large costs can be imposed on “crisis bystanders” (countries with relatively strong fundamentals) from systemic shocks.
  - These findings reinforce the case for a GFSN designed to ring-fence such countries from systemic shock contagion.

### Mapping the Linkages and Stylized Facts
- Stylized facts and patterns:
  - Interconnectedness is much stronger for AEs than EMs and has generally increased in the last decade across asset classes.
  - In EMs, interconnectedness is higher for cross-border bank claims than for portfolio claims.
  - Concentration is significantly higher in EMs than AEs but has generally declined over the past decade, with the notable exception of cross-border bank claims in European EMs where concentration increased rapidly pre-crisis.
  - More than 90 percent of claims issued by residents in EMs are held by residents in AEs or financial centers; share held in EMs is in most cases fairly small, generally 5 percent or less.
  - Cross-EM linkages increased very rapidly over the last decade but remain small compared to linkages from AEs.
- Role of core nodes:
  - A few core AEs and financial centers dominate the web of cross-border linkages across assets and regions.
  - Many core countries are core for at least two of the three asset classes (portfolio equity, portfolio debt, bank claims); Germany, France, Japan, Ireland, Netherlands, U.K., U.S. are core sources and recipients for all three classes (as identified where data permit).
  - Large overlap between top sources and top recipients implies shocks can be transmitted in both directions, accelerating spread.
- Data and scope caveats:
  - CPIS and BIS coverage limitations (e.g., China non-participation in BIS/CPIS) constrain full assessment; identification of core sources limited to participating countries.
  - Paper excludes FDI and off-balance sheet banking positions due to data limitations.

### Systemic Instability and Shock Transmission
- Network externalities and fragility:
  - Increased interconnectedness creates a network externality that raises systemic risk; individual countries do not internalize this externality when choosing linkages.
  - Complexity and incomplete information increase incentives for herding, flight to quality, and liquidity crunches; small or less connected countries can trigger systemic market responses if crises act as “wake-up” calls.
- Concentration risks:
  - Recipient countries with concentrated exposures to a few sources face amplified deleveraging if a main source is hit.
  - Higher concentration increases exposure to local/regional shocks and reduces effectiveness of regional/local risk-sharing.
- Empirical stylized facts on capital flows to EMs:
  - Deleveraging:
    - Many EMs experienced sharp reversals in net capital inflows during the crisis, often driven by sudden stops in external liabilities with limited action on external assets.
    - Time-series correlation between gross and net capital inflows: correlation often close to zero in non-EMs; tends to be closer to one in EMs.
  - Synchronization:
    - Shifts in cross-border exposures can be highly synchronized at times of stress; episodes of large inflows often end together within a narrow period (e.g., 1997−98 and 2008−09).
  - Volatility:
    - Post-crisis recovery saw a shift toward portfolio flows, historically relatively more volatile.
    - Compared to past episodes, average pace of portfolio inflows during the ongoing wave has more than quadrupled.
    - Portfolio flows are more closely related to global factors and potentially more exposed to global shocks.

### Determinants of Linkages — Gravity and Micro Evidence
- Empirical determinants (gravity framework and bank-group analysis):
  - Geography and history:
    - Distance is a statistically significant determinant; higher exposure tends to be built toward nearer countries.
    - Common language and common legal system are associated with stronger exposure, particularly for equity and bank claims.
  - Economic factors:
    - Larger countries (source and recipient) have larger cross-border exposures; size and income explain about 40 percent of interquartile differences across asset classes.
    - Increased trade integration and common currency are associated with stronger linkages.
  - Financial factors:
    - Financial development in both source and recipient countries feeds stronger cross-border linkages—strongest for bank claims.
    - Financial center status in source and recipient countries is associated with stronger cross-border exposures.
    - Recipient countries with more open capital accounts tend to have stronger portfolio exposure across all asset classes.
- Bank-group findings:
  - In selected European and Latin American EMs, foreign-owned group assets represent between 30 and 40 percent of total domestic assets on average; parents mostly reside in AEs and financial centers.
  - Cross-border groups have tended to move “closer to home” (regional agglomeration) and show heterogeneous intra-group concentration trends across parent countries.

### National, Regional, and Global Defenses
- National mitigants:
  - Accumulating international reserves (self insurance), taxing externality-generating flows, administrative/prudential measures.
  - Constraints:
    - Taxing the externality is difficult due to measurement challenges and multilateral benefits of interconnectedness.
    - Self insurance constrained by fiscal costs of holding low-yielding foreign assets, diminishing returns (IMF, 2010d), and reluctance to use reserves in crises.
    - Recourse to taxes and self insurance has varied considerably across countries (IMF, 2011c; Magud and others, 2011).
- Regional and global mechanisms:
  - Where national defenses are insufficient, regional and global financing mechanisms can cushion residual volatility’s impact on individual countries and the system.
  - The 2008–09 crisis underscored the value of an effective global mechanism to coordinate liquidity injections and policy responses.
  - Evidence: several “crisis bystanders” with relatively strong fundamentals suffered deep output contractions in 2008–09 despite being hit later and less severely.

### Empirical Evidence — Macro and Micro Quantification
- Global drivers and variance decomposition (Annex IV):
  - Global factors explained on average around 25 percent of the cross-time variation of gross total capital inflows to EMs.
  - For portfolio inflows, the estimated share explained by global factors rose to 54 percent.
  - When domestic variables were omitted, global-factor shares rose to 65 percent for total inflows and 87 percent for portfolio inflows (interpreted as upper bounds).
  - Seven-country panel shares (percent) for Aggregate Inflows and Portfolio Inflows:
    - Brazil: 30; 59
    - Indonesia: 18; 88
    - Korea: 17; 37
    - Peru: 5; 17
    - South Africa: 22; 5
    - Thailand: 20; 88
    - Turkey: 11; 41
  - Alternative three-step variance-decomposition (48 EMs): for net inflows global factors explained less than 20 percent; for total gross inflows less than a third.
- Macro spillovers and VARs:
  - VAR evidence: shocks from AE bloc have quantitatively important effects on EM bloc; reverse effects much smaller; an important fraction of growth spillovers is attributable to financial shocks in the AE bloc.
- Bank-group level regressions:
  - EM subsidiaries’ lending is affected by subsidiary leverage/liquidity, local macroeconomic conditions, shocks to parent financial conditions, and parent-country macro conditions.
  - Parent-to-subsidiary effects weaker when subsidiaries are located in AEs.
- Interconnectedness and crisis performance:
  - Regression relating 2008 output contraction to fundamentals and network indices:
    - In-degree coefficient: 14.33* (7.66)
    - HHI coefficient: -11.64* (6.62)
    - Foreign bank claims (percent of GDP, in logs): -2.44*** (0.81) and -1.44 (0.86)
    - GIR in percent of (short-term debt at residual maturity plus current account deficit, in logs): 3.11*** (0.90) and 2.10** (0.97)
    - Observations: 40; R-squared: 0.53 (two reported columns).
  - Interpretation: higher in-degree (more diversified) associated with smaller output contractions; higher HHI (more concentrated) associated with larger output contractions.
- Deleveraging regression (change in log claims between 2007 and 2008):
  - Sample observations: 1,262; 2,576; 1,212 (three columns).
  - R-square: 0.11; 0.17; 0.08.
  - Selected coefficients:
    - Log of claims in 2007: -0.09 P> |t| 0.02; -0.16 P> |t| 0.01; -0.08 P> |t| 0.01.
    - EM recipient dummy: -0.23 P> |t| 0.15; -0.17 P> |t| 0.07; -0.13 P> |t| 0.07.
    - EM source dummy: 0.20 P> |t| 0.12; 0.26 P> |t| 0.07; -0.19 P> |t| 0.09.
  - Key finding: the larger the initial exposure, the larger the subsequent deleveraging; EMs experienced larger deleveraging than AEs.
- Explaining bilateral financial exposures (Annex VII baseline random-effects coefficients, 2001–09):
  - Log GDP, recipient: Equity 0.572*** (0.024); Debt 0.604*** (0.052); BIS bank claims 0.727*** (0.035).
  - Log GDP, source: Equity 0.122*** (0.027); Debt 0.532*** (0.065); BIS bank claims 0.372*** (0.051).
  - Log distance: Equity -0.597*** (0.056); Debt -0.702*** (0.088); BIS bank claims -0.763*** (0.081).
  - Common language: Equity 0.622*** (0.119); Debt 0.420*** (0.161); BIS bank claims 0.879*** (0.130).
  - Common currency: Equity 1.618*** (0.161); Debt 1.542*** (0.136); BIS bank claims 0.226 (0.139).
  - Financial depth indicator, recipient: Equity 0.243*** (0.036); Debt 0.235*** (0.055); BIS bank claims 0.696*** (0.075).
  - Financial center dummy, source: Equity 0.483*** (0.097); Debt 1.460*** (0.116); BIS bank claims 1.136*** (0.097).
  - Observations: Equity 22,631; Debt 10,711; BIS bank claims 10,498.
  - Number of country pairs: Equity 3,511; Debt 1,396; BIS bank claims 1,504.
  - R2: Equity 0.664; Debt 0.715; BIS bank claims 0.695.

