## 1korea2020006 ## Source details **Canonical URL:** [1korea2020006](https://www.imf.org/-/media/files/publications/cr/2020/english/1korea2020006.pdf) ## Other formats - [Markdown version](/-/media/files/publications/cr/2020/english/1korea2020006.pdf.md) - [Structured JSON version](/-/media/files/publications/cr/2020/english/1korea2020006.pdf.json) --- ### EXECUTIVE SUMMARY — Overview and context - Systemic risk analysis part of the 2019 Korea FSAP comprising: - forward-looking solvency analysis for banks, insurers, and pension funds; - a liquidity stress test for banks; - an assessment of network and interconnectedness across financial sector entities and the real economy. - Structural characteristics: - Korea’s financial system has grown by 40 percentage points of GDP since 2013. - Mortgage insurance schemes are widely used and influenced the bank risk assessment approach. - Life and non-life insurance sector is large, highly concentrated, and saturated. - Fintech developments (Open Banking system and e-money providers) are accelerating. - Demographics among the most adverse world-wide, implying continuous drag on demand, downward pressure on interest rates, financial firms’ income, and capitalization unless business models change. - Adverse macro-financial scenario: - Scenario judged severe enough to encapsulate a COVID-19 implied fallout; deep downturn spans over two full years before normalizing. ### BANK SOLVENCY ANALYSIS — Key findings and results - System resilience and heterogeneity: - The Korean banking system is broadly resilient to the FSAP severe adverse economic downturn scenario. - Performance under the downturn is heterogeneous: - Specialized banks exhibit the most pronounced drop in capital ratios. - ODIs (including Mutual Savings Banks and Credit Cooperatives) show the most dispersed capital ratio outcomes. - None of the nation-wide, regional, and specialized banks are found to fall short of regulatory capital minima. - Aggregate outcomes and losses: - System-wide capital depletion from the starting point amounts to about 22 percent of initial capital levels. - This represents about 2.9 percent of GDP as of 2018. - Drivers of pressure: - Credit risk losses and diminished net interest income are the most pronounced sources of pressure on banks’ capital ratios. - Mortgage insurance shielding: - Insured portion of mortgage portfolios (exposure-weighted medians at end-2018): nation-wide 26 percent, regional 45 percent, specialized 27 percent (assumed constant along horizon). - Counterfactual (insurance switched off) weighted average impact per bank group on CET1/RWA: nation-wide banks 21 basis points, regional banks 28 basis points, specialized banks 7 basis points; about 70 percent of impacts stem from the RWA effect. - Fintech Overlay impacts: - Fintech-induced competition raises system-wide capital depletion to 4 percent, from 3 percent without the overlay. - Regional banks’ capital ratios may fall by 0.6 to 1.3 percentage points by end of the five-year horizon under Fintech pressures. - Aggregate consumption of CCBs when overlay activated: nation-wide 0.4 percentage points, regional 0.7 percentage points, specialized 2.2 percentage points. - Nonlinear solvency→funding-cost feedback: - Nonlinearity estimated; example for regional banks: a 1p.p. shift to the capital ratio starting at 1 percent (4 percent) would induce a -0.25 p.p. (-0.1 p.p.) response of their cost of debt. - For eight banks, the additional feedback explains between -0.3 and -0.7 percentage points in the fall of their regulatory capital ratios. ### BANK LIQUIDITY ANALYSIS — Key findings and stress-test diagnostics - Coverage and sample: - Top-down LST based on 2018Q4 regulatory data; sample of 18 banks plus foreign bank branches; coverage ≈ 90 percent of banking system assets. - Current liquidity posture: - LCRs stand above the recently introduced requirement of 100 percent (60 percent for foreign bank branches) and FX LCR requirement of 80 percent in 2019; most banks met these in 2018. - NSFR: all banks subject to NSFR met the 100 percent minimum regulatory threshold in 2019Q1. - Stress-test