## _cr1506

## Source details

**Canonical URL:** [_cr1506](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/scr/2015/_cr1506.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/scr/2015/_cr1506.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/scr/2015/_cr1506.pdf.json)

---

### EXECUTIVE SUMMARY — Overview and key findings
- The Financial Sector Assessment Program (FSAP) for Korea combined top-down (TD) and bottom-up (BU) stress testing of Korea’s commercial banks and certain non-bank depository institutions (NBDIs).
- The FSAP team did not have access to confidential supervisory data for stress tests and contagion analyses; analyses were implemented by the authorities in cooperation with participating banks, with the FSAP team performing methodological and estimation validations.
- Stress tests covered banks’ solvency and liquidity and included contagion evaluation stemming from funding pressures and potential defaults.
- Overall finding: Korean banks are well-positioned to withstand severe macrofinancial shocks; all major banks remain well capitalized and meet the Basel III capital requirements in the tested scenarios.
- Aggregate capital adequacy ratio (CAR) of the banking system was 14.3 percent as of end-September 2013, up from 12.5 percent at end-2008.

### Stress testing framework and data
- TD solvency tests used the Bank of Korea’s SAMP framework complemented by BOK macroeconomic model projections; BU solvency tests were carried out by individual banks.
- TD liquidity tests were carried out by the Financial Supervisory Service (FSS); BU liquidity tests were performed by individual banks using different outflow assumptions.
- All stress-test assumptions and parameters were agreed between the authorities and the FSAP team.
- Coverage:
  - TD solvency tests covered all commercial banks (seven nationwide banks and regional banks aggregated) — these account for 100 percent of aggregate assets of commercial banks, 63 percent of aggregate assets of the entire banking system, and 30 percent of aggregate assets of financial institutions.
  - BU solvency tests covered the seven nationwide commercial banks, accounting for 76 percent of aggregate assets of commercial banks.
- Key limitations:
  - No direct FSAP access to supervisory data; certain segments excluded (policy banks, foreign bank branches, some NBDIs).
  - The BOK macroeconomic model does not fully embed feedback loops from higher mortgage rates to household debt burdens and consumption; authorities applied an exogenous ad hoc consumption shock specified by the FSAP team to capture some second-round effects.

### Solvency stress-test scenarios and calibrations
- Four macro scenarios evaluated (five-year horizon through end-2017, baseline from April 2013 WEO):
  - Mild global double-dip (GDD): shock equivalent to one standard deviation of real GDP growth from baseline in the first two years.
  - Severe global double-dip (GDD): shock equivalent to two standard deviations of real GDP growth from baseline in the first two years; cumulative two-year 6.4 percent drop in real GDP under the severe shock.
  - Prolonged slow growth (PSG): cumulative real GDP growth deviating by one standard deviation below baseline over a five-year horizon; PSG assumes a 2.5 percentage point deviation from baseline each year over five years.
  - Interest rate increase: 250 basis point upward shock over 2 ½ years, starting in the second half of 2014.
- Single-factor sensitivity tests included:
  - Interest rate spikes of 300 basis points and 200 basis points.
  - FX depreciation of 50 percent and 30 percent.
  - Drop in real house prices of 20 percent and 15 percent.

### Solvency stress-test results and implications
- PD dynamics:
  - Under severe and mild GDD, PDs rise over initial two years then decline but remain elevated at 2.9 percent in 2016 (severe) and 1.6 percent in 2016 (mild).
  - Maximum PD increases in 2015: severe scenario: 181 percent rise to 3.3 percent; mild scenario: 73 percent increase to 2 percent.
  - Under PSG, PDs increase over time to about 2 percent in 2017.
- System-wide CAR outcomes:
  - System-wide CARs remain considerably above the Basel III requirement even under severe macro conditions (Scenario 2).
  - By end-2017 under the severe scenario: aggregate CAR with contagion declines to 12 percent, aggregate CAR without contagion 13.6 percent — a 3.6 percentage point drop relative to the baseline.
  - Single-factor impacts: a 300 basis point interest rate spike accounts for up to a 1.9 percentage point drop in aggregate CAR; all banks remain at or above 10.7 percent.
  - A sizable drop in housing prices causes more than 2 percentage point drop in CARs, but banks remain well capitalized.
- Credit risk composition:
  - Credit risk mostly related to corporate exposures: SMEs and large corporations drive potential credit costs; household lending–linked costs are more muted.
  - Residential mortgage loan credit costs are small, reflecting recourse debt nature and average system-wide loan-to-value (LTV) ratio of about 50 percent.
- Market risk:
  - Market risk exposures at end-2012 about eight times smaller than credit risk exposures.
  - Cumulative five-year market losses account for about 0.3 percent of RWA at end-2017 under the most adverse scenario.
  - TD five-year market losses up to 0.8 percent of RWA a year; BU 0.1 percent. Aggregate TD loss 0.3 percent of RWA by end-2017; BU aggregate gain of 1.1 percent.

### Liquidity stress tests — TD and BU findings
- TD liquidity stress test (30-day horizon) focused on FX and local currency liquidity conditions using an implied cash flow method; simulated complete freeze of unsecured short-term wholesale funding and severe run-offs.
- TD FX findings:
  - Aggregate FX liquidity gap under a global wholesale funding shutdown: US$3.8 billion (KRW 4.4 trillion) without committed credit lines from international banks.
  - After accounting for committed credit lines: small surplus of around US$155 million (KRW 172 billion).
  - Committed credit lines from international banks equal about 1.2 percent of Korea’s international reserves.
- BU liquidity tests (seven nationwide banks, 30-day horizon) scenarios included worst bank-specific historical outflows and a uniform extreme scenario:
  - System-wide net cash positions after the pre-specified shock: Local currency: KRW 34 trillion; Foreign currency: KRW 9 trillion.
  - Under a repeat of each bank’s worst historical deposit outflow: Net cash positions: KRW 57 and KRW 11 in local currency and foreign currency, respectively.
  - All seven nationwide commercial banks would have adequate liquid assets to withstand a 30-day freeze in wholesale funding and deposit withdrawals.

### Non-bank depository institutions (NBDIs / NBFIs) assessment
- NBDIs collectively account for 20 percent of domestic household credit (and 14.3 percent of financial sector assets at end-2012).
- Stress-test approach: reverse stress test with conservative assumptions (e.g., collateral value subject to a 20 percent discount; only 70 percent of this discounted value recognizable; required provisioning: 20 percent for secured loans and 100 percent for unsecured loans).
- Key result:
  - Some NBDIs—specifically mutual savings banks (MSBs) and credit unions—have thin capital buffers: these institutions were found to be able to handle only up to a 6–8 percentage point increase in nonperforming loans before capital fell below the regulatory minimum (elsewhere results indicate around 10 percentage points for some entities).
  - Even NBDIs that can withstand larger NPL increases should be closely monitored given very low regulatory minimum capital requirements.
- Spillovers: Potential spillovers from NBDI difficulties to other parts of the financial sector are likely to be minimal because NBDI funding is mostly in deposits.