### Network Analysis — Metrics and Thresholds (Annex II)
- Link definition and threshold:
  - A link is the stock of claims issued by a recipient country to a source country.
  - A link is active if the recipient‘s cross-border liabilities are above 0.2 percent of the recipient‘s GDP (threshold preserved).
- In-degree and out-degree:
  - In-degree: number of active creditor locations a country borrows from; normalized between 0 and 1.
  - Out-degree: number of recipients a source lends to; active links require exposures above 0.2 percent of the source‘s GDP.
- Concentration — Herfindahl-Hirschman Index (HHI):
  - HHI defined as sum of squared shares si^2, normalized to range between 0 and 1.
- Data sample and periods:
  - Cross-border claims studied for 57 emerging markets and 24 advanced economies.
  - Networks constructed for period 1999-2009.
  - CPIS data on portfolio holdings available annually for 2001-2009.
  - BIS analysis restricted to 24 source economies reporting consistently 1999-2009.

### Takeaways and Implications for GFSN Design
- Main high-level takeaways:
  - (i) As financial linkages mainly emanate from AEs, shifts in cross-border exposures are a key channel for the propagation of shocks globally; EMs, in particular, are exposed to such shifts, which can be extremely large, given EMs relatively shallow financial markets.
  - (ii) Growing global financial linkages also imply increasing importance of global drivers for EM asset prices movements.
  - (iii) Both macro and micro (banking) empirical results show that cross-border financial shocks from AEs and network interconnectedness could have significant macroeconomic impact on EMs.
  - (iv) Empirical evidence also suggests that, conditional on a systemic shock occurring, more interconnected countries suffered a smaller output decline during the crisis; however, countries with more concentrated exposure suffered a more pronounced output contraction.
- Design-relevant principles:
  - A GFSN must account for the dominance of financially-developed AEs as both sources of shocks and large counterparties.
  - Geography and regional agglomeration limit the effectiveness of purely regional risk-sharing; global mechanisms are necessary.
  - Insurance mechanisms addressing sudden shifts in cross-border exposures driven by aggregate/global shocks are essential complements to local or regional defenses.
  - National self-insurance has limits; residual volatility requires regional or global financing buffers to prevent localized liquidity runs from becoming systemic crises.

### Conclusions (Section VI)
- Cross-border financial linkages have increased dramatically in holdings and asset-price co-movements.
- Linkages are dominated by a few AEs and financial centers that form "core nodes" in the system.
- EMs' cross-border holdings remain relatively small despite rapid growth; the bulk of EM liabilities are held by core nodes. EMs with shallow domestic capital markets are particularly exposed to large shifts in cross-border exposures.
- Shocks can be highly synchronized and transmission can be non-linear, with potentially significant real costs to EM economies.
- There is a trade-off between country-level benefits from increased international risk diversification and increased systemic risk from heightened interconnectedness.
- National defenses may not eliminate network externalities; regional and especially global financing mechanisms have a role to cushion residual volatility.
- The 2008-09 crisis underscores the value of an effective global mechanism to coordinate liquidity injections and other policy responses: limited opportunities to diversify against cross-border shock risk point to significant gains from an effective GFSN.

*Source: IMF staff paper excerpt (Executive Summary; Sections I–VI; Annexes II, IV, VII) as provided in content unit _060111.*

### 1. Cross-Border Portfolio Claims .......................................................................................

### 1. Cross-Border Portfolio Claims

### Executive Summary
- Objectives:
  - Map cross-border financial linkages and identify factors that drive them to inform the design of a global financial safety net (GFSN).
  - Build on previous staff work and complement the companion paper on the Analytics of Systemic Crises and the Role of Global Financial Safety Nets.
  - Emphasize the value of a GFSN that forestalls localized liquidity shocks propagating through the global financial network into large-scale systemic crises.
- Key findings:
  - Cross-border financial linkages have increased dramatically and become more complex.
  - A few “core” advanced economies (AEs), including some financial centers, continue to dominate linkages across asset classes and regions, both as sources and recipients.
  - Emerging markets’ (EMs) strongest linkages remain with AEs, although cross-EM linkages rose very rapidly during the last decade (from a low base).
  - Increased cross-border linkages:
    - Promote risk diversification at the individual country level.
    - Generate a network externality that makes the global financial network more prone to systemic risk—risk that shocks to a “core” node lead to a breakdown of the entire network.
    - Reduce investor information about specific exposures as complexity grows, amplifying systemic risks from panic responses.
- Implications:
  - Countries with shallow domestic financial markets and concentrated exposures to a few lenders are more prone to synchronized shifts in cross-border flows.
  - Common factors (such as global risk aversion) increasingly drive global financial markets and tend to intensify abruptly during stress, amplifying shock transmission.
  - Potentially large costs can be imposed on “crisis bystanders” (countries with relatively strong fundamentals) from systemic shocks.
  - These findings reinforce the case for a GFSN designed to ring-fence such countries from systemic shock contagion.

### Context and Motivation
- Global trends:
  - Global economic linkages have intensified dramatically over the past two decades, underpinned by an exponential rise in trade and financial flows (Figure 1).
  - Cross-border linkages have been dominated by financial flows among AEs; flows to and among EMs have risen in importance both in absolute terms and relative to their economies.
  - Patterns of linkages have grown in complexity; example: thickening of financial links among European EMs during the last decade (Figure 2).
- Focus of the paper:
  - Understand the evolving nature of cross-border financial linkages—the “plumbing” of the global economy—to map channels for shock transmission and tailor policy responses.
  - Internalize lessons from the recent crisis and complement the Systemic Crises paper.
  - A key goal is mapping and explaining drivers of cross-border linkages and their macroeconomic consequences to support establishment and design of a GFSN that mitigates global liquidity shocks.
- Trade-off highlighted:
  - Tension between (a) country-level benefits from increased international risk diversification (which pushes toward increased interconnectedness) and (b) increased systemic risk at the global level from heightened interconnectedness.
  - Network theory provides insights: increased links can improve resilience at low interconnectedness but can create latent fragility where shocks to a core node propagate non-linearly through the network.

