## 1. External Borrowing and Lending: Korea, Australia, and Hungary

## Source details

**Canonical URL:** [1. External Borrowing and Lending: Korea, Australia, and Hungary](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2012/_wp12268.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2012/_wp12268.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2012/_wp12268.pdf.json)

---

### Introduction
- Two well-known sources of external vulnerabilities in Korea’s financial system:
  - Capital markets that are large and open and sensitive to global risk factors, leading to high price volatility during bouts of financial stress.
  - Sharp increases in short-term external debt by Korean banks in the run-up to the Global Financial Crisis (GFC), exposing them to rollover risks.
- Interaction studied: how these two vulnerabilities interacted during the GFC and post‑Lehman risk aversion episodes, and whether the interaction has weakened over time.
- Cyclical and structural changes since the GFC:
  - Exporters’ demand for medium- to long-term foreign exchange hedging has declined.
  - Offshore entities have increasingly replaced foreign bank branches as the investor base of local currency bonds.
  - International reserves have substantially increased relative to short-term debt.
- Key finding: Sensitivity of exchange rate volatility to changes in VIX has declined very significantly since the GFC, mainly reflecting lower FX liquidity mismatches.

### Why were Korean banks hit so hard during the GFC?
- Stylized facts explaining Korea’s large crisis impact:
  - Fact 1 — Weak link between banks’ external borrowing and domestic lending:
    - Korea was an outlier: despite being a current account surplus country, banks accumulated large external debt.
    - Pre-GFC dynamics in Korea: a threefold pre-GFC increase in external borrowing paired with only a one half increase in overall domestic lending.
    - A one dollar increase in external borrowing prior to the GFC was accompanied by only a 20 cent increase in foreign currency lending.
  - Fact 2 — Banks’ external borrowing mainly driven by demand for FX hedging:
    - Expectations of trend appreciation in exchange rates created large hedging demand by exporters and asset managers expanding overseas securities investment.
    - Shipbuilders were a prominent example, driven by rapidly expanding order books.
    - Selected entries from Table 1 (Companies’ FX Forwards Selling and Buying, in billions of U.S. dollars):
      - FX Forwards sold (A): 71.7 (2005), 99.7 (2006 1st), 126.0 (2006 2nd), 136.6 (2007), 92.8 (2008), 43.8 (2009 1st), 70.9 (2009 2nd), 103.2 (2010), 112.5 (2011)
      - Shipbuilders: 22.3 (2005), 43.1 (2006 1st), 62.3 (2006 2nd), 40.9 (2007), 33.5 (2008), 7.4 (2009 1st), 15.7 (2009 2nd), 21.9 (2010), 27.7 (2011)
      - (Orders): 35.2 (2005), 54.6 (2006 1st), 97.4 (2006 2nd), 69.8 (2007), 51.5 (2008), 18.4 (2009 1st), 14.4 (2009 2nd), 34.6 (2010), 49.3 (2011)
      - FX Forwards bought (B): 42.5 (2005), 50.4 (2006 1st), 54.2 (2006 2nd), 74.6 (2007), 34.4 (2008), 40.2 (2009 1st), 49.8 (2009 2nd), 70.3 (2010), 98.2 (2011)
      - Oil refineries: 6.1 (2005), 1.0 (2006 1st), 2.5 (2006 2nd), 3.5 (2007), 1.8 (2008), 1.7 (2009 1st), 3.0 (2009 2nd), 4.3 (2010), 12.8 (2011)
      - (Oil imports): 42.5 (2005), 55.8 (2006 1st), 60.5 (2006 2nd), 85.0 (2007), 44.4 (2008), 40.6 (2009 1st), 50.7 (2009 2nd), 68.6 (2010), 91.1 (2011)
      - Net selling (=A-B): 29.2 (2005), 49.3 (2006 1st), 71.8 (2006 2nd), 62.0 (2007), 58.3 (2008), 3.7 (2009 1st), 21.1 (2009 2nd), 33.0 (2010), 14.4 (2011)
  - Resulting build-up in short-term external debt:
    - Aggregate short-term external debt owed by Korean banks rose to US$160 billion in the third quarter of 2008 from US$60 billion in the first quarter of 2006.
    - Bank of Korea internal data: increase in banks’ net forward positions (after netting out portions backed by swaps) can explain up to 60 percent of the increase in short-term external borrowing during this period.
    - BOK and Financial Supervisory Services joint thematic examination (January 2008): banks tend to square about three quarters of their forward overbought positions by external borrowing and 20 percent by swap transactions.
- Empirical model: a panel data model of short-term external debt estimated using GMM (sample period 2004Q4-2011Q4; 15 domestic banks and 27 foreign bank branches included).