### Household loan structure and vulnerabilities
- High share of mortgage loans—33.6 percent as of end-March 2013—are bullet loans with three-year refinancing, exposing households to refinancing risk.
- Variable-rate loan dominance exposes households to rising financing costs; share of fixed rate loans increased from 5.1 percent at end-2010 to 36.3 percent in March 2013.
- Household debt as share of disposable income rose from 116 percent in 2004 to 156 percent at end-2011.
- Aggregate debt at risk of households, after accounting for asset holdings, accounted for 0.7 percent of household debt at end-2012; a first-order house price decline would raise household debt at risk from 0.7 percent to 1 percent of total household debt.
- Year-on-year growth in banks’ household loans: 2.8 percent as of end-June 2013 (compared to 6.1 percent as of end-June 2011); NBDIs growth: 4.2 percent as of end-June 2013 (down from 15.2 percent); non-depository NBFIs household lending: 13.4 percent year-on-year as of end-June 2013. NBDI lending accelerated to 7 percent by end-2013.

### Key limitations and caveats
- FSAP could not independently run models or directly validate supervisory data; BU results could not be fully validated due to lack of direct FSAP-bank communication.
- Certain financial segments were excluded: policy banks, foreign bank branches, and certain small NBDIs (e.g., community credit cooperatives collectively accounting for about 10 percent of aggregate loans).
- Complex macrofinancial feedback effects only partially captured; interest rate scenario results are subject to caution because nonlinear propagation channels (higher household debt-servicing costs → higher bank credit costs → reduced credit availability → falling house prices) are difficult to model fully.

### Policy recommendations and process enhancements
- Strengthen stress-testing process and inter-agency coordination:
  - FSS should maintain closer interaction with banks to ensure BU stress testing results are fully validated.
  - FSS and BOK should enhance cross-sharing of TD and BU stress testing results to eliminate discrepancies and better inform policy actions.
  - Ideally, stress tests conducted by different agencies should be made public to encourage closer coordination.
- Banks’ internal risk management:
  - Banks should run stress test scenarios distinct from common FSS-provided scenarios and conduct event-specific tests tailored to operational needs.
- NBFI regulatory framework:
  - Apply regulatory and supervisory framework consistent with that for banks; specifically, Basel capital adequacy framework should be applied to all NBDIs, with larger entities of an equivalent size to banks subjected to stricter supervision.
- Ongoing monitoring:
  - Monitor FX liquidity gaps, contingent liabilities, and potential second-round macrofinancial feedback loops closely.

*Source: EXECUTIVE SUMMARY and extracted chapter content, Republic of Korea, International Monetary Fund (content unit: _cr1506).*

### EXECUTIVE SUMMARY ___________________________________________________________________________ 5

### EXECUTIVE SUMMARY

### Overview of the FSAP assessment
- The Financial Sector Assessment Program (FSAP) for Korea was carried out in close collaboration with the authorities and included top-down (TD) and bottom-up (BU) stress testing of Korea’s commercial banks and certain non-bank depository institutions (NBDIs).
- The FSAP team did not have access to confidential supervisory data for the stress tests and the contagion analyses; analyses were implemented by the authorities in cooperation with participating banks, with the FSAP team performing methodological and estimation validations.
- Stress tests covered banks’ solvency and liquidity, and included contagion evaluation stemming from funding pressures and potential defaults.

### Stress testing framework and data
- Solvency TD tests were based on the internal Systemic Risk Assessment Model for Macroprudential Policy (SAMP) developed by the Bank of Korea (BOK), complemented by BOK macroeconomic model projections; BU solvency tests were carried out by individual banks.
- Liquidity TD tests were carried out by the Financial Supervisory Service (FSS), with separate assessments of local currency and foreign exchange (FX) liquidity risks; BU liquidity tests were performed by individual banks using a different set of outflow assumptions.
- All stress-test assumptions and parameters were agreed between the authorities and the FSAP team.
- Limitations: lack of direct access to supervisory data prevented independent implementation; certain financial segments were not included (policy banks, foreign bank branches, some NBDIs); complex macrofinancial feedback effects were only partially captured (BOK macroeconomic model does not fully embed feedback loops from higher mortgage rates to household debt burdens and consumption).

### Solvency stress-test scenarios
- Four scenarios were used to capture key macrofinancial risks:
  - Mild global double-dip.
  - Severe global double-dip (with shocks of one and two standard deviations of real GDP growth, respectively).
  - Prolonged slow growth (PSG): cumulative real GDP growth deviating by one standard deviation below baseline over a five-year horizon.
  - Increase in domestic interest rates in response to a potential unwinding of U.S. unconventional monetary policy (UMP): policy rates assumed to increase by 250 basis points over two and a half years, starting in mid-2014.
- Growth shocks are commensurate with those in other FSAPs; the severe double-dip shock is more extreme compared to the 1997 Asian crisis.

### Solvency stress-test results and implications
- Overall finding: Korean banks are well-positioned to withstand severe macrofinancial shocks.
- Comparative resilience:
  - Under the severe double-dip scenario, Korean banks can absorb shocks of a larger magnitude compared to the 1997 Asian crisis.
  - All major banks remain well capitalized and meet the Basel III capital requirements in the tested scenarios.
- Interest-rate shock:
  - A rise in interest rates due to potential unwinding of U.S. UMP is found to have only a limited negative impact on domestic banks, though results should be treated with caution due to model limitations.

### Liquidity stress tests
- Banks’ liquidity risks in local currency and FX are generally manageable, with a small gap in FX liquidity identified.
- Top-down liquidity tests and bottom-up liquidity tests used different assumptions and produced assessments of 30-day cash flows and liquid asset ratios (detailed assumptions and results presented in the full assessment).

### Assessment of non-bank depository institutions (NBDIs)
- NBDIs collectively account for 20 percent of domestic household credit.
- Some NBDIs—specifically mutual savings banks (MSBs) and credit unions—have thin capital buffers:
  - These institutions were found to be able to handle only up to a 6–8 percentage point increase in nonperforming loans before capital fell below the regulatory minimum.
- Stress tests of non-bank institutions were essential given their sizable share of household lending; however, certain types of individually small but collectively large NBDIs were excluded due to lack of supervisory data.