### Mapping the Linkages and Stylized Facts
- Stylized facts from network measures (Annex II):
  - Interconnectedness is much stronger for AEs than EMs and has generally increased in the last decade across asset classes.
  - In EMs, interconnectedness is higher for cross-border bank claims than for portfolio claims.
  - Concentration is significantly higher in EMs than AEs but has generally declined over the past decade, with the notable exception of cross-border bank claims in European EMs where concentration increased rapidly pre-crisis.
- Role of “core” nodes:
  - A few core AEs and financial centers dominate the web of cross-border linkages across assets and regions.
  - EMs maintain the strongest linkages with these AEs despite rising cross-EM linkages.

### Systemic Instability and Shock Transmission
- Network externalities and fragility:
  - Increased interconnectedness creates a network externality that raises systemic risk; individual countries do not internalize this externality when choosing linkages.
  - Complex networks increase incomplete information, raising potential for herding, flight to quality, and liquidity crunches.
  - Small or less connected countries can trigger systemic market responses if crises act as a wake-up call to creditors.
- Concentration risks:
  - Recipient countries with concentrated exposures to a few sources face amplified deleveraging if a main source is hit.
  - Higher concentration increases exposure to local/regional shocks and reduces effectiveness of regional/local risk-sharing.
- Empirical implications:
  - Common factors increasingly drive global financial markets; these factors tend to spike during stress periods, intensifying shock transmission.
  - Countries with shallow financial markets and concentrated lender exposures are especially prone to synchronized shifts in cross-border flows.
  - Large costs can be imposed on crisis bystanders—supports need for GFSN to ring-fence them.

### Determinants of Linkages
- Empirical evidence:
  - Geographical and historical factors remain important determinants of cross-border linkages.
  - Stronger linkages occur among economies that are closer, larger, more developed, and financially more advanced.
- Policy implication:
  - Insurance mechanisms against sudden shifts in cross-border exposures driven by aggregate or global shocks are essential complements to local or regional risk-sharing mechanisms.

### National Defenses and Limitations
- National-level mitigants:
  - Accumulating international reserves (self insurance), taxing externality-generating flows, or administrative/prudential measures can mitigate capital flow volatility.
- Constraints and complications:
  - Taxing away the externality is difficult due to measurement challenges and accounting for multilateral benefits of interconnectedness.
  - Self insurance constrained by fiscal costs of holding low-yielding foreign assets, diminishing returns to reserve accumulation (IMF, 2010d), and reluctance to use reserves in crises.
  - Recourse to taxes and self insurance has varied considerably across countries (IMF, 2011c; Magud and others, 2011).

### Regional and Global Defenses
- Role of regional and global mechanisms:
  - Where national defenses are insufficient, regional and global financing mechanisms can cushion residual volatility’s impact on individual countries and the system.
  - The 2008–09 crisis underscored value of an effective global mechanism to coordinate liquidity injections and policy responses.
- Evidence from crisis:
  - In 2008–09, several “crisis bystanders” (countries with relatively strong fundamentals) suffered deep output contractions despite being hit later and less severely than more vulnerable countries (Figure 3).
  - This outcome and limited opportunities to diversify against cross-border shocks point to significant global welfare gains from an effective GFSN.

### Country Coverage and Scope
- Analytical focus:
  - Empirical analysis mainly focuses on financially-developed EMs with (partially) open capital accounts, for whom linkages with AEs and financial centers dominate.
- Scope limitations:
  - Some linkages—such as spillovers from reserve accumulation in large EMs, trade and commodity price linkages among EMs, AEs, and LICs—are important but not analyzed here, in part due to data limitations (Box 2) and because they are less likely to propagate global shocks.
- Annexes and supporting analysis:
  - The paper includes network measures, country groupings, and additional empirical analysis in Annexes (e.g., Annex II on network analysis).

*Source: IMF staff paper excerpt (Executive Summary and Sections I–II, including Box 1 and Figures 1–3).*

### 7.      Relation to other staff work. This paper complements recent staff work on financial

### 7. Relation to other staff work. This paper complements recent staff work on financial

### Relation to other staff work
- Complements recent staff work on financial linkages presented in IMF 2009a, 2010b and 2010c.
- Builds on recent staff work on cross-border capital flows (IMF 2010a and 2011e).
- Contribution: brings together aspects of the global financial infrastructure relevant for the design of a GFSN.
- Should be read in conjunction with the Systemic Crises paper, which focuses on triggers, propagation, and policy responses to past systemic crises to assess adequacy of the existing global financial safety net for future systemic shocks.

### Data sources and limitations (Box 2)
- Main datasets used:
  - BIS Consolidated Banking Statistics (immediate borrower basis; group worldwide-consolidated, including claims of subsidiaries and branches).
    - Only a subset of source countries (24) reported data consistently through 1999-2009.
    - Only a few EM countries participated in recent years as source countries.
    - Series contain a few breaks; these breaks were not taken into account in the analysis.
  - IMF Coordinated Portfolio Investment Survey (CPIS), annual survey of bilateral portfolio holdings.
    - Limitations:
      - Not all economies participate (e.g., some oil-exporting economies with large sovereign wealth funds, offshore centers, and economies with large holdings such as China and Taiwan province of China may be absent).
      - Possible under-reporting of cross-border assets due to incomplete institutional coverage.
      - May not capture portfolio holdings of entities resident in a reporting country but owned by foreign investors; holdings in financial centers typically do not capture ultimate destination.
      - Implied external liabilities from CPIS typically below those in a country‘s International Investment Position.
    - Ongoing enhancements:
      - Efforts to increase frequency and shorten timeliness of data.
      - Collect data on institutional sector of foreign debtors on an encouraged basis.
      - Implementation of enhancements beginning with the 2013 data and increasing participating countries are part of the G-20 Data Gaps Initiative.
- Exclusions and caveats:
  - Paper does not consider foreign direct investment (FDI) as it is generally viewed as relatively stable, though special purpose vehicles and conduits may reduce its stability.
  - Banking sector linkage analysis excludes off-balance sheet positions owing to data limitations.