### Determinants of Korean Banks’ Short-Term External Debt (empirical findings)
- Dependent variable: ST external debt/ total asset.
- Key significant coefficients (foreign bank branches):
  - ST external debt(-1)/ total asset(-1): 0.4424*** (Robust St.err 0.0960)
  - FX derivatives positions /total asset: 0.3186*** (Robust St.err 0.0558)
  - BIS capital adequacy ratio: -0.0740** (Robust St.err 0.0352)
  - Inflation(-1): -3.9555*** (Robust St.err 0.7462)
  - △Log(exr)(-1): -0.1272* (Robust St.err 0.0750)
  - Exr volatility(-1): 0.1242*** (Robust St.err 0.0260)
  - CA(-1)/GDP(-1): -0.5967** (Robust St.err 0.2313)
  - Reserve(-1)/GDP(-1): 0.3312** (Robust St.err 0.1326)
  - Incentive: 2.0640*** (Robust St.err 0.6774)
  - VIX: -0.3836*** (Robust St.err 0.0988)
- Key significant coefficients (domestic banks):
  - ST external debt(-1)/ total asset(-1): 0.6962*** (Robust St.err 0.0611)
  - FX derivatives positions /total asset: 0.1110 (Robust St.err 0.0704)
  - BIS capital adequacy ratio: -0.0431** (Robust St.err 0.0208)
  - Inflation(-1): -0.1493 (Robust St.err 0.1077)
  - △Log(exr)(-1): 0.0087 (Robust St.err 0.0096)
  - Exr volatility(-1): -0.0023 (Robust St.err 0.0023)
  - CA(-1)/GDP(-1): -0.0395 (Robust St.err 0.0268)
  - Reserve(-1)/GDP(-1): -0.0824 (Robust St.err 0.0977)
  - Incentive: 0.0004 (Robust St.err 0.0910)
  - VIX: 0.0087 (Robust St.err 0.0171)
- Model diagnostics:
  - AR (1) for foreign bank branches: -2.5596 <p-value> <0.0105>
  - AR (2) for foreign bank branches: 0.6172 <p-value> <0.5371>
  - AR (1) for domestic banks: -2.0906 <p-value> <0.0366>
  - AR (2) for domestic banks: 1.3315 <p-value> <0.1830>
- Note: ***,**, * denote significance in 1, 5, and 10 percent range.

### Bank-specific positioning and mechanics of short positions
- Two primary ways banks squared forward overbought positions:
  - Borrowing foreign currency, selling it into Korean won, and investing the proceeds in domestic bonds (typically foreign bank branches).
  - Creating a short FX position by combining cross-currency (or FX) swap and a spot selling of foreign currency (typically domestic banks).
- Foreign bank branches’ typical trade flow:
  - Buy FX forward from client; borrow FX from headquarter; convert into KRW in spot market; use KRW to buy Monetary Stabilization Bonds (MSBs) or Korea Treasury Bonds (KTBs).
  - Generated risk-free profits from gaps between interest differentials and swap rates but created FX liquidity mismatches because forwards often had longer maturities (particularly with shipbuilders) than their borrowing.
- Foreign bank branches used short-term external debt to offset longer-term exposures in FX forwards because of cheaper interest cost of funding and low perceived liquidity risk amid abundant global liquidity pre-GFC.

### Evidence of liquidity mismatches and vulnerabilities
- Fact 3 — Foreign bank branches built up liquidity mismatches:
  - FX derivative composition (Q3 2008): FX forwards (a bulk exceeding 1 year) and FX swaps (1 to 3 months) on the long side; cross-currency swaps (1 to 3 years) on the short side.
  - Branches raised short-term external debt to offset net open positions related to:
    - Hedging services to nonbank clients (gap between net long forward and net short CRS; ~one half of net open positions).
    - Intermediary roles in FX swap market (net long FX swap position; pairs with domestic banks’ net short FX swap position; ~remaining half).
  - Main source of liquidity mismatches: hedging operations (net long forward positions), not short-term FX swap positions.
- Vulnerability dynamic:
  - Global interbank dollar funding freeze (e.g., Lehman) realized tail risk, causing retrenchment and drying up of FX funding that exposed liquidity mismatches.

### Linkages between short-term external debt, FX funding, and exchange rate volatility
- Mechanism during the GFC:
  - Sudden stop in capital flows hit foreign bank branches; rollover difficulties on short-term external debt forced conversion of KRW liquidity into FX.
  - Cost of onshore dollar funding rose sharply as foreign bank branches retrenched and offshore arbitrageurs curtailed CRS funding; market liquidity in dollar funding dissipated.
  - Foreign bank branches liquidated bond positions to obtain FX, generating downward pressure on the won.
  - Depreciation could stoke demand for dollar swap funding as market participants expected exchange-rate overshoot.
- Empirical association: Positive association between exchange rate volatility and short-term external debt.