### Background and recent macrofinancial context
- Growth developments:
  - Economic growth moderated sharply to 2 percent in 2012 from 6.3 percent in 2010; recovery was modest in 2013 supported by private consumption and construction.
  - Over the past decade, Korean growth has been relatively stable and less volatile relative to G20 peers.
- Policy response:
  - The government introduced a fiscal stimulus of 1¼ percent of GDP via a supplementary budget in April 2013.
  - The 2013 fiscal deficit (excluding social security funds) is projected to widen to 2.1 percent of GDP.
  - The Bank of Korea introduced cumulative cuts in the policy rate of 50 basis points since July 2012.
  - In April 2013, the government introduced a comprehensive policy package to stimulate the housing market.
- Post-2008 crisis stabilization measures (selected):
  - BOK provided KRW 16.8 trillion in liquidity via long-term repos and set aside KRW 5 trillion (KRW 2.1 trillion used) for corporate bond/commercial paper purchases.
  - Currency swap lines and reserves provided up to US$56 billion in USD liquidity.
  - Government guaranteed banks’ foreign debt in 2009 (limited use).
  - A Bank Recapitalization Fund of KRW 20 trillion was established in December 2008; a one-time injection of KRW 4 trillion into eight banks was carried out in March 2009.
  - KAMCO stood ready to issue up to KRW 40 trillion government-backed bonds to purchase corporate NPLs (only a fraction used).
  - SME credit guarantees and subsidized lending facilities were expanded.
  - Macroprudential tools included a ceiling on long-term FX loan-to-deposit ratio (2009), a cap on forward FX contracts (2009), a leverage cap on banks’ FX derivatives positions (2010), and a levy on non-deposit FX liabilities (2011).
- Banking-sector metrics and exposures:
  - Aggregate capital adequacy ratio (CAR) of the banking system was 14.3 percent as of end-September 2013, up from 12.5 percent at end-2008.
  - FX lending by Korean commercial banks was 9 percent of total loans at end-2012.
  - At end-2011, household debt as a share of disposable income was 156 percent.
  - At end-June 2013, NBFIs accounted for roughly a half of outstanding household loans.
  - Bank lending is concentrated in the corporate sector (61 percent of total loans overall; 56 percent for commercial banks; 70 percent for specialized banks). Three industries—manufacturing, wholesale and retail, and real estate—account for roughly two thirds of corporate lending, with manufacturing alone at 44 percent of the total.

### Key limitations of the assessment
- No direct access to supervisory data; implementation relied on authorities and participating banks with FSAP validation.
- Stress tests excluded policy banks, foreign bank branches, and certain NBDIs due to data limitations.
- Macroeconomic model limitations: the BOK macroeconomic model does not fully account for feedback from higher mortgage rates to household debt servicing burdens and consumption; authorities applied an exogenous ad hoc consumption shock specified by the FSAP team to capture some second-round effects.

*Source: EXECUTIVE SUMMARY, REPUBLIC OF KOREA, INTERNATIONAL MONETARY FUND.*

### 6. The current structure of household loans poses a vulnerability to the financial system,

### 6. The current structure of household loans poses a vulnerability to the financial system,

### Household loan structure and risks
- A high share of mortgage loans—33.6 percent as of end-March 2013—is structured as bullet loans, with households repaying interest and refinancing loans every three years, exposing households to considerable refinancing risks.
- Household debt is dominated by variable rate loans, exposing households to a sharp rise of financing costs and banks to rising credit risks in case of an upward shift in interest rates.
- The share of fixed rate loans increased from 5.1 percent at end-2010 to 36.3 percent in March 2013, reflecting regulatory guidance:
  - This has lowered interest rate risk for households, but has effectively transferred it to banks, potentially raising banks’ vulnerabilities if they do not adequately hedge this risk or price the fixed-rate loans properly.
- As of end-June 2013, year-on-year growth in banks’ household loans was 2.8 percent, compared to 6.1 percent as of end-June 2011; for NBDIs, growth was down to 4.2 percent from 15.2 percent. Household lending by non-depository NBFIs continued to rise rapidly, at 13.4 percent year-on-year as of end-June 2013. NBDI lending accelerated to 7 percent by end-2013.

### Banking sector profitability and income composition
- Banking sector profitability has faced downward pressures due to:
  - (i) the sustained low interest rate environment, compressing net interest margins (NIMs);
  - (ii) heightened credit costs from deteriorating macroeconomic conditions, with higher write-offs and distressed asset sales;
  - (iii) conservative regulatory provisioning standards (up to 150 percent of expected losses); and
  - (iv) supervisory pressures on banks to cut interest rates and fees.
- Return on assets (ROA) of commercial banks in 2012 was 0.5 percent, compared to 1–1¼ percent in 2005–07.
- The share of non-interest income since the onset of the crisis averaged 17 percent of total income in 2008–11.

### Asset quality, NPLs, and asset disposal practices
- Commercial banks’ NPLs accounted for 1.3 percent of total loans at end-2012 and have generally been persistently low (at less than 1 percent) in recent years.
- Low NPL levels reflect active disposal of bad assets, with an average of about 1 percent per year written off or sold, mostly via market mechanisms, to Asset Management Companies (AMCs), such as the United Asset Management Corporation (UAMCO).
- New NPL flows outstrip outstanding NPLs and account for about 1.7-2.6 percent of total loans.
- AMCs’ recovery rates are allegedly high, with lenders legally having senior claims on up 120 percent of the face value of a loan.
- Risk: If housing prices decline sharply, disposal mechanisms may break down if there is scarcity of counterparties in asset auctions.

### Corporate sector debt at risk
- The aggregate debt at risk of the corporate sector accounted for 29 percent of corporate debt at end-2012, only marginally lower than 31 percent at end-2008.
- Risks are concentrated in fewer companies in certain sectors, including construction, ship-building and cross-border transportation.
- Definition: Companies are considered financially distressed, and their debt at risk, if their earnings before tax and interest cannot fully cover their interest expenses—i.e., their interest coverage ratios are below 1.

### Household indebtedness and medium-term vulnerabilities
- Household debt as a share of disposable income increased from 116 percent in 2004 to 156 percent at end-2011.
- Both the banking sector and non-banking sector have considerable exposures to households, with the latter accounting for roughly half of household loans.
- Near-term risks appear muted due to high household wealth, prudential buffers, and improving growth outlook.
- Aggregate debt at risk of households, after accounting for asset holdings, accounted for 0.7 percent of household debt at end-2012.
- A first-order effect of house price decline (without second-round effects) would raise household debt at risk from 0.7 percent to 1 percent of total household debt.
- Concern: Decline in credit quality and an upsurge in credit costs could create a negative feedback loop via slower growth and further declines in household creditworthiness.
- Note: Debt at risk is defined as the debt owed by financially distressed borrowers with debt service ratio (DSR) over 40 percent.