design and calibration: - Four scenarios (asset price, retail funding, wholesale funding, extreme combined); assessed in KRW, FX, and both combined; hurdle rate 100 percent. - Liquidity shocks calibrated more severely than historical Korea experience (e.g., LCR assumed runoff 40 percent on unsecured non-operating deposits vs LST 50 percent). - Results and vulnerabilities: - Overall system shortfall in KRW would reach 2 percent of initial counterbalancing capacity. - Nation-wide banks would face a shortfall close to 9 percent of counterbalancing capacity. - Banking system’s FX liquidity shortfall from FX funding shocks would reach 4 percent of FX counterbalancing capacity; nation-wide banks 8 percent FX shortfall. - In extreme no-sharing scenario: aggregate KRW shortfalls = KRW 43tr; aggregate FX shortfalls = USD 3.5bn (less than 1 percent of BOK FX reserves). - Specialized banks’ reliance on unsecured wholesale funding leads to FX liquidity coverage falling from 112 percent to 85 percent under wholesale funding stress calibration. - Internet-only banks: KRW coverage ratio falls from 284 percent to 140 percent under retail run-off shock. - Central bank capacity: - BOK has ample space to provide liquidity; estimated FX shortfall (USD 3.5bn) < 1 percent of BOK FX reserves. ### INSURANCE AND PENSION SECTOR — Stress test findings and vulnerabilities - Insurance sector coverage and design: - Top-down solvency stress test covered about 75 percent of the life and non-life insurance market (seven life and six non-life insurers). - Adverse scenario shocks included sovereign bond yields (up to +250 bps), equity prices (-39 percent), corporate spreads and defaults, FX, and mutual funds (-12.5 percent); shocks modeled as instantaneous haircuts. - Held-to-maturity designation shields around 23 percent of sample investment assets from market price changes. - Insurance stress-test outcomes: - Required capital assumed to decline marginally by 2 percent in the stress model. - Life insurers: - Starting median RBC coverage: 239 percent. - Median life company drops to 169 percent under stress. - Two insurers fall slightly below recommended level of 150 percent but remain above regulatory minimum 100 percent. - Overall reduction in available capital: KRW 27.4 trillion (a 38.7 decline, or 4.3 percent of total assets). - Equity shock: 35 percent of the KRW 27.4 trn reduction; corporate bond shock: 30 percent; sovereign yield shock: 18 percent; mutual fund shock: 15 percent. - Investment income in 2019 would be KRW 2.2 trillion (12 percent) lower than 2018 under instantaneous stress; net income before tax would drop to KRW 2.3 trn (-42 percent) in 2019. - Underwriting results muted; need to restructure toward lower guarantees and protection business. - Non-life insurers: - Median RBC coverage declines from 214 to 158 percent. - Available capital reduced by KRW 8.9 trillion (-27.8 percent, or -3.9 percent of assets). - Corporate bond shock contributes 50 percent of the reduction. - Pension funds and NPS: - NPS assets: USD 600bn (earlier section) / KRW 700 trn (later section — source presents both currency-denominated measures in different parts). - NPS projected growth to KRW 1,778 trn in 2041 when benefit payments will start exceeding contributions; fund depletion projected by 2057 under unchanged parameters. - NPS current asset concentration: domestic assets ≈ 60 percent; holds >7 percent of stock market capitalization and 18 percent of Korean sovereign bonds. - Alternative investments: 12 percent of assets (real estate 37 percent, private equity 32 percent, infrastructure 30 percent, hedge funds 2 percent). - Past performance: return -0.9 percent in 2018; average annual performance since inception 5.6 percent. ### NETWORK, INTERCONNECTEDNESS, AND CONTAGION — Key findings - Network analysis scope: - Cross-exposures for 64 large financial institutions; default simulation following Espinosa-Vega and Solé (2010) with LGD = 100 percent. - Main outcomes: - Specialized banks are about as systemic as nationwide banks (confirmed by centrality metrics). - Nationwide banks are net lenders to specialized banks and