### Cross-country comparison: Korea, Australia, and Hungary
- Overall lending and external borrowing patterns:
  - Hungary: Bank external borrowing closely tracked total lending; causality appears to run from easier access to foreign finance to increased domestic credit.
  - Australia: Increased reliance on wholesale funding and capital inflows (carry-trade motives) underpinned a sharp pre-GFC run-up in external borrowing and domestic lending; ability to raise external borrowing in local currency or swap foreign currency debt into local currency weakened link between external borrowing and foreign currency lending.
  - Korea: Divergent pattern — a threefold run up in external borrowing in the 30 months leading to the GFC was associated with only a 50 percent increase in overall lending. The GFC reduced a quarter of the external debt stock, yet overall lending continued to grow on a moderated path.
- Foreign currency lending patterns:
  - Hungary: close tie between external borrowing and currency mismatches.
  - Australia and Korea: weak links between external borrowing and foreign currency lending.

### Exchange rate volatility and regression evidence
- May 2012 observation: Korean banks were "riding out global shocks much better than they did three or four years ago."
- CDS spreads for major Korean banks widened to only 150 basis points, about half the peak reached during the third quarter of 2011, and compared with 700–800 basis points seen during the GFC.
- Regression sample: Jan. 2006 to Dec. 2011. Dependent: Won-Dollar Exchange Rate Volatility.
- Key regression coefficients (coefficient (Newey-West standard error)):
  - Banks’ short-term external debt/GDP: 0.015** (0.006)
  - Yield spread(US) (3Yr – FFR): 0.188** (0.082)
  - Trade growth(-1): -3.457** (1.695)
  - Inflation(CPI): -0.032 (0.059)
  - GDP growth(-1): -0.025* (0.014)
  - Foreigners’ stock buying(net)(-1): 0.000 (0.000)
  - D(Foreigners’ share in bond markets): -0.415* (0.240)
  - D(FBBs’ share in bond markets): -0.089 (0.117)
  - D(VIX): -0.062** (0.025)
  - D(VIX)*Short-term external debt/GDP: 0.002*** (0.000)
  - Constant: 0.083 (0.300)
  - R-squared: 0.787
  - Adjusted R-squared: 0.752
- Notable regression implications:
  - Sensitivity of exchange rate volatility to changes in VIX is a function of the external debt-to-GDP ratio: peaked at 0.08 (October 2008), fell to 0.03 (April 2010), and further declined to 0.01 (prior to euro deleveraging in Q3 2011).
  - Counterfactual simulations: VIX shocks of GFC magnitude replicated in later episodes would have produced peak exchange rate volatility that was 50 percent less (first Greek crisis) and 70 percent less (euro deleveraging) than during the GFC peak.
  - Negative association between foreigners’ share in the Korean bond market and exchange rate volatility suggests investor-base diversification reduces herding risk.

### CIP deviations (Annex summary)
- CIP deviations derived from 3-month FX and 1-year cross-currency swap markets; regressors include changes in VIX, TED, CDS, net forwards sold by companies, and net bond investment by foreign investors and foreign bank branches (lagged).
- Sample periods and main findings:
  - Pre-crisis (Jan. 2003–Jul. 2007): Sovereign CDS spread and net forward positions sold by companies positively and significantly associated with CIP deviations; global risk factors not significant.
  - Crisis (Aug. 2007–Jun. 2009): VIX significantly positive for CIP deviations; net bond investment by foreign investors significantly negative (sell-offs exacerbated global risk aversion).
  - Post-crisis (Jul. 2009–Dec. 2011): Country risk, hedging demand, and global risk factors lost significance; net bond investments by foreign investors and foreign bank branches significantly affected three-month CIP deviation; rise in foreign investor share (vs. foreign bank branches) brought stability.
- Reported Adjusted R-squared ranges:
  - Table A-1 (three-month CIP deviation): 0.443, 0.412, 0.379, 0.367, 0.364, 0.421 (across columns).
  - Table A-2 (one-year CIP deviation): 0.515, 0.506, 0.582, 0.587, 0.451, 0.383 (across columns).

### Changes since the GFC that reduced vulnerabilities
- Decline in demand for medium- to long-term FX hedging by shipbuilders due to a sharp global decline in ship orders and changed exchange rate expectations.
- Macroprudential measures introduced:
  - Ceilings on banks’ foreign exchange derivative positions.
  - Tighter foreign exchange liquidity standards and more frequent stress tests.
  - A macroprudential levy on foreign currency-denominated noncore bank liabilities.
  - Reinstated withholding tax on foreign investors’ interest income on government bonds.
- Shift in investor base for domestic bonds:
  - Offshore entities, including real money and sovereign investors, increasingly replaced foreign bank branches as investors in domestic bonds.
  - Increase in share of central banks in foreign investor holdings of domestic bonds to 35 percent by end-2011.
- Improved reserve coverage: reserve-to-short-term-debt cover significantly improved.
- Evidence of weakened feedback loop: May 2012 episode — one year cross-currency swap spread widened to minus 150 basis points, compared to minus 300 basis points at its trough during the third quarter of (period referenced).