### Snapshot of cross-border financial linkages (III)
- Stylized facts:
  - AEs still dominate cross-border financial linkages.
  - More than 90 percent of claims issued by residents in EMs are held by residents in AEs or financial centers; share held in EMs is in most cases fairly small, generally 5 percent or less.
  - Exception: debt holdings of Asian EMs where linkages to EMs are relatively more important (Milesi-Ferretti and others, 2010).
  - Cross-EM linkages increased very rapidly over the last decade but remain small compared to linkages from AEs.
- Core nodes:
  - Relatively few countries act as ―core‖ nodes; same countries tend to dominate across asset classes.
  - Core nodes are mostly AEs or financial centers; only a few EMs appear in the core list.
  - Overlap across asset classes: many core countries are core for at least two of the three asset classes (portfolio equity, portfolio debt, bank claims); several (Germany, France, Japan, Ireland, Netherlands, U.K., U.S.) are core sources and recipients for all three classes.
  - Large overlap between top sources and top recipients implies shocks can be transmitted in both directions, accelerating spread.
- Participation and measurement limits:
  - Identification of core sources limited to participating countries in BIS and CPIS; prominent non-participants (e.g., China) constrain full assessment.
- Empirical snapshot (Figure 5 summary points preserved from figure labels):
  - Composition by residence of claimholders (percent of total claims, 2009) shows large shares attributable to Advanced economies and Financial centers across Portfolio Debt, Portfolio Equity, and Bank Claims for regions including EM Asia, EM Europe, and EM Latin America.

### Have EMs been overlooked? (paragraph 11 and Box 4)
- Using a more comprehensive database including reserve assets, Milesi-Ferretti and others, 2010 confirm EMs account for a small part of cross-border financial linkages.
  - In 2007, share of emerging Asia including China in external assets holdings was only about 5 percent of total global external assets; same for external liabilities.
- China:
  - Difficult to establish role as source in cross-border linkages because China does not report to BIS and CPIS.
  - U.S. Treasury TIC data shows China‘s penetration in U.S. asset market capitalization remains limited and concentrated in sovereign debt markets.
  - Flow example: during 2001-2010, the U.S. imported US$5.8 trillion from the rest of the world (measured as cumulative current account balance over the period); this flow was largely supplied by a limited number of economies, most notably Japan, China and oil producers, and largely took the form of accumulation of official reserves.
- Core-node EMs identified:
  - Brazil appears as a core top source; Brazil and Korea appear as core top recipients (based on BIS and CPIS participation).

### Bank ownership linkages (paragraphs 12 and Figure 6)
- Cross-border bank ownership has increased cross-border financial linkages between AEs and EMs.
- Staff-constructed dataset shows:
  - Importance of cross-border banking groups has grown over time, most notably in European and Latin American EMs.
  - In these EMs, on average assets belonging to foreign-owned groups represent between 30 and 40 percent of total domestic assets, with group parents mostly residing in AEs and financial centers (Figure 6).
  - Importance of cross-border asset ownership is much lower in AEs.
  - Cross-border groups are still largely owned by a parent bank residing in AEs and financial centers.

### Takeaways (paragraph 13)
- (i) Cross-border financial linkages are still overwhelmingly to AEs.
- (ii) There are relatively few countries, mostly AEs and financial centers, that act as ―core‖ nodes in the global financial system.

### Cross-border financial linkages and shock transmission (IV)
- Capital flows and shock transmission:
  - Shifts in cross-border exposures, especially rapid synchronized deleveraging, are a key source of systemic risks.
  - EMs are most exposed since bulk of their exposures are toward core nodes (AEs or financial centers), which tend to propagate global shocks.
- Stylized facts about capital flows to EMs (drawn from IMF, 2011e):
  - Deleveraging:
    - Many EMs experienced sharp reversals in net capital inflows during the crisis, often driven by sudden stops in external liabilities with limited action on external assets.
    - Changes in EMs‘ gross external liabilities often associated with changes in net external liabilities more than in AEs.
    - Time-series correlation between gross and net capital inflows:
      - Correlation often close to zero in non-EMs; tends to be closer to one in EMs.
      - Implication: shifts in gross inflows are normally offset by changes in outflows in AEs but not in EMs.
      - Consequence: EMs lack the flexibility afforded by greater interconnectedness and face higher one-way risk of deleveraging; EMs stand to benefit most from a global insurance mechanism.
  - Synchronization:
    - Shifts in cross-border exposures can be highly synchronized at times of stress.
    - Episodes of capital inflow surges often start at different times but often end together within a narrow period (Figure 9); e.g., sudden stop episodes of 1997−98 and 2008−09.
    - Suggests exogenous factors (e.g., shocks to global risk appetite) drive reversals; insurance cannot rely exclusively on local/regional mechanisms.
  - Volatility:
    - Post-crisis recovery saw a shift toward portfolio flows, historically relatively more volatile.
    - Compared to past episodes of capital flow surges, average pace of portfolio inflows during this ongoing wave has more than quadrupled.
    - Portfolio flows are more closely related to global factors and potentially more exposed to global shocks.
- Empirical summaries preserved from figures:
  - Figure 7: Change in Capital Flows (2008Q4 & 2009Q1 over 2008Q2 & 2008Q3) reported in percent of GDP for multiple EMs and AEs (figure lists country-level changes visually).
  - Figure 8: Correlation between Gross and Net Capital Inflows (12-month rolling windows) shows EMs, G7 excl. UK, Financial centers, and Other advanced markets trends through 2003Q4–2009Q4.
  - Figure 9: Capital Inflows to AEs and EMs - Gradual Buildups but Synchronized Stops shows interrupted episodes statistics (1997-98, 2001, 2008, 2010, etc.) and that 50% to 80% of cases experienced interrupted episodes.
  - Figure 10: Share of Gross Capital Inflows during Large Inflows Episodes (In percent of total inflows) compares 1995Q4-1998Q2, 2006Q4-2008Q2, and 2009Q3-2010Q2 for Emerging Markets and Advanced Markets across Direct Inflows, Portfolio Inflows, Other Inflows.
  - Figure 11: Net Private Capital Flows to EMs During Periods of Low Interest Rates and VIX, and High Growth Differentials (percent of GDP) shows flows by category: Bank and other private flows; Portfolio debt flows; Portfolio equity flows; Foreign direct investment (before, during, after).

*Source: Extracted from content unit _060111 (IMF staff paper excerpt).*

### 15.      Accounting for shifts in exposures. The stylized facts just summarized show that

### _060111 - 15.      Accounting for shifts in exposures. The stylized facts just summarized show that

### Global drivers
- Global factors can explain only around 25 percent of variations in total gross inflows to EMs (Annex IV).
- The importance of global factors for portfolio inflows is estimated, on average, at around 50 percent of total variation in portfolio inflows to EMs.
- The importance of global factors tends to shift over time; capital flows to EMs tend to be large during periods of:
  - low global interest rates,
  - low global risk aversion,
  - high growth differentials between emerging markets and advanced economies.
- Footnote caveat: omitting domestic variables increases the role of global factors significantly, suggesting country-specific domestic factors are themselves influenced, directly or indirectly, by global factors.

### Asset price co-movements
- Correlations among a broadening range of asset markets have been increasing over time.
- Common factors explain a significant fraction of EM cross-country asset price variation.
- The contribution of the first principal component to the total variation in EM external yields has grown over time, reaching almost 80 percent.
- EMs have been affected mostly on the receiving end of spillovers, meaning EMs tend to be on average net receivers of global/AM shocks.
- Global risk aversion (captured by the VIX index) has become an increasingly important source of volatility for global markets, with a spike at the time of the global crisis.

### Common characteristics (imitation and association)
- Shock transmission can occur through investors' perception of countries' common characteristics; countries sharing characteristics with an epicenter can be hit harder.
- Example: Latvia and other Eastern European countries in the 2008 global crisis—common features included:
  - significant presence of Western European banks,
  - a hard peg to the euro (currency boards),
  - rapid credit growth, asset price inflation, and large current account deficits.
- Even with weak direct financial links between some countries, common policy frameworks and vulnerabilities can create perceived association and contagion ("wake-up" calls to investors and depositors).