### External funding and global market risk exposure
- Post-2008, banks decreased dependence on short-term wholesale funding in global markets but remain exposed to potential funding pressures if global financial risks increase sharply (e.g., re-intensification of stress in the euro area where most foreign funding originates).
- Korea has so far been largely unaffected by capital outflows related to a potential UMP exit, but its status as a safe haven is untested. A disorderly UMP exit and sudden reversal of capital flows could impair business sentiment and investment, tighten financial conditions, and raise funding costs for banks and corporates.

### Stress testing framework and limitations
- The FSAP stress testing exercise combined TD and BU stress tests of commercial banks’ solvency and liquidity, including contagion evaluations due to funding liquidity pressures and potential defaults.
- The FSAP team did not have access to confidential supervisory data; analyses were carried out by the authorities with methodological and estimation validations by the FSAP team.
- TD solvency tests used the BOK’s SAMP framework, complemented by macro projections from the BOK macroeconomic model. BU tests were carried out by individual banks. TD liquidity stress tests (local currency and FX) were carried out by the FSS.
- Key limitations:
  - The FSAP team did not have direct access to supervisory data and could not independently run models nor verify parameters against underlying data.
  - The FSAP team did not have direct communication with banks on the BU stress testing process; BU results and follow-up inquiries were communicated by the FSS.
  - Stress tests excluded certain segments: policy institutions and foreign bank branches; certain small NBDIs (e.g., community credit cooperatives) that collectively account for about 10 percent of aggregate loans were excluded due to insufficient supervisory data.
  - Complex macrofinancial feedback effects were captured only partially. The BOK macroeconomic model only partially accounts for links between higher interest rates and banks’ credit risk and does not embed full feedback loops from higher mortgage rates to households’ debt payment burdens and consumption. To account for ‘second-round’ effects, an exogenous ad hoc consumption shock specified by the FSAP team was applied.

### Authorities’ stress testing practices (Box summary)
- The BOK conducts TD stress tests for the commercial banking sector, focusing on credit risks from: (i) loans to large enterprises; (ii) SME loans; (iii) mortgages; (iv) credit card loans; and (v) other household loans, and market risks on trading and AFS accounts from movements in interest rates, stock prices and exchange rates.
- The BOK’s SAMP integrates macro-financial systemic risk analyses, projects macro-financial factors from a joint fat tail-risk distribution, and permits assessment of propagation of second-round default and liquidity contagion effects via network models.
- The FSS conducts TD FX liquidity stress tests and requires BU stress testing by banks (biannual for solvency; monthly for FX liquidity). Scenarios for FX liquidity are aligned with deterioration experienced during the 2008 crisis and the Asian crisis.
- Areas for enhancement:
  - Expand coordination and result-sharing between the BOK and the FSS to enable cross-validation of BU and TD results.
  - Improve validation of BU exercise results supervised by the FSS (e.g., validation of banks’ loss estimation methods).
  - Adopt a framework for ongoing stress testing of NBFIs.

### Banking sector solvency stress test design (framework highlights)
- Solvency tests evaluated credit and market risks under extreme but plausible macroeconomic scenarios with a five-year risk horizon through end-2017, using end-2012 data.
- Coverage:
  - TD stress tests covered all commercial banks (seven nationwide banks and regional banks aggregated). These account for 100 percent of aggregate assets of commercial banks, 63 percent of aggregate assets of the entire banking system, and 30 percent of aggregate assets of financial institutions. TD tests were carried out on a bank-level consolidated basis using bank-by-bank data by the BOK.
  - BU tests covered the seven nationwide commercial banks, which account for 76 percent of aggregate assets of commercial banks, 56 percent of aggregate assets of the entire banking sector and 27 percent of aggregate assets of financial institutions.
- Risk types assessed:
  - Credit risks related to on- and off-balance sheet exposures; exposures to specific sectors including corporations (large corporations and SMEs) and households (mortgages, credit cards and other lending). In BU, banks provided breakdowns of losses by type of off-balance sheet exposures (derivatives and non-derivatives). Banks did not provide estimates of credit risks due to large exposures, nor industry-specific breakdowns for certain sectors.
  - Market risks based on changes in prices of held securities, broken down by account (trading and AFS) and by type of security (domestic sovereign debt, other domestic debt, foreign debt, equities and other securities). HTM positions were assumed not subject to shocks. Trading account gains/losses recognized through profits and losses; AFS through reserves.
- Behavioral assumptions:
  - Balance sheets assumed to grow based on (i) projected level of credit for each year to 2017 (projections estimated using the BOK’s SAMP) and (ii) an assumed constant share of non-loan assets in total assets based on average long-term growth.
  - For TD, baseline profitability projections based on the BOK’s SAMP bank profit and loss module linking banks’ profits to twelve macro risk factors.
  - For BU, banks projected net interest income under baseline using internal models; BOK’s SAMP provided projections of non-interest income.
  - Existing NPLs were adjusted up to include NPL sales given active disposal practices. New problem loans were assumed to maintain a default status and banks to provision fully any losses over the entire five-year stress testing horizon.
- Solvency benchmarks:
  - Potential risks assessed using Tier 1 capital and total regulatory capital benchmarks. The hurdle rate is based on Basel II for 2013 and on the Basel III schedule starting 2014, including a conservation buffer but excluding SIFI surcharges.
  - Definition of capital based on Basel III; definition of risk-weighted assets based on Basel 2.5 and III.

*Source: Republic of Korea staff report, IMF Financial Sector Assessment Program (FSAP) — extracted chapter content.*

### 15. The stress testing exercise also included sensitivity tests to individual risk factors. The

### _cr1506 - 15. The stress testing exercise also included sensitivity tests to individual risk factors. The

### Sensitivity tests and single-factor analyses
- Single-factor sensitivity analyses quantified the impact of shocks associated with one-time changes in individual risk factors, all else equal.
- Sensitivity tests assessed credit risks related to:
  - a one-time spike in interest rates,
  - exchange rate depreciation, and
  - a fall in real house prices.
- Motivation: high share of variable-rate loans to households and households’ vulnerability to housing price shocks given high mortgage debt.
- Single-factor shocks in the TD included:
  - interest rate spikes of 300 basis points and 200 basis points;
  - FX depreciation of 50 percent and 30 percent;
  - drop in real house prices of 20 percent and 15 percent.