### Policy implications and recommendations
- Main findings:
  - Increased resilience of the Korean financial system since the GFC, particularly due to smaller short-term external debt weakening the link between exchange rate volatility and global risk factors.
  - Excessive exchange rate volatility during the GFC resulted from massive buildup of short-term external debt driven by expectations of trend appreciation of the Korean won.
- Macroeconomic policy trade-offs:
  - Surges in capital inflows can cause currency overshooting, hurting international competitiveness.
  - Central bank intervention to prevent overshooting complicates monetary policy and can build downside tail risks to financial stability; in Korea, tail risks manifested as excessive short-term debt accumulation.
- Macroprudential stance:
  - Measures aimed at curbing short-term external borrowing have helped mitigate FX funding risk and can act as speed bumps moderating capital flow cycles.
  - Macroprudential measures have efficiency costs and risks of overkill; optimization needed focusing on strengthening institutional framework, minimizing risk of overkill, and closing loopholes.
- Ongoing policy actions recommended:
  - Continuous surveillance of financial markets.
  - Close monitoring and analysis of banks’ and corporate sector balance sheets.
  - Frequent stress tests to ensure robustness to shocks.

### Key statistics (preserved exactly)
- Aggregate short-term external debt owed by Korean banks: US$60 billion in the first quarter of 2006; US$160 billion in the third quarter of 2008.
- Bank of Korea internal data: increase in banks’ net forward positions (after netting out portions backed by swaps) can explain up to 60 percent of the increase in short-term external borrowing during this period.
- BOK/FSS thematic examination (January 2008): banks square about three quarters of their forward overbought positions by external borrowing and 20 percent by swap transactions.
- Panel estimation sample period: 2004Q4-2011Q4; 15 domestic banks and 27 foreign bank branches included.
- CDS spreads for major Korean banks in May 2012: widened to only 150 basis points; compared with 700–800 basis points seen during the GFC.
- Cross-currency swap spreads: minus 150 basis points (May 2012 episode) vs. minus 300 basis points at trough during the third quarter of (period referenced).
- Share of central banks in foreign investor holdings of domestic bonds: 35 percent by end-2011.
- Regression R-squared: 0.787; Adjusted R-squared: 0.752.

*Extracted from IMF source document: _wp12268 - 1. External Borrowing and Lending: Korea, Australia, and Hungary .................................6*

### 1. External Borrowing and Lending: Korea, Australia, and Hungary  .................................6

### 1. External Borrowing and Lending: Korea, Australia, and Hungary  .................................6

### Introduction
- Two well-known sources of external vulnerabilities in Korea’s financial system:
  - Capital markets that are large and open and sensitive to global risk factors, leading to high price volatility during bouts of financial stress.
  - Sharp increases in short-term external debt by Korean banks in the run-up to the Global Financial Crisis (GFC), exposing them to rollover risks.
- Interaction studied: how these two vulnerabilities interacted during the GFC and post‑Lehman risk aversion episodes, and whether the interaction has weakened over time.
- Cyclical and structural changes since the GFC noted in the source:
  - Exporters’ demand for medium- to long-term foreign exchange hedging has declined.
  - Offshore entities have increasingly replaced foreign bank branches as the investor base of local currency bonds.
  - International reserves have substantially increased relative to short-term debt.
- Key finding summarized in the source:
  - Sensitivity of exchange rate volatility to changes in VIX has declined very significantly since the GFC, mainly reflecting lower FX liquidity mismatches.
- Paper organization (as per source): causes and dynamics of short-term external debt (Section II), exchange rate volatility (Section III), changes since the GFC (Section IV), regressions and simulations (Section V), and policy implications (conclusion).

### Why were Korean banks hit so hard during the GFC?
- Context: Lehman Brothers bankruptcy in September 2008 caused a sudden stop in capital flows; many capital-flow recipient countries lost funding to cope with current account deficits. Korea, despite a prior current account surplus, was among the hardest hit in Asia.