### Relative size of capital markets (amplification)
- EMs represented about a third of world GDP in 2009.
- EMs' stock markets and bank assets were around one fifth of these asset classes on a global level in 2009.
- Debt markets were less than one tenth of global debt markets when public and private debt markets are combined (2009).
- Example reallocation impact: a reallocation of 1 percent of assets from AE markets stock, public debt or bank assets corresponds to a shift of between 4 and 6 percent in terms of EM market size, and 20 percent for private debt markets.
- Larger AE financial markets are generally deeper and more liquid than EM counterparts, increasing the potential impact beyond size differences alone.

### Shift in EM assets (longer-run structural shifts)
- Rapid growth of EM holdings of external assets implies some large EMs' shifts in asset allocation could have significant repercussions for global financial markets; China highlighted as particularly important.
- China is projected to contribute to more than one third of global net wealth accumulation between 2010 and 2015.
- Staff estimates: further reserve accumulation by China in the order of US$600 billion during 2011-2015 would be needed in addition to the US$2 trillion accumulation under the baseline scenario to keep the real price of Chinese financial assets in line with the real price of U.S. assets.

### Macroeconomic effects of financial linkages
- Four pieces of empirical evidence indicate financial linkages explain a large portion of macroeconomic spillovers from AEs to EMs.

1) VAR evidence
- Impulse response functions from a vector autoregressive model on two economic blocs (AEs and EMs) show shocks from the AE bloc have a quantitatively important effect on the EM bloc, with the reverse not being the case.
- The exercise confirms an important fraction of growth spillovers can be attributed to financial shocks in the AE bloc.

2) Bank-group-level evidence
- Regression analysis on bank groups shows lending behavior of EM subsidiaries is affected by:
  - their own leverage and liquidity conditions,
  - local macroeconomic conditions,
  - shocks to the parent's financial conditions (liquidity, capital adequacy, profitability, non-performing loans),
  - macroeconomic conditions in the parent's home country.
- Parent-to-subsidiary effects are less statistically robust when a subsidiary is located in an AE, reinforcing that EMs tend to be particularly subject to shocks emanating from AEs.
- Figure 17 description: the impact is calculated as change in new lending (in percent of assets) for an inter-quartile change in each factor, relative to median new lending.

3) Interconnectedness and crisis performance
- Regression analysis relating output contraction during the 2008 global crisis to fundamentals and network indices found:
  - More diversified countries (higher in-degree) suffered less output contraction than less diversified countries.
  - Countries with more concentrated exposure (higher HHI) suffered a more pronounced output contraction.
- Table 2 (selected coefficients reproduced exactly as presented):
  - In-degree 14.33* (7.66)
  - HHI -11.64* (6.62)
  - Domestic demand growth in AE trading partners 1.12* (0.57) and 1.36** (0.58)
  - Foreign bank claims (percent of GDP, in logs) -2.44*** (0.81) and -1.44 (0.86)
  - GIR in percent of (short-term debt at residual maturity plus current account deficit, in logs) 3.11*** (0.90) and 2.10** (0.97)
  - Observations 40 40
  - R-squared 0.53 0.53
- Interpretation: higher interconnectedness (in-degree) is consistent with risk diversification; higher concentration (HHI) is consistent with shock amplification.

4) Total spillover index based on real variables
- A "real" interconnectedness index for industrial production indices across G7, a group of EMs, China, and the VIX remained broadly stable in the years leading up to the crisis, jumped with the crisis, and has remained elevated since then.
- Network theory explanation: during the "great moderation", small shocks could be dispersed because they did not hit core nodes; increased financial interconnectedness raised system fragility to a systemic shock, producing a spike in measured real spillovers when a large shock occurred.

*Source: IMF staff analysis and figures as presented in the supplied content.*

### 17.      Takeaways. (i) As financial linkages mainly emanate from AEs, shifts in cross-border

### 17. Takeaways.

### Main high-level takeaways
- (i) As financial linkages mainly emanate from AEs, shifts in cross-border exposures are a key channel for the propagation of shocks globally; EMs, in particular, are exposed to such shifts, which can be extremely large, given EMs relatively shallow financial markets.
- (ii) Growing global financial linkages also imply increasing importance of global drivers for EM asset prices movements.
- (iii) Both macro and micro (banking) empirical results show that cross-border financial shocks from AEs and network interconnectedness could have significant macroeconomic impact on EMs.
- (iv) Empirical evidence also suggests that, conditional on a systemic shock occurring, more interconnected countries suffered a smaller output decline during the crisis; however, countries with more concentrated exposure suffered a more pronounced output contraction.

### Motivation for explaining cross-border linkages
- The section explains the strength and drivers of cross-border linkages by relating the size of cross-border financial linkages to structural and economic features of source and recipient countries.
- The intent is to identify determinants of channels through which shocks can be transmitted and to distill principles relevant for the design of a global financial safety net (GFSN).

### Deleveraging exercise (2007–2008)
- A simple exercise related the change in bilateral exposure between 2007 and 2008 to a number of determinants including the size of the initial exposure.
- Key finding: the larger the initial exposure, the larger the subsequent deleveraging.
- EMs experienced larger deleveraging than AEs.

### Methodology and data
- Regression analysis related a country‘s cross-border portfolio and bank liabilities to structural and economic features of source and recipient countries (Annex VII for details).
- Empirical model: a gravity equation of financial linkages between a source and a destination country; both source and destination groups included EM and AE countries.
- Explanatory variables included: distance, time difference, common legal system, common language; size, income per capita, goods trade; financial depth and capital account openness; regional group dummies; year fixed effects.
- Model estimated for the 2009 cross-section and panel 2001-09 using country-pair random effects and source and recipient country fixed effects.

### Deleveraging regression (Table 3) — selected statistics
- Dependent variable: Change in log claims between 2007 and 2008 1,2/
- Sample observations: 1,262; 2,576; 1,212
- R-square: 0.11; 0.17; 0.08
- Selected coefficient entries as reported in table header row:
  - Log of claims in 2007: -0.09 P> |t| 0.02; -0.16 P> |t| 0.01; -0.08 P> |t| 0.01
  - EM recipient dummy: -0.23 P> |t| 0.15; -0.17 P> |t| 0.07; -0.13 P> |t| 0.07
  - EM source dummy: 0.20 P> |t| 0.12; 0.26 P> |t| 0.07; -0.19 P> |t| 0.09
- Notes in table: Other control variables include bilateral exports, financial openness, financial depth, exchange rate depreciation and financial center dummies. Change in log claims are not corrected by valuation change which could be significant in the case of portfolio equity.