### Solvency scenarios (Section B)
- Four adverse macroeconomic scenarios evaluated against a baseline based on the April 2013 IMF World Economic Outlook (WEO) projections:
  - Scenario 1 (mild GDD): shock equivalent to one standard deviation of real GDP growth from baseline in the first two years of the five-year horizon; positive (linear) adjustment to baseline in the third year; linked to a global shock including a significant slowdown in China.
  - Scenario 2 (severe GDD): shock equivalent to two standard deviations of real GDP growth from baseline in the first two years; (linear) positive adjustment to baseline in the third and fourth year; linked to deterioration in global conditions.
  - Scenario 3 (PSG, Prolonged slow growth): cumulative real GDP growth rate deviating by one standard deviation below baseline over a five-year horizon; magnitude based on deviation from trend of five-year cumulative GDP growth rate between 1990 and 2012; assumed linked to lingering uncertainty in advanced countries.
  - Scenario 4 (interest rate increase): increase in domestic interest rates in response to unwinding of UMP globally; scenario captures household repayment stress given prevalence of variable-rate mortgages and second-round effects via the real economy (consumption, house prices, credit costs).
- Calibration and magnitudes:
  - GDP shocks for severe (two) and mild (one) standard deviations based on GDP growth between 1990 and 2012.
  - Cumulative two-year 6.4 percent drop in real GDP under the severe shock.
  - PSG assumes a 2.5 percentage point deviation from baseline each year over a five-year period, accounting for cumulative five-year GDP growth of 6.4 percent (below historical minimum 13.1 percent since 1980).
  - Interest rate scenario: 250 basis point upward shock over 2 ½ years, starting in the second half of 2014; motivated by potential U.S. interest rate normalization.
- Scenario inputs used projected macroeconomic and financial risk factors estimated via the BOK macro model (BOK12), including: GDP growth (for interest rate scenario); inflation; unemployment rate; policy interest rate; credit growth; house prices; equity prices; corporate (A-rated) spreads; short- and long-term government bond yields; exchange rate vis-à-vis the USD.
- For scenario 4, authorities applied an additional exogenous ad hoc consumption shock specified by the FSAP team to account for feedback loops from higher mortgage rates to households’ debt payment burdens and consumption.

### Credit risk modeling (Section B, continued)
- Credit risk estimated based on EAD, PD, and LGD for each principal loan type:
  - (i) large enterprise loans;
  - (ii) SME loans;
  - (iii) mortgage loans;
  - (iv) credit card loans;
  - (v) other household loans.
- EAD:
  - Estimated based on projected exposures for each year to end-2017.
  - Credit growth for the stress testing horizon determined endogenously via the BOK SAMP.
  - EAD did not include committed (but undrawn) credit lines that may be withdrawn by borrowers in distress.
- PD:
  - Estimated via the SAMP model using a logit transformation of banks’ NPL ratio as proxy for default rate.
  - New NPLs adjusted for write-offs and sales to reflect newly incurred NPLs.
  - Proxied PDs related to macrofinancial variables distinct for each loan type to estimate stressed PDs.
- LGD:
  - Approximated by a variant of the S&P model (2010), linking LGD dynamics to PD dynamics.

### PD outcomes and dynamics (Figure 7 discussion)
- PDs rise considerably under all scenarios.
- Under severe and mild GDD, PDs rise over initial two years then decline but remain elevated at:
  - 2.9 percent in 2016 (severe),
  - 1.6 percent in 2016 (mild).
- Maximum increase in PDs occurs in 2015:
  - severe scenario: 181 percent rise to 3.3 percent;
  - mild scenario: 73 percent increase to 2 percent.
- Under PSG, PDs increase over time to about 2 percent in 2017.
- Interest rate scenario produces a more muted PD rise due to model constraints in accounting for interest rate shocks.

### Second-round contagion assessment (TD)
- In TD solvency tests, contagion-related losses estimated as second-round losses due to liquidity contagion using BOK’s SAMP model.
- Default contagion model unfeasible because high capitalization buffers and lack of bank defaults after first-round solvency shocks (contagion model assumes contagion preceded by default).
- Use of SAMP motivated by inability to access supervisory interbank exposure data for the FSAP team’s own network model.

### Results (Section C)
- System-wide findings:
  - Korean banking system appears sound even under extreme shocks to economic growth or protracted sluggish movement.
  - System-wide CARs remain considerably above the Basel III requirement even under severe macro conditions (Scenario 2), including a 3.8 percent contraction of GDP in 2013 and a subsequent 2.8 percent decline in 2014.
  - On an individual basis, all banks likely to meet Basel III capital requirements, with CARs remaining at about or above 10 percent for all years and scenarios.
- Capital buffers and management:
  - Resilience reflects very high initial capital buffers and conservative capital management, partly via retained earnings (e.g., limiting dividend payments).
- Contagion impact:
  - Banks remain solvent after accounting for potential second-round contagion effects.
  - Aggregate CAR declines marginally with contagion but remains above regulatory thresholds at aggregate and individual bank level.
  - By end-2017 under the severe scenario:
    - aggregate CAR with contagion declines to 12 percent,
    - aggregate CAR without contagion 13.6 percent.
    - This accounts for a 3.6 percentage point drop relative to the baseline.
- Credit cost composition:
  - Credit risk mostly related to corporate exposures: SMEs and large corporations drive potential credit costs under both TD and BU approaches; household lending–linked costs are more muted.
  - BU approach tends to generate higher credit costs related to SME exposures.
  - Residential mortgage loan credit costs are small, reflecting recourse debt nature and average system-wide loan-to-value (LTV) ratio of about 50 percent.
  - No indications of a housing bubble; housing affordability index considerably below OECD peers.
- Market risk:
  - Market risk exposures at end-2012 about eight times smaller than credit risk exposures; mostly related to losses on domestic sovereign debt securities holdings.
  - Cumulative five-year market losses account for about 0.3 percent of RWA at end-2017 under the most adverse scenario.
  - More than half of market losses related to domestic sovereign debt holdings.
  - In the severe GDD scenario, market shocks reduce some banks’ CARs by up to 0.5 percent of RWA during peak of shock in 2015.
- Divergence TD vs BU on market risk:
  - TD five-year losses up to 0.8 percent of RWA a year; BU 0.1 percent.
  - Aggregate TD loss of 0.3 percent of RWA by end-2017; BU aggregate gain of 1.1 percent.
  - Differences partly due to different models and datasets (TD: supervisory balance sheet data; BU: detailed bank-level data) and limited ability to validate BU results due to lack of direct interaction with banks.
- Single-factor impacts on CAR (Figure 11):
  - 300 basis point interest rate spike accounts for up to a 1.9 percentage point drop in aggregate CAR, even after contagion; all banks remain at or above 10.7 percent.
  - Sizable drop in housing prices causes more than 2 percentage point drop in CARs, but banks remain well capitalized.
  - FX risks limited given net open FX position less than 1 percent of capital.