Key stylized facts from the source explaining Korea’s large crisis impact:
- Fact 1: Weak link between banks’ external borrowing and domestic lending
  - Typical macro link: banks’ external borrowing often tied to domestic credit demand and current account deficits. Larger current account deficits generally increase need for banks’ external borrowing.
  - Korea was an outlier: despite being a current account surplus country, banks accumulated large external debt.
  - Cross-country comparison (referenced Figure 1): Korea, Russia, and Ukraine built up large amounts of external bank debt (above 10 percent of GDP) while running current account surpluses since early 2000s.
  - Pre-GFC dynamics in Korea:
    - Threefold pre-GFC increase in external borrowing paired with only a one half increase in overall domestic lending.
    - A one dollar increase in external borrowing prior to the GFC was accompanied by only a 20 cent increase in foreign currency lending (specific phrasing preserved).
- Fact 2: Banks’ external borrowing mainly driven by demand for FX hedging
  - Expectations of trend appreciation in exchange rates created large hedging demand by exporters and asset managers expanding overseas securities investment.
  - Shipbuilders were a prominent example, driven by rapidly expanding order books.
  - Expectations of currency appreciation, lengthy production cycles, and lower import contributions led firms to raise hedge ratios.
  - Table 1 (Companies’ FX Forwards Selling and Buying, in billions of U.S. dollars) — selected figures preserved exactly:
    - FX Forwards sold (A): 71.7 (2005), 99.7 (2006 1st), 126.0 (2006 2nd), 136.6 (2007), 92.8 (2008), 43.8 (2009 1st), 70.9 (2009 2nd), 103.2 (2010), 112.5 (2011)
    - Shipbuilders: 22.3 (2005), 43.1 (2006 1st), 62.3 (2006 2nd), 40.9 (2007), 33.5 (2008), 7.4 (2009 1st), 15.7 (2009 2nd), 21.9 (2010), 27.7 (2011)
    - (Orders): 35.2 (2005), 54.6 (2006 1st), 97.4 (2006 2nd), 69.8 (2007), 51.5 (2008), 18.4 (2009 1st), 14.4 (2009 2nd), 34.6 (2010), 49.3 (2011)
    - FX Forwards bought (B): 42.5 (2005), 50.4 (2006 1st), 54.2 (2006 2nd), 74.6 (2007), 34.4 (2008), 40.2 (2009 1st), 49.8 (2009 2nd), 70.3 (2010), 98.2 (2011)
    - Oil refineries: 6.1 (2005), 1.0 (2006 1st), 2.5 (2006 2nd), 3.5 (2007), 1.8 (2008), 1.7 (2009 1st), 3.0 (2009 2nd), 4.3 (2010), 12.8 (2011)
    - (Oil imports): 42.5 (2005), 55.8 (2006 1st), 60.5 (2006 2nd), 85.0 (2007), 44.4 (2008), 40.6 (2009 1st), 50.7 (2009 2nd), 68.6 (2010), 91.1 (2011)
    - Net selling (=A-B): 29.2 (2005), 49.3 (2006 1st), 71.8 (2006 2nd), 62.0 (2007), 58.3 (2008), 3.7 (2009 1st), 21.1 (2009 2nd), 33.0 (2010), 14.4 (2011)
  - Resulting build-up in short-term external debt:
    - Aggregate short-term external debt owed by Korean banks rose to US$160 billion in the third quarter of 2008 from US$60 billion in the first quarter of 2006.
    - Bank of Korea internal data: increase in banks’ net forward positions (after netting out portions backed by swaps) can explain up to 60 percent of the increase in short-term external borrowing during this period.
    - BOK and Financial Supervisory Services joint thematic examination (January 2008) found banks tend to square about three quarters of their forward overbought positions by external borrowing and 20 percent by swap transactions.
- Empirical analysis mentioned:
  - A panel data model of short-term external debt was estimated using GMM (sample period 2004Q4-2011Q4, 15 domestic banks and 27 foreign bank branches included; see Tables 2, 3 in source).
  - The source reports: "The estimation shows that a one percentage point increase in net forward position (scaled by the total asset size) of a foreign bank branch is associated with a" — (estimation result statement is truncated in the supplied content).

### Cross-country comparison: Korea, Australia, and Hungary (Box 1 summary)
- Overall lending and external borrowing patterns:
  - Hungary: Bank external borrowing closely tracked total lending; causality appears to run from easier access to foreign finance to increased domestic credit.
  - Australia: Increased reliance on wholesale funding and capital inflows (carry-trade motives) underpinned a sharp pre-GFC run-up in external borrowing and domestic lending. Australia can raise external borrowing in local currency or swap foreign currency debt into local currency, weakening the link between external borrowing and foreign currency lending.
  - Korea: Divergent pattern — a threefold run up in external borrowing in the 30 months leading to the GFC was associated with only a 50 percent increase in overall lending. The GFC reduced a quarter of the external debt stock, yet overall lending continued to grow on a moderated path.
- Foreign currency lending patterns:
  - Hungary exemplifies a close tie between external borrowing and currency mismatches; banks tend to denominate loans in foreign currency when funding is in foreign currency.
  - Australia and Korea show weak links between external borrowing and foreign currency lending.

### Key statistics and exact figures (as presented in the source)
- Short-term external debt of Korean banks:
  - US$60 billion in the first quarter of 2006.
  - US$160 billion in the third quarter of 2008.
- Net forward positions explanation: up to 60 percent of increase in short-term external borrowing attributable to increases in net forward positions (after netting out portions backed by swaps), per Bank of Korea internal data.
- BOK/FSS thematic examination (January 2008): banks square about three quarters of forward overbought positions by external borrowing and 20 percent by swap transactions.
- Panel estimation details: sample period 2004Q4-2011Q4; 15 domestic banks and 27 foreign bank branches included.
- Table 1 (Companies’ FX forwards) preserved numeric entries exactly (see bullets above under Fact 2).