### Key empirical findings on determinants of cross-border linkages (Figure 19 and Annex VII)
- Geography and history:
  - Distance between source and recipient is a statistically significant determinant of cross-border financial linkages; higher exposure tends to be built toward nearer countries.
  - Historical and cultural factors (same legal system, common language) are associated with stronger exposure, with particular relevance for equity and bank claims.
- Economic factors:
  - Economic size of both source and recipient matters: larger countries have larger cross-border exposures.
  - More developed countries (overall GDP and per capita GDP) tend to have larger portfolio exposures.
  - Size and income level explain about 40 percent of the interquartile differences across asset classes.
  - Increased trade integration is associated with higher cross-border exposures.
  - Common currency tends to be associated with stronger linkages.
- Financial factors:
  - Increased financial development in both source and recipient countries feeds stronger cross-border financial linkages—strongest for bank claims.
  - Financial center status in source and recipient countries is associated with stronger cross-border exposures; for source countries this is stronger for portfolio debt and bank claims.
  - Recipient countries with more open capital accounts tend to have stronger portfolio exposure across all asset classes; capital account openness effect is not statistically significant for source countries.
- Regional groups:
  - Country group dummies capture differences in exposure levels across groups (Table 4).
  - From the recipient side, G-7 countries remain dominant (statistically significant for portfolio debt and equity, not for bank claims).
  - EMs in general have lower exposures compared to advanced economies.
  - G-7 dominance as source countries is especially strong across asset classes.

### Interconnectedness of bank groups and geographical distance
- Using a bank-group dataset, there has not been a common trend in within-group asset concentration across source countries; some countries (e.g., banks with parents from Spain or Hong Kong SAR) have seen marked increases in asset concentration of their external operations.
- There has been a consistent trend toward a lower average "distance" between subsidiaries and their parent—cross-border groups have tended to move "closer to home," indicating regional agglomeration.
- Figure 20 documents changes in intra-group asset concentration (HHI) and average distance to parents (thousands of kilometers) for selected parent countries (averages of 1996-2000 vs. 2006-09).

### Section takeaways relevant to GFSN design
- (i) Larger and more developed countries tend to have stronger cross-border linkages; financial development in recipient and source economies reinforces this. A GFSN design needs to account for the importance of financially-developed economies as both sources of shocks and potential targets.
- (ii) Geography matters: despite globalization, forces push toward geographical agglomeration; linkages are stronger with closer countries. This could limit the benefit of risk-sharing schemes at a regional level and argues for a global response.
- (iii) Historical factors (shared language, legal system) create "special" linkages—e.g., Spanish banks' large relevance in Latin America.

### Conclusions (section VI)
- Cross-border financial linkages have increased dramatically over time in terms of cross holdings of external assets and co-movements of asset prices across asset classes and economies.
- Linkages are dominated by a few AEs and financial centers that form "core nodes" in the system.
- EMs' cross-border holdings remain relatively small despite rapid growth; the bulk of EM liabilities are held by core nodes. EMs with shallow domestic capital markets are particularly exposed to large shifts in cross-border exposures.
- Shocks can be highly synchronized across global financial markets and their transmission can be non-linear, with potentially significant real costs to EM economies.
- There is a trade-off between benefits of increased international risk diversification (which increases interconnectedness) and increased systemic risk from heightened interconnectedness.
- National defenses, including self-insurance, may not eliminate externalities intrinsic to an interconnected financial network.
- Regional and especially global financing mechanisms have a role to cushion residual volatility when national defenses are insufficient.
- The 2008-09 crisis underscores the value of an effective global mechanism to coordinate liquidity injections and other policy responses: in 2008-09, some "crisis bystanders" with relatively strong fundamentals were hit later and less severely, and recovered more rapidly, yet still suffered deep output contractions.
- Limited opportunities to diversify against cross-border shock risk point to significant gains from a global financial safety net effective in forestalling localized liquidity runs from turning into systemic crises.

*Source: IMF staff analysis (excerpted from sections 17–VI of the supplied chapter).*

### Annex II. Network Analysis

### Annex II. Network Analysis

### Insights from network theory
- Network analysis interprets many financial market phenomena as outcomes of increased interconnectedness in a financial network (see Haldane, 2009; Allen and Gale, 2005; Battiston and others, 2009; Gai and Kapadia, 2008; Garrat and others, 2011; Kubelec and Sa, 2010; von Peter, 2007).
- Interconnectedness is a double-edged sword:
  - At low levels, increased interconnectedness improves resilience and risk sharing (Allen and Gale, 2005).
  - At high levels, interconnectedness can raise latent fragility and vulnerability to systemic breakdowns (negative network externality from core-node shocks).
- Nonlinearities from asymmetric information and herd behavior:
  - Financial integration under incomplete information strengthens incentives for herding (Calvo and Mendoza, 1997).
  - Unusual events may trigger perceptions of "immeasurable risk" and flight to quality (Caballero, 2009).
  - Incomplete information in complex networks creates environments prone to fire sales and liquidity crunches (Caballero and Simsek, 2011).
  - Smaller, ex ante non-systemic countries can become epicenters of systemic events if they trigger creditor reassessments of similarly characterized assets.
- Concentration risk:
  - Recipient countries with unusually large concentration of exposures to a few sources face greater deleveraging risk if a main source is hit.
  - Empirical example: countries with the largest portfolio holdings in emerging Europe at end-2007 experienced the largest portfolio adjustment in 2008 (Galstyan and Lane, 2010).

### Measuring interconnectedness and concentration — definitions and methodology
- Link definition and threshold:
  - A link is represented by the stock of claims issued by a recipient country to a source country.
  - A link is considered active if the stock of the recipient‘s cross-border liabilities is above 0.2 percent of the recipient‘s GDP.
- In-degree:
  - In-degree is the number of locations a country borrows from, counting only active links (threshold 0.2 percent of recipient GDP).
  - In-degree is normalized by the number of sources so it lies between 0 and 1.
  - Country-level in-degree can be aggregated to region-level interconnectedness by averaging.
  - Out-degree is a variant capturing the number of recipients a source country lends to; active links require exposures above 0.2 percent of the source‘s GDP.
- Concentration — Herfindahl-Hirschman Index (HHI):
  - Let N be the number of creditors and si the share of creditor i in country j‘s foreign liabilities; the HHI is the sum of si^2.
  - The HHI is normalized to range between 0 (no concentration) and 1 (only one source per recipient).
  - Country-level HHIs are averaged to obtain regional HHI.
- Data and sample:
  - Cross-border claims studied for 57 emerging markets and 24 advanced economies (listed in Annex I).
  - Networks constructed for period 1999-2009 using BIS and CPIS databases.
  - BIS analysis restricted to the 24 source economies that reported consistently 1999-2009.
  - CPIS data on cross-border portfolio holdings available annually for 2001-2009; analysis restricted to sources that also reported consistently to BIS.
- Data caveats:
  - CPIS and BIS follow different principles to determine the "location" of the source; CPIS tends to overemphasize financial centers relative to BIS consolidated statistics.
  - Commercial bank holdings of portfolio securities may cause overlap between datasets.
  - Neither dataset is adjusted for valuation effects.
  - BIS locational statistics would avoid the financial-center residence issue but are not publicly available.

### Empirical findings and stylized facts (1999–2009)
- General trends:
  - Interconnectedness has generally increased in the last decade across EMs and AEs and across asset classes; this trend was partially reversed during the 2008 deleveraging but may have resumed in 2009.
  - Concentration has generally declined over the past decade across EMs and AEs and across asset classes, with notable exceptions.
- Cross-region and asset-class patterns:
  - Interconnectedness is much stronger for AEs than EMs (figures use different scales).
  - Smaller AEs tend to be more interconnected.
  - For EMs, interconnectedness is higher for cross-border bank claims than for portfolio claims.
  - No clear pattern across asset classes for AEs.
  - Concentration across asset classes is still significantly higher in EMs than in AEs.
  - Notable exception: cross-border bank claims in European EMs saw concentration increase rapidly in the years before the crisis.
- Directionality of linkages:
  - Out-degree analysis shows AEs‘ outward linkages are almost as spread out as inward linkages.
  - For EMs, outward linkages are fairly limited — interconnectedness in EM economies is generally "one way", affecting mostly their cross-border liability side.
  - Limited data: out-degree index for EMs summarized for five EM countries reporting consistently (Brazil, Mexico, Chile, Panama, and Turkey).