### Further considerations (Section D)
- Cautionary notes due to stress testing limitations and this exercise’s constraints.
- Potential for complex, difficult-to-model macro-financial feedback loops is a major risk requiring close monitoring and robust bank risk management.
- Example: fairly benign modeled impact of sharp interest rate spike (Scenario 4) compared to other scenarios may understate consequences because nonlinear propagation (higher household debt-servicing costs → higher bank credit costs → reduced credit availability → falling house prices) could be more severe than modeled.
- Such loops are difficult to model but could be highly detrimental to macrofinancial stability.

### Banking sector liquidity stress tests — framework (Section A)
- TD liquidity stress test focused on FX and local currency liquidity conditions (30-day horizon).
- Coverage and data:
  - Covered all commercial banks: seven nationwide banks individually and all regional banks as aggregate entity (these account for 86 percent of aggregate assets of commercial banks).
  - Conducted bank-by-bank using end-March 2013 FSS consolidated data; estimations by the FSS and validated by the FSAP team.
- Stress design:
  - Assessed ability to withstand 30-day persistent withdrawals of wholesale funding and customer deposits using implied cash flow method of Schmieder et al (2011).
  - Simulated: complete freeze of unsecured short-term wholesale funding market; 20 percent withdrawal of secured short-term wholesale funding; run-off rates on term and demand deposits slightly exceeding worst past 1-month deposit withdrawals.
  - Liquidity assessed via funding gap: difference between liquid liabilities and assets.
  - Test did not take into account access to central bank liquidity.
- Severity:
  - Applied liquidity shocks more extreme than historical Korean experience, in line with and exceeding Lehman liquidity crisis to gauge resilience to extreme liquidity event.
- Haircuts and committed lines:
  - Assumed haircuts on liquid assets and committed credit lines consistent with extremely severe market liquidity stress, applied to zero risk-weighted exposures and riskier liquid assets.
  - Haircuts broadly in line with severe scenario of Schmieder (2012), applied to FX and local currency positions.
  - Liquidity stress tests estimated both with and without full cut-off of committed credit lines (haircuts of 100 percent and 0 percent). Complete cut-off viewed as extreme tail event but used to evaluate importance of credit lines in severe global liquidity crisis.

*Italic: Source — IMF staff chapter text from _cr1506 (excerpt) contained in the provided PDF content.*

### 30. FX liquidity stress tests suggest that even in the event of unprecedented liquidity

### _cr1506 - 30. FX liquidity stress tests suggest that even in the event of unprecedented liquidity

### Top-down (TD) FX and local currency liquidity stress tests — scenarios and assumptions
- TD scenario highlighted: shutdown of global wholesale funding.
- Liquidity outflows: withdrawal rate (in percent)
  - Customer deposits: Demand 620; Term 4030
  - Short-term wholesale funding: Secured 2020; Unsecured 100100
  - Contingent liabilities 00
- Liquidity inflows: haircut on liquid assets (in percent)
  - Committed credit lines 1/: 75
  - Sovereign debt securities and other zero risk-weighted exposures: 0 and 100
  - Other trading assets: 5
  - Other debt securities: 20
  - Other equity securities: 20
  - Other securities: 30
  - Derivatives: 50

### TD findings — system-wide and bank-level FX liquidity outcomes
- Aggregate FX liquidity gap under a global wholesale funding shutdown:
  - Aggregate gap: US$3.8 billion (KRW 4.4 trillion) without taking into account committed credit lines from international banks.
  - Committed credit lines from international banks equal about 1.2 percent of Korea’s international reserves.
  - After accounting for these credit lines: small surplus of around US$155 million (KRW 172 billion).
- Local currency liquidity stress tests:
  - Individual banks have sufficient liquid assets to handle potential outflows under extreme shocks.
- Distributional results:
  - Liquid asset ratios of individual commercial banks remain sound following severe shocks (e.g., dry-out of global wholesale markets or sizable run-off of FX demand and term deposits).
  - A confluence of FX deposit outflows and wholesale funding pressures can be taxing on individual banks despite relatively small aggregate FX reserve implications.
- Caveat noted:
  - Outflows of banks’ contingent liabilities were not incorporated in the TD stress tests and need active monitoring by authorities.

### Bottom-up (BU) liquidity stress tests — coverage, scenarios, and key assumptions
- Coverage and data:
  - Tests covered the seven nation-wide banks on consolidated bank-level basis. These institutions account for 76 percent of the aggregate assets of commercial banks.
  - Tests used internal models and data as of end-March 2013; stress horizon 30 days.
- Two scenarios:
  - Scenario based on worst bank-specific historical outflows.
  - Extreme scenario applied uniformly across banks.
- Common assumptions across scenarios:
  - Dry-up of wholesale funding, including a complete shutdown of the unsecured wholesale funding market.
  - Roll-over of up to 20 percent of maturing secured wholesale funding.
  - Deposit run-offs calibrated based on worst 30-day bank-specific historical outflows since 1990 or on pre-specified extreme outflows.
  - Banks’ deposit outflows differ based on currency.
  - Banks are subject to outflows from contingent liabilities, such as unused credit lines extended by other parties.
- BU behavioral outflows and inflows (Table 3):
  - Behavioral outflows — General description:
    - New FX swaps: The magnitude of new FX swaps is limited to the initial position.
    - Other new derivatives: The magnitude of new other new derivatives is limited to the initial position.
  - Withdrawal rate (in percent):
    - Demand deposits 4020
    - Contingent liabilities 1010
  - Behavioral inflows — General description:
    - New FX swaps: The magnitude of new FX swaps is limited to the initial position.
    - Other new derivatives: The magnitude of new other new derivatives is limited to the initial position.
    - New debt issuance: Cannot issue any new long-term debt instrument.
    - New unsecured wholesale funding: No new unsecured wholesale funding.
  - Rollover rate (in percent of maturing funding):
    - Secured wholesale funding Up to 20Up to 20
    - New term deposits 8060
  - Haircut on liquid assets (in percent):
    - Domestic sovereign debt securities 5
    - Other debt securities 20
    - Equity securities 50
    - Other securtiies 75
- Cash flow methodology:
  - Tests based on cash flow analysis; assessed banks’ ability to maintain positive net cash positions over the 30-day horizon.
  - Did not account for possible spikes in funding costs or link distress to solvency conditions.
  - Banks allowed to raise liquidity via sale of unencumbered securities, use excess reserves (as BOK deposits), and access BOK liquidity facilities.
  - For potential margin calls, exchange rates assumed to follow the worst month-to-month depreciation of the won since 1990.
  - New FX swaps and other derivatives limited to initial positions as of March 29, 2013 (rollover allowed but position cannot increase).