*Italic: Extracted from IMF source document: _wp12268 - 1. External Borrowing and Lending: Korea, Australia, and Hungary .................................6*

### 0.3 percentage point increase in the short-term external debt (scaled the same way) in the

### _wp12268 - 0.3 percentage point increase in the short-term external debt (scaled the same way) in the

### Determinants of Korean Banks’ Short-Term External Debt (empirical findings)
- Dependent variable: ST external debt/ total asset.
- Key significant coefficients (foreign bank branches):
  - ST external debt(-1)/ total asset(-1): 0.4424*** (Robust St.err 0.0960)
  - FX derivatives positions /total asset: 0.3186*** (Robust St.err 0.0558)
  - BIS capital adequacy ratio: -0.0740** (Robust St.err 0.0352)
  - Inflation(-1): -3.9555*** (Robust St.err 0.7462)
  - △Log(exr)(-1): -0.1272* (Robust St.err 0.0750)
  - Exr volatility(-1): 0.1242*** (Robust St.err 0.0260)
  - CA(-1)/GDP(-1): -0.5967** (Robust St.err 0.2313)
  - Reserve(-1)/GDP(-1): 0.3312** (Robust St.err 0.1326)
  - Incentive: 2.0640*** (Robust St.err 0.6774)
  - VIX: -0.3836*** (Robust St.err 0.0988)
- Key significant coefficients (domestic banks):
  - ST external debt(-1)/ total asset(-1): 0.6962*** (Robust St.err 0.0611)
  - FX derivatives positions /total asset: 0.1110 (Robust St.err 0.0704)
  - BIS capital adequacy ratio: -0.0431** (Robust St.err 0.0208)
  - Inflation(-1): -0.1493 (Robust St.err 0.1077)
  - △Log(exr)(-1): 0.0087 (Robust St.err 0.0096)
  - Exr volatility(-1): -0.0023 (Robust St.err 0.0023)
  - CA(-1)/GDP(-1): -0.0395 (Robust St.err 0.0268)
  - Reserve(-1)/GDP(-1): -0.0824 (Robust St.err 0.0977)
  - Incentive: 0.0004 (Robust St.err 0.0910)
  - VIX: 0.0087 (Robust St.err 0.0171)
- Model diagnostics:
  - AR (1) for foreign bank branches: -2.5596 <p-value> <0.0105>
  - AR (2) for foreign bank branches: 0.6172 <p-value> <0.5371>
  - AR (1) for domestic banks: -2.0906 <p-value> <0.0366>
  - AR (2) for domestic banks: 1.3315 <p-value> <0.1830>
- Note: ***,**, * denote significance in 1, 5, and 10 percent range.

### Bank-specific positioning and mechanics of short positions
- Two primary ways banks squared forward overbought positions:
  - Borrowing foreign currency, selling it into Korean won, and investing the proceeds in domestic bonds (typically foreign bank branches).
  - Creating a short FX position by combining cross-currency (or FX) swap and a spot selling of foreign currency (typically domestic banks).
- Foreign bank branches’ typical trade flow:
  - Buy FX forward from client; borrow FX from headquarter; convert into KRW in spot market; use KRW to buy Monetary Stabilization Bonds (MSBs) or Korea Treasury Bonds (KTBs).
  - This generated risk-free profits from gaps between interest differentials and swap rates but created FX liquidity mismatches because forwards often had longer maturities (particularly with shipbuilders) than their borrowing.
- Foreign bank branches used short-term external debt to offset longer-term exposures in FX forwards because of:
  - Cheaper interest cost of funding.
  - Low perceived liquidity risk amid abundant global liquidity pre-GFC.

### Evidence of liquidity mismatches and vulnerabilities (facts and mechanisms)
- Fact 3: Foreign bank branches have built up liquidity mismatches.
  - Foreign bank branches were mainly accountable for FX liquidity mismatches in the banking system before Lehman (Figure 3).
  - FX derivative composition (Q3 2008): FX forwards (a bulk exceeding 1 year) and FX swaps (1 to 3 months) on the long side; cross-currency swaps (1 to 3 years) on the short side.
  - Branches raised short-term external debt to offset net open positions related to:
    - Hedging services to nonbank clients (gap between net long forward and net short CRS; ~one half of net open positions).
    - Intermediary roles in FX swap market (net long FX swap position; pairs with domestic banks’ net short FX swap position; ~remaining half).
  - Main source of liquidity mismatches: hedging operations (net long forward positions), not short-term FX swap positions.
- Vulnerability dynamics:
  - Foreign banks’ global operations and access to liquidity backstops made liquidity mismatches appear low-risk until a global interbank dollar funding freeze occurred (e.g., Lehman), which realized tail risk and stressed Korea’s banking system.
- Domestic banks’ behavior:
  - More conservative: offset bulk of net long forwards by net short CRS positions (long term).
  - Obtained FX swap funding mainly to offset short-term net long forward positions.
  - Domestic banks incurred little net open positions in their forward books.