### Key quantitative thresholds, periods, and sample counts preserved from the source
- Link active threshold: 0.2 percent of recipient‘s (or source‘s) GDP.
- Networks constructed for the period 1999-2009.
- CPIS data availability used: 2001-2009.
- Sample size: 57 emerging markets and 24 advanced economies.
- BIS analysis restricted to 24 source economies reporting consistently 1999-2009.

*Source: Annex II. Network Analysis, prepared by Ran Bi and Sergi Lanau.*

### Annex IV. Importance of Global Factors in Driving Capital Inflows to EMs

### Annex IV. Importance of Global Factors in Driving Capital Inflows to EMs

### Objective and approach
- Investigates the quantitative importance of global factors and country-specific fundamentals in driving cross-border capital inflows to EMs.
- Uses the same panel framework as in IMF, 2011e, including both global and domestic variables as determinants of capital inflows.
- Addresses potential bias from large cross-country variation by decomposing total variance of underlying flows (in log levels) into two components: variance across countries and variance over time. The contribution of global factors is computed as the ratio of their variance to the variance of gross capital inflows over time.

### Global factors considered
- U.S. long term interest rates.
- VIX risk index.

### Main quantitative findings
- Global factors explained on average around 25 percent of the cross-time variation of gross total capital inflows to EMs.
- For portfolio inflows, the estimated share explained by global factors rose to 54 percent.
- When domestic variables were omitted from regressions (treating domestic variables as potentially endogenous to global factors), the share of cross-time variation explained by global factors rose to:
  - 65 percent for total inflows.
  - 87 percent for portfolio inflows.
  - These omission-based estimates are interpreted as an upper bound for the role of global factors.

### Country-level robustness (seven-country panel results)
- Separate regressions for seven countries confirmed similar magnitudes:
  - Average share of global factors in explaining total inflows: about one fifth (≈20 percent).
  - Average share of global factors in explaining portfolio inflows: 48 percent.
- Table of shares (percent):
  - Brazil: Aggregate Inflows 30; Portfolio Inflows 59
  - Indonesia: Aggregate Inflows 18; Portfolio Inflows 88
  - Korea: Aggregate Inflows 17; Portfolio Inflows 37
  - Peru: Aggregate Inflows 5; Portfolio Inflows 17
  - South Africa: Aggregate Inflows 22; Portfolio Inflows 5
  - Thailand: Aggregate Inflows 20; Portfolio Inflows 88
  - Turkey: Aggregate Inflows 11; Portfolio Inflows 41

### Alternative variance-decomposition methodology (IMF, 2011b approach)
- Three-step procedure applied to sample of 48 EMs for both gross and net flows (normalized as a percent of GDP):
  - Step 1: For each period t, decompose flows Y(i,t)=a(t) + error(i,t), where a(t) is the cross-section average of flows (global factor component) and error(i,t) is the country-specific component. Compute Residual Sum of Squares (RSS) for each period.
  - Step 2: Derive Total Sum of Squares (TSS) for each period as the cross-country sum of squared differences of flows from their sample mean.
  - Step 3: Compute R-squared as 1-(RSS/TSS) for each period; this yields the share of variance attributable to global factors.
- Results (presented in Figures IV.1 and IV.2):
  - For net inflows: global factors explained less than 20 percent of total capital inflows.
  - For total gross inflows: global factors explained less than a third.
- The exercise highlights that the relative importance of global factors shifts over time and across different types of flows.

### Interpretation and caveats
- Conservative estimates indicate global factors could explain at least 25 percent of variations in total gross inflows; for portfolio inflows, global factors are likely much more important (above 50 percent).
- The relative importance of global factors is time-varying and reflects global financial and liquidity conditions.
- A simple outright variance decomposition from panel regressions yields that the variance of global factors accounts for a mere 5 percent of total sample variation; this estimate likely has a significant downward bias because cross-country variation driven by idiosyncratic country factors dominates total variance.
- When domestic variables are omitted, global-factor shares increase substantially, indicating domestic variables are significantly affected by global factors; omitted-variable estimates can be interpreted as upper bounds.

*Source: IMF staff calculations and analysis in "Annex IV. Importance of Global Factors in Driving Capital Inflows to EMs".*

### Annex VII. Explaining Bilateral Financial Exposures

### Annex VII. Explaining Bilateral Financial Exposures

### Literature summary
- Portes and Rey, 2005: proxies for informational asymmetries and host country stock market size are key determinants of international equity flows.
- Rose and Spiegel, 2004: positive association between bilateral trade and bilateral bank lending.
- Lane and Milesi-Ferretti, 2008: strongly positive association between bilateral trade and portfolio equity allocations; bilateral portfolio equity holdings correlated with informational/cultural linkages (common language, legal origins).
- Galstyan and Lane, 2010: systemic positive relation between the scale of pre-crisis equity holdings and the subsequent pull back during the global crisis.
- Contribution of this annex: revisit prior studies applying a common approach to different asset classes (bank lending and portfolio debt and equity claims).

### Empirical strategy (gravity framework)
- Estimated equation:
  - ln(y_ijt) = α + β1 ln(size_it) + β2 ln(size_jt) + β3 ln(distance_ij) + w_it + x_jt + z_ijt + ε_ijt
  - where i = source, j = recipient, t = year (2001 to 2009); y_ijt = country i‘s cross-border holdings of country j‘s financial assets at time t; w_it = source country-specific variables; x_jt = recipient country-specific variables; z_ijt = bilateral explanatory variables.
- Dependent variable specification: ln(1+x) to account for many zero observations (especially in CPIS).
- Assets analyzed: portfolio debt securities, portfolio equity securities, and bank claims.
- Primary estimation: country-pair random effects (Breusch-Pagan Lagrange Multiplier test suggested superiority over pooled OLS).
- Robustness: fixed effects for source and recipient countries were also used; main variables of interest were robust to this specification change.

### Data and explanatory variables
- Data sources:
  - Portfolio equity and portfolio debt: CPIS (1997-2009).
  - Cross-border bank claims: BIS banks‘ foreign claims (Table 9a).
  - Bilateral trade: IMF Direction of Trade Statistics.
  - Distance, common currency, common language, common legal origin: CEPII dataset.
  - GDP and GDP per capita: IMF/WEO.
  - Financial openness: Chinn-Ito index (2008 update; 2009 dummy uses 2008).
  - Financial depth indicators: World Bank Financial Structure Database (stock market capitalization to GDP for equity regression; outstanding bond to GDP for debt regression; private sector credit to GDP for bank claims regression).
  - Financial center dummy: international financial center indicator.
- Sample coverage: 81 countries (27 advanced economies and 54 emerging markets); actual sample size varied with data availability for each regression.

### Baseline regression results (country-pair random effects), 2001-09
- Estimation reported for three dependent variables: Equity securities, Debt securities, BIS bank claims.
- Coefficients (robust standard errors in parentheses). Significance: *** p<0.01, ** p<0.05, * p<0.1.