### BU findings — resilience and system net cash positions
- All seven nationwide commercial banks would have adequate liquid assets to withstand a 30-day freeze in wholesale funding and deposit withdrawals, with no liquidity shortages for both local and foreign currencies separately.
- System-wide net cash positions after the pre-specified shock:
  - Local currency: KRW 34 trillion
  - Foreign currency: KRW 9 trillion
- Under a repeat of each bank’s worst historical deposit outflow:
  - Net cash positions: KRW 57 and KRW 11 in local currency and foreign currency, respectively.
- Typical funding responses:
  - To meet local currency funding needs banks rely roughly equally on sale of liquid assets, renewal of term deposits, and inflows from maturing investment.
  - To meet foreign currency funding needs banks rely largely on inflows from maturing investments.

### Non-bank financial institutions (NBFIs / NBDIs) — role, stress-test framework, and results
- Structure and importance:
  - Non-bank depository institutions (NBDIs) include credit cooperatives, merchant banks, mutual savings banks and postal savings institution — they account for 14.3 percent of financial sector assets at end-2012.
  - Non-depository financial institutions include insurance companies, asset management companies, securities companies, and credit specialized companies — accounting for 37.5 percent of total assets.
  - The share of NBFIs in total financial sector assets increased to 52 percent at end-2012 from 41 percent at end-2000.
- Risk characteristics:
  - NBDIs are not closely interlinked with the rest of the financial system, but their common exposure to macro risk factors and large collective size can pose credit risks.
  - Some larger NBDIs lend to the same pool of borrowers from which they draw deposits; NBDIs also tend to lend to lower-income households, increasing exposure to adverse macro shocks.
- Reverse stress-test approach and conservative assumptions:
  - Pre-loss net income based on the lowest level of pre-loss net income to total net income since 2007, adjusted for total assets as of end-2012.
  - Collateral value subject to a 20 percent discount; only 70 percent of this discounted value recognizable for provisioning.
  - Required provisioning: 20 percent for secured loans and 100 percent for unsecured loans (more stringent than regulation).
  - Stress test conducted at aggregate level for each type of NBDI (mutual savings banks, credit unions, mutual banking entities).
- Key result:
  - Certain NBDIs have relatively thin capital buffers. For example, mutual savings banks, credit unions, and fishery cooperatives can only handle up to around 10 percentage point increase in NPLs before their capital drops below the minimum required threshold (Table 4).
  - Even NBDIs that can withstand larger NPL increases should be monitored closely given generally very low regulatory minimum capital requirements.
  - Potential spillovers from NBDI difficulties to other parts of the financial sector are likely to be minimal because NBDI funding is mostly in deposits.

### Further considerations and policy recommendations (process and regulatory enhancements)
- Strengthen stress-testing process and inter-agency coordination:
  - Financial Supervisory Service (FSS) should maintain closer interaction with banks to ensure BU stress testing results are fully validated; BU results showed discrepancies across banks related to data and methodologies (e.g., loss estimation) that would benefit from further validation.
  - FSS and Bank of Korea (BOK) should enhance cross-sharing of TD and BU stress testing results to eliminate discrepancies (e.g., divergent market loss results), establish more consistent views on banking sector risks, and better inform policy actions.
  - Ideally, stress tests conducted by different agencies should be made public to encourage closer coordination.
- Banks’ internal risk management:
  - Banks would benefit from applying stress test scenarios distinct from common FSS-provided scenarios; banks comply with regulatory requirements and follow FSS scenarios but do not appear to run event-specific tests tailored to their operational needs.
- NBFI regulatory framework:
  - NBFIs’ resilience can be strengthened by applying a regulatory and supervisory framework consistent with that for banks.
  - Specifically, the Basel capital adequacy framework should be applied to all NBDIs, with larger entities of an equivalent size to banks subjected to stricter supervision.

*Source: IMF staff and Financial Supervisory Service (content unit: _cr1506 - 30. FX liquidity stress tests suggest that even in the event of unprecedented liquidity).*

### Appendix III. Solvency Test Assumptions

### Appendix III. Solvency Test Assumptions

### Key Solvency Stress Test Assumptions (Table AIII.1)
- Risk Factors
  - Credit losses: Based on satellite models, using various macro-financial variables.
  - Market losses: Based on changes in asset prices.
- Profitability
  - Profits are based on satellite models, using various macro and bank-specific variables.
- Off-balance sheet positions
  - Off-balance sheet positions are included in the stress tests.
- Risk Measurement
  - PDs and LGDs: Point in time
- Regulatory Standards
  - Hurdle rates: Solvency to be assessed in accordance with the hurdle rates in the graduated Basel III schedule, taking into account the capital conservation buffer, expected to be introduced in the later part of the stress testing (five-year) horizon.
  - Capital definition: Based on Basel 2.5 and III.
- Risk-weighted assets
  - Risk-weighted assets are adjusted in line with PDs and balance sheet growth.
- Behavioral Adjustments
  - Dividend payout: Zero if income is negative or capital falls below regulatory limit; else latest actual dividend payout ratio.
  - Balance sheet growth: Identical to the projected credit growth, which is projected by the BOK’s macroeconomic model.
  - Asset disposal: Not permitted.
  - Income composition: Assumed to remain constant.