therefore more vulnerable to specialized bank stress than vice versa. - Insurers are net lenders to all other sub-segments and vulnerable to banking sector stress; insurers are not as systemically central. - Banks and insurers most hit in solvency stress tests are not tightly linked to rest of system; spillover potential appears contained at current juncture. - Recommendations on data and modeling: - Maintain and expand bilateral exposure dataset compiled by the FSS; collect exposures regularly (e.g., quarterly); develop dynamic network model methodologies. ### MACRO-FINANCIAL SCENARIO — Adverse scenario calibration and projections - Scenario configuration and horizon: - Adverse scenario assumed to start unfolding from 2019Q1 onward; horizon covers 20 quarters from 2019Q1-2023Q4. - Calibrated to match Growth-at-Risk (GaR) estimate: at 5 percent tail probability GaR ≈ -3.1 percent year-on-year for 2019. - Real GDP shock severity ≈ 2.1x standard deviation multiple (2-year cumulative) historically. - Comparison with IMF WEO (April 2020): - WEO April 2020 baseline for 2020: real GDP growth = -1.2 percent year-on-year. - FSAP adverse scenario for 2020: real GDP growth = -3.1 percent year-on-year. - Unemployment: WEO April 2020 expects 4.5 percent; FSAP adverse expects 4.8 percent. - FSAP adverse judged sufficiently conservative as of end-April 2020 to encompass COVID-19 consequences. - Inflation and asset price projections: - CPI/deflator-based inflation moves up in first year from import-cost pass-through, then temporarily negative before recovery. - Nation-wide nominal house prices: cumulative drop of -16 percent between 2018Q4 and 2021Q2. - Equity prices: drop of -39 percent from 2018Q4 to 2020Q2 (trough), before recovery. ### MODELING APPROACHES — Solvency and credit modules (methodology highlights) - Core features of Solvency Analysis Tool Suite: - i) account for IFRS 9 loan loss provision principles; - ii) nonlinear solvency→funding-cost feedback mechanism; - iii) dynamic balance sheets; - iv) a “Fintech Overlay” capturing competitive and liquidity dynamics. - Credit Risk (CR) Module: - Hybrid PD–satellite–Z-factor methodology: - PD-satellite models estimated using Bayesian Model Averaging (BMA) for four portfolio segments: large NFCs, SMEs, household mortgages (including Jeonse), consumer credit. - Z-factor methodology calibrated from historical transition matrices (2010Q1-2018Q4) to map to transition matrices and default rates. - Bank-portfolio-specific Z path for 20-quarter forward horizon set to match exposure-weighted PD path implied by satellite models. - IFRS 9 mapping: - Prudential “normal” → Stage 1; “precautionary” → Stage 2; “substandard”, “doubtful”, “loss” → Stage 3. - Perfect foresight assumption used: baseline and adverse scenarios considered separately with 100 percent weight each (multi-scenario probability-weighting not used). - LGD and RWA treatment: - Structural LGD model links housing-collateralized portfolios’ LGDs to house price paths for residential real estate including Jeonse loans. - Other portfolios: LGDs kept constant in baseline and scaled by 1.15 under adverse scenario. - Conservative overlay: take maximum of model-implied LGDs and bank-reported regulatory downturn LGDs. - RWs for STA exposures kept constant at end-2018; RWs for IRB exposures allowed to move dynamically as TTC-PDs change. - Pass-through smoothing parameter α set to 0.5 in baseline (sensitivity range 0–1). - Interest income/expense models: - Panel econometric models estimated separately for nation-wide, regional, and specialized banks; include feedback from capital ratios to funding cost (nonlinear). - IIR and IER definitions and key coefficients preserved in Appendix V. - Dynamic balance sheets: - Stock evolution formulas define S1, S2, S3 stocks; portfolio gross loan growth gt scenario-determined and equal across banks. - Write-off and asset-sale parameters held at observed end-sample positions; allowing write-offs and sales materially affects solvency outcomes. - Static vs dynamic balance sheet comparison: static outcomes more benign; median differences to full dynamic balance sheet