- Log GDP, recipient:
  - Equity: 0.572*** (0.024)
  - Debt: 0.604*** (0.052)
  - BIS bank claims: 0.727*** (0.035)

- Log GDP, source:
  - Equity: 0.122*** (0.027)
  - Debt: 0.532*** (0.065)
  - BIS bank claims: 0.372*** (0.051)

- Log GDP per capita, recipient:
  - Equity: 0.0764 (0.050)
  - Debt: 0.227*** (0.081)
  - BIS bank claims: 0.00349 (0.062)

- Log GDP per capita, source:
  - Equity: 0.430*** (0.051)
  - Debt: 0.478*** (0.087)
  - BIS bank claims: -0.18 (0.123)

- Log exports from recipient to source:
  - Equity: 0.0784*** (0.008)
  - Debt: 0.0364*** (0.013)
  - BIS bank claims: 0.0613*** (0.011)

- Financial openness, recipient:
  - Equity: 0.0612*** (0.021)
  - Debt: 0.109*** (0.042)
  - BIS bank claims: 0.0744*** (0.028)

- Financial openness, source:
  - Equity: -0.0320* (0.017)
  - Debt: 0.101** (0.046)
  - BIS bank claims: -0.035 (0.050)

- Financial depth indicator, recipient:
  - Equity: 0.243*** (0.036)
  - Debt: 0.235*** (0.055)
  - BIS bank claims: 0.696*** (0.075)

- Financial depth indicator, source:
  - Equity: 0.171*** (0.028)
  - Debt: 0.126** (0.053)
  - BIS bank claims: 0.135* (0.081)

- Time difference:
  - Equity: 0.0582*** (0.014)
  - Debt: -0.0307 (0.023)
  - BIS bank claims: 0.00122 (0.021)

- Log distance:
  - Equity: -0.597*** (0.056)
  - Debt: -0.702*** (0.088)
  - BIS bank claims: -0.763*** (0.081)

- Common legal system:
  - Equity: 0.179*** (0.069)
  - Debt: 0.173* (0.101)
  - BIS bank claims: -0.0151 (0.096)

- Common language:
  - Equity: 0.622*** (0.119)
  - Debt: 0.420*** (0.161)
  - BIS bank claims: 0.879*** (0.130)

- Common currency:
  - Equity: 1.618*** (0.161)
  - Debt: 1.542*** (0.136)
  - BIS bank claims: 0.226 (0.139)

- Financial center dummy, destination:
  - Equity: 0.764*** (0.107)
  - Debt: 0.0669 (0.120)
  - BIS bank claims: 0.621*** (0.129)

- Financial center dummy, source:
  - Equity: 0.483*** (0.097)
  - Debt: 1.460*** (0.116)
  - BIS bank claims: 1.136*** (0.097)

- G7 dummy, recipient:
  - Equity: 0.573*** (0.134)
  - Debt: 0.689*** (0.170)
  - BIS bank claims: 0.218 (0.143)

- EMEUR dummy, recipient:
  - Equity: -0.493*** (0.118)
  - Debt: -0.606*** (0.235)
  - BIS bank claims: -0.461*** (0.156)

- Middle east EM dummy, recipient:
  - Equity: -0.444*** (0.153)
  - Debt: 0.0784 (0.314)
  - BIS bank claims: -0.967*** (0.180)

- Asian EM dummy, recipient:
  - Equity: 0.0876 (0.169)
  - Debt: -0.611** (0.256)
  - BIS bank claims: -0.115 (0.158)

- Latin American EM, recipient:
  - Equity: -0.385*** (0.130)
  - Debt: -0.132 (0.200)
  - BIS bank claims: 0.316** (0.142)

- G7 dummy, source:
  - Equity: 1.532*** (0.119)
  - Debt: 1.452*** (0.180)
  - BIS bank claims: 1.259*** (0.130)

- EMEUR dummy, source:
  - Equity: -1.362*** (0.113)
  - Debt: -2.199*** (0.247)
  - BIS bank claims: -1.485*** (0.262)

- Middle east EM dummy, source:
  - Equity: -0.634*** (0.146)
  - Debt: -1.180*** (0.360)
  - BIS bank claims: (no estimate reported)

- Asian EM dummy, source:
  - Equity: -0.898*** (0.142)
  - Debt: -0.416* (0.244)
  - BIS bank claims: (no estimate reported)

- Latin American EM, source:
  - Equity: -0.921*** (0.124)
  - Debt: -1.282*** (0.225)
  - BIS bank claims: -2.234*** (0.230)

- Constant:
  - Equity: -1.798** (0.786)
  - Debt: -4.068*** (1.412)
  - BIS bank claims: 6.393*** (1.383)

- Observations:
  - Equity: 22,631
  - Debt: 10,711
  - BIS bank claims: 10,498

- Number of country pairs:
  - Equity: 3,511
  - Debt: 1,396
  - BIS bank claims: 1,504

- R2:
  - Equity: 0.664
  - Debt: 0.715
  - BIS bank claims: 0.695

- Source note: BIS, CPIS and staff estimates. Robust standard errors in parentheses.

### Interpretation of key findings
- Economic size matters: recipient Log GDP coefficients are positive and highly significant across equity (0.572***), debt (0.604***), and bank claims (0.727***).
- Distance is a strong negative determinant across all asset classes: Log distance coefficients are -0.597*** (equity), -0.702*** (debt), -0.763*** (bank claims).
- Cultural and institutional linkages increase holdings: Common language (0.622*** equity; 0.420*** debt; 0.879*** bank claims) and common currency (1.618*** equity; 1.542*** debt) show large positive associations.
- Financial depth of recipient is strongly positive, especially for bank claims (0.696***).
- Financial centers as sources and destinations significantly boost cross-border holdings, notably source financial center for debt (1.460***) and bank claims (1.136***).
- Regional/emerging-market dummies reveal heterogeneous effects: EMEUR and some EM regional dummies are associated with lower bilateral holdings in many specifications.

### Robustness and specification notes
- Pooled OLS considered but country-pair random effects preferred by Breusch-Pagan Lagrange Multiplier test.
- Country-pair fixed effects not considered because that would require dropping time-invariant country-pair variables of interest.
- Robustness checks using source and recipient fixed effects (dropping time-invariant source and recipient variables) produced consistent results for main variables of interest.

### Data description highlights (Table VII.2)
- Bilateral portfolio equity holdings: Portfolio equity instruments issued by recipient country residents and held by source country residents, 1997-2009 CPIS.
- Bilateral portfolio debt holdings: Portfolio debt instruments issued by recipient country residents and held by source country residents, 1997-2009 CPIS.
- Bilateral foreign bank claims: BIS banks‘ foreign claims (cross-border + local claims) on a country.
- Distance: simple distance in km between the most populated cities (CEPII).
- Common currency, language, legal origin: CEPII dataset; common currency data up to 2006, 2006 values used for 2007-2009.
- Financial openness: Chinn-Ito index (Chinn-Ito 2008 Update); dummy for 2009 from 2008.
- Financial depth indicators: Stock market capitalization to GDP (equity), outstanding bond to GDP (debt), private sector credit to GDP (bank claims) from World Bank Financial Structure Database.

*Source: Annex VII, "Explaining Bilateral Financial Exposures" (prepared by Mali Chivakul). Continuous references and data sources as listed in the annex.*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/np/pp/eng/2011/_060111.pdf_