### Additional note
- Content appears in the context of the Republic of Korea financial sector stress testing.

---

### Appendix IV. Stress Test Matrices

### Banking Sector: Solvency Risk (Table AIV.1)

- Institutions included
  - Bottom-up by Banks: All nationwide commercial banks (7).
  - Top-down by Authorities: All nationwide commercial banks (7); all other commercial banks (on an aggregate basis).
- Market share
  - Bottom-up: 89 percent of commercial bank assets.
  - Top-down: 60 percent of banking sector assets.
  - Top-down: 40 percent of broad financial sector assets.
  - Top-down: 100 percent of assets of commercial banks.
- Data and baseline date
  - Bottom-up: Banks’ internal data as of December 2012. Bank consolidated.
  - Top-down: Bank-by-bank supervisory data as of December 2012. Bank consolidated.
- Methodology
  - Bottom-up: Banks’ internal models.
  - Top-down: BOK’s model (SAMP).
- Stress test horizon
  - Five years up to December 2017. (Top-down: Five years up to December 2017 (instantaneous for sensitivity test).)
- Shocks (Scenario analysis; all are macro scenarios, stressing asset prices and macroeconomic variables)
  - Baseline: April 2013 World Economic Outlook (WEO); real GDP growth rate is 2.8 ppts for 2013 and 3.9 ppts for 2014.
  - A mild global double-dip: A GDP growth rate shock of 1 standard deviation (-3.3 ppts) from baseline in the first two years of a five-year horizon, with linear adjustment back to baseline in the third year. Growth rates at -0.5 percent (2013) and 0.6 percent (2014).
  - A severe global double-dip: A GDP growth rate shock of 2 standard deviation (-6.6 ppts) from baseline in the first two years of a five-year horizon, with linear adjustment back to baseline in the third and fourth years after the shock. Growth rates at -3.8 percent (2013) and -2.7 percent (2014). Standard deviations for both scenarios are based on 1990-2012 data.
  - Prolonged slowdown: A cumulative decline of real GDP by 1 standard deviation from baseline over five years (-2.5 ppts/year). This shock is estimated based on the five-year cumulative growth rate between 1990 and 2012.
  - An increase in interest rates: A 250 bps cumulative increase of domestic interest rates over 2½ years, starting in the second half of 2014, as a result of the advanced countries’ potential exit from unconventional monetary policy.
- Sensitivity Analysis (Single-factor analysis)
  - Interest rate-related credit risk: instantaneous impact of an interest rate spike (extreme shock: 300 bps; moderate: 200 bps).
  - Exchange rate depreciation (extreme shock: 50 percent; moderate: 30 percent).
  - A fall in real house prices (extreme shock: 20 percent; moderate: 15 percent).
- Risks/factors assessed
  - Credit risk of loan portfolios by type of exposure (large corporations, SMEs, residential mortgage, credit cards, and others).
  - Market risk of trading and AFS securities, and off-balance sheet exposure.
  - All macro scenarios assess both credit and market risks.
- Risk factor assumptions
  - Credit losses are based on satellite models, using various macro/financial variables.
  - Mark-to-market valuation of trading and AFS securities.
  - Profitability is based on satellite models, using various macro and bank-specific variables.
  - Off-balance sheet positions are taken into account.
  - Contagion effects are incorporated through the channels of funding costs, credit crunch, and fire-sale of assets.
- Behavioral Adjustments
  - Net interest income estimated based on banks’ internal model (bottom-up).
  - Net interest income estimated based on the authorities’ model (top-down).
  - Balance sheet growth identical to projected credit growth.
  - A constant share of non-loan assets in total assets based on average long-term growth.
  - Asset disposal not permitted.
  - Income composition assumed to remain constant.
  - Risk-weighted Assets (RWA) assumed to be adjusted in line with PDs (IRB approach).
  - Dividend payout ratio assumed to be zero if income is negative or capital falls below regulatory limit; else latest actual dividend payout ratio.
- Regulatory standards
  - RWA as per Basel 2.5 and III.
  - Hurdle rate based on Basel III schedule. Relevant ratios include tier-1 and total capital ratio (8 percent).
  - Capital definition per Basel 2.5 and III.
- Reporting format to the FSAP team
  - Bottom-up: Post-shock solvency ratios and losses by type of exposure (individual institutions).
  - Top-down: Distribution of capital ratios across the commercial banking system.
  - Top-down: Post-shock solvency ratios and losses by type of exposure (7 nationwide banks individually, aggregate by type of institutions).
- Note
  - The authorities did not provide the supervisory data. The FSAP team was unable to carry out independent stress test due to the unavailability of publicly available data.

### Banking Sector: Liquidity Risk (Table AIV.2)

- Institutions included
  - Bottom-Up: All nationwide commercial banks (7).
  - Top-Down: All nationwide commercial banks (7); all other commercial banks (on an aggregate basis).
- Market share
  - Bottom-Up: 89 percent of commercial bank assets.
  - Top-Down: 60 percent of banking sector assets.
  - Top-Down: 40 percent of broad financial sector assets.
  - Top-Down: 100 percent of the assets of commercial banks.
- Data and baseline date
  - Internal bank data as of March 2013. Bank consolidated.
- Risk factors
  - KRW and FX funding risk (deposit withdrawal and market freeze; uniform shocks across banks, independent of solvency tests).
  - Focus on access to funding, rather than funding cost.
- Test horizon
  - 30 days.
- Methodology
  - Bottom-Up: Cash flow analysis, based on internal models.
  - Withdrawal rates are applied to deposits, wholesale funding, derivatives, and committed credit lines.
  - Banks are assumed to be able to raise additional liquidity via the sale of unencumbered securities (with haircuts) and use of excess reserve deposits.
  - Top-Down: Implied cash flow analysis.
  - Stress tests incorporate wholesale funding difficulties and deposit withdrawals (funding risk), and fire sales of assets to meet liquidity constraints (market liquidity risk). Asset-specific haircuts are assumed.
  - Assessment of the risk of inability to rollover funding.
- Shocks
  - The magnitude of the wholesale funding shocks is in line with the severe liquidity difficulties experienced by banks globally after the Lehman collapse. The deposit withdrawal rates are more severe than the historical experience in Korea.
  - Disruption of 100 percent of unsecured short-term wholesale funding; 80 percent of secured wholesale funding (both KRW and FX); outflows of 20 percent of maturing retail term deposits (40 percent for FX), 40 percent of retail demand deposits (20 percent for FX), and 10 percent of contingent liabilities (both KRW and FX).
  - Disruption of 100 percent of new unsecured short-term wholesale funding; 80 percent of secured wholesale funding (both KRW and FX); outflows of 6 percent of retail term deposits (30 percent for FX), and 40 percent of retail demand deposits (20 percent for FX).
  - Severance of committed credit lines: at 0 and 100 percent.
- Assessment criteria
  - Maintaining no liquidity shortfall.
- Reporting format to the FSAP team
  - Liquidity shortfall, and inflows and outflows by type of funding (individual institutions); domestic and foreign currency separately.
- Note
  - The authorities did not provide supervisory data. The FSAP team was unable to carry out independent stress test due to the unavailability of publicly available data.

### Non-bank Depository Institutions: Solvency Risk (Table AIV.3)

- Institutions included (Top-down by Authorities)
  - All mutual savings banks; all credit unions; all agricultural cooperatives; all fisheries cooperatives; and all forestry cooperatives (on an aggregate basis of each type).
- Market share
  - 69 percent of the assets of non-bank depository institutions.
  - 19 percent of the assets of non-bank financial institutions.
  - 10 percent of the assets of all financial institutions.
- Data and baseline date
  - Data as of December 2012.
- Methodology
  - Reverse stress test.
  - An ad hoc increase in NPLs.
- Shocks
  - A 10 percent increase in NPLs.
- Assessment criteria
  - Maximum magnitude of NPL shock that would account for a drop of capital adequacy below minimum regulatory level.
- Reporting format to the FSAP team
  - Post-shock solvency ratios and capital shortages.
- Note
  - The authorities did not provide the supervisory data. The FSAP team was unable to carry out independent stress test due to the unavailability of publicly available data.

*Source: _cr1506 - Appendix III. Solvency Test Assumptions (Appendices III and IV).*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/scr/2015/_cr1506.pdf_