### Linkages between short-term external debt, FX funding, and exchange rate volatility
- Regression analysis (Section V referenced) establishes pre-crisis accumulation of short-term external debt as a cause of exchange rate volatility spikes during the GFC.
- Mechanism during the GFC:
  - Sudden stop in capital flows hit foreign bank branches; rollover difficulties on short-term external debt forced conversion of KRW liquidity into FX.
  - Cost of onshore dollar funding rose sharply as foreign bank branches retrenched and offshore arbitrageurs curtailed CRS funding; market liquidity in dollar funding dissipated.
  - Foreign bank branches liquidated bond positions to obtain FX, generating downward pressure on the won.
  - Depreciation could stoke demand for dollar swap funding as market participants expected exchange-rate overshoot.
- Empirical association:
  - Positive association between exchange rate volatility and short-term external debt is consistent with existing studies.

### Why is the Korean won so volatile? (drivers identified)
- Characteristics linked to high volatility:
  - High degree of capital market openness and deep capital markets combined with relatively liberal capital account regulation.
  - Exchange rate regime: since late 1997 free floats; market perception of authorities being more accommodative during depreciating spells than appreciating spells, generating expectations of trend appreciation during upswings in global capital flows.
  - Market infrastructure: despite growth of onshore swap and offshore NDF trading, won’s share in global FX markets falls below what would align with Korea’s nominal GDP; relatively small and less diversified trader base leads to shallow FX market susceptible to liquidity drying-up.
  - Linkages with FX funding market amplify volatility: FX liquidity mismatches, sudden stop in capital flows, drying up of dollar funding, forced asset liquidation, and depreciation-feedback loops.
- Empirical observations and indicators:
  - KRW implied volatility (3M option) and stock index returns show pronounced volatility surges during risk-off episodes (e.g., Lehman crisis; Greek crises).
  - Bid-ask spreads and their volatility are closely linked with spikes in exchange rate volatility (direction of causality is difficult to determine).

### Changes since the GFC that reduced vulnerabilities
- Decline in demand for medium- to long-term FX hedging by shipbuilders due to a sharp global decline in ship orders and changed expectations over the won, reducing FX liquidity mismatches of foreign bank branches.
- Macroprudential measures introduced:
  - Ceilings on banks’ foreign exchange derivative positions.
  - Tighter foreign exchange liquidity standards complemented by more frequent stress tests.
  - A macroprudential levy on foreign currency-denominated noncore bank liabilities.
  - Reinstated withholding tax on foreign investors’ interest income on government bonds.
- Shift in investor base for domestic bonds:
  - Offshore entities, including real money and sovereign investors, increasingly replacing foreign bank branches as investors in domestic bonds.
  - Reversal in relative prominence of these investor groups as net buyers of domestic bonds since late 2009.
  - Increase in share of central banks in foreign investor holdings of domestic bonds to 35 percent by end-2011.
- Improved reserve coverage:
  - Korea’s reserve to short-term debt cover has significantly improved, giving the BOK more capacity to mitigate shocks in onshore dollar funding and FX markets.
- Evidence of weakened feedback loop:
  - May 2012 bout of risk aversion: onshore dollar funding market showed little signs of stress, with one year cross-currency swap spread widening to minus 150 basis points, compared to minus 300 basis points at its trough during the third quarter of (period referenced).

*Source: IMF Working Paper content (_wp12268).*

### 2011. The CDS spreads for major Korean banks also widened to only 150 basis points, about

### _wp12268 - 2011. The CDS spreads for major Korean banks also widened to only 150 basis points, about

### Exchange Rate Volatility and Resilience
- In May 2012, Korean banks were "riding out global shocks much better than they did three or four years ago."
- The CDS spreads for major Korean banks widened to only 150 basis points, about half the peak reached during the third quarter of 2011, and compared with 700–800 basis points seen during the GFC.
- The spread widened significantly less than those of many international comparators (e.g., Australian banks).
- Regression analysis tests sensitivity of exchange rate volatility to:
  - external shocks proxied by VIX,
  - foreign currency liquidity risk proxied by the banking sector short-term external debt-to-GDP ratio,
  - the interaction between VIX and the short-term external debt-to-GDP ratio.
- The analysis included standard control variables from existing studies and variables capturing structural changes such as diversification of the investor base of domestic bonds.

### Regression Findings (Determinants of the Exchange Rate Volatility)
- Sample: Jan. 2006 to Dec. 2011. Dependent: Won-Dollar Exchange Rate Volatility.
- Key coefficients and statistics (coefficient (Newey-West standard error)):
  - Banks’ short-term external debt/GDP: 0.015** (0.006)
  - Yield spread(US) (3Yr – FFR): 0.188** (0.082)
  - Trade growth(-1): -3.457** (1.695)
  - Inflation(CPI): -0.032 (0.059)
  - GDP growth(-1): -0.025* (0.014)
  - Foreigners’ stock buying(net)(-1): 0.000 (0.000)
  - D(Foreigners’ share in bond markets): -0.415* (0.240)
  - D(FBBs’ share in bond markets): -0.089 (0.117)
  - D(VIX): -0.062** (0.025)
  - D(VIX)*Short-term external debt/GDP: 0.002*** (0.000)
  - Constant: 0.083 (0.300)
  - R-squared: 0.787
  - Adjusted R-squared: 0.752
- Three notable points from the regression:
  - The sensitivity of exchange rate volatility to changes in VIX is a function of the external debt-to-GDP ratio. Sensitivity peaked at 0.08 (October 2008), fell to 0.03 (April 2010), and further declined to 0.01 (prior to euro deleveraging in Q3 2011).
  - Counterfactual historical simulation: shocks in VIX of the same magnitude as the GFC peak (August–October 2008) replicated during April–June 2010 and July–September 2011 yield peak exchange rate volatility that would have been 50 percent less (first Greek crisis) and 70 percent less (euro deleveraging) than during the GFC peak. Findings suggest increased FX market resilience mainly because of smaller FX liquidity risk exposure.
  - A significantly negative association exists between foreigners’ share in the Korean bond market and exchange rate volatility, suggesting diversification of the investor base reduces risk of market herding.

### CIP Deviations (Annex: Determinants of CIP Deviations)
- CIP deviations derived from three-month won-dollar FX and one-year won-dollar cross-currency swap markets.
- Regressors included month-on-month changes in VIX, TED, CDS spread, net forward exchanges sold by companies, net bond investment by foreign investors (KTBs and MSBs), and net bond investment by foreign bank branches. Net bond investment variables lagged by one period.
- Global risks measured by first principal components of changes in VIX and TED (VIX_TED).
- Sample periods: January 2003 to July 2007 (pre-crisis), August 2007 to June 2009 (crisis), July 2009 to December 2011 (post-crisis).
- Summary of results by period:
  - Pre-crisis period:
    - Sovereign CDS spread and net forward positions sold by companies were positively and significantly associated with CIP deviations.
    - Global risk factors (VIX, TED) and interaction terms were not significant.
  - Crisis period:
    - VIX became significantly positive: coefficients ranged from 0.127 to 0.165 for three-month CIP deviation and 0.045 to 0.066 for one-year CIP deviation.
    - Net bond investment by foreign investors was significantly negative, indicating sell-offs exacerbated impact of global risk aversion.
  - Post-crisis period:
    - Country risk factor (CDS spread), demand for hedging (net forwards sold by companies), and global risk factors lost significance.
    - Net bond investments by both foreign investors and foreign bank branches significantly affected CIP deviation in the three-month FX swap market.
    - An increase in investor share by foreign investors, at the expense of foreign bank branches, brought some stability to the market.
    - Interaction terms were significant, indicating changes in supplier base and interactions matter post-crisis.
- Tables A-1 and A-2 report detailed coefficients across pre-GFC, GFC, and post-GFC periods for three-month and one-year CIP deviations with Adjusted R-squared values reported:
  - Table A-1 (three-month CIP deviation) Adjusted R-squared: 0.443, 0.412, 0.379, 0.367, 0.364, 0.421 (across reported columns).
  - Table A-2 (one-year CIP deviation) Adjusted R-squared: 0.515, 0.506, 0.582, 0.587, 0.451, 0.383 (across reported columns).
- Interpretation: Pre-crisis CIP deviations explained mainly by country risk and hedging demand; crisis CIP deviations by global risk factors; post-crisis CIP deviations by changes in supplier base and interaction effects.

### Policy Implications
- Main findings:
  - Evidence supports increased resilience of the Korean financial system since the GFC, particularly due to smaller short-term external debt weakening the link between exchange rate volatility and global risk factors.
  - Excessive exchange rate volatility during the GFC resulted from massive buildup of short-term external debt driven by expectations of trend appreciation of the Korean won.
- Macroeconomic policy trade-offs:
  - Surges in capital inflows can cause currency overshooting, hurting international competitiveness.
  - Central bank intervention to prevent overshooting complicates monetary policy and can build downside tail risks to financial stability.
  - In Korea, tail risks manifested as excessive short-term debt accumulation due to expectations of trend appreciation.
- Macroprudential measures:
  - Since the GFC, Korea introduced measures mainly aimed at curbing short-term external borrowing.
  - These measures have helped mitigate FX funding risk associated with short-term debt buildup and can act as speed bumps moderating capital flow cycles.
  - Macroprudential measures are not a panacea and may have efficiency costs; optimization is needed focusing on strengthening institutional framework, minimizing risk of overkill, and closing loopholes.
- Recommendations for ongoing policy actions:
  - Continuous surveillance of financial markets.
  - Close monitoring and analysis of banks’ and corporate sector balance sheets.
  - Frequent stress tests to ensure robustness of the economy and financial system to shocks.

*Source: Authors’ estimates and analysis as presented in the supplied content.*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2012/_wp12268.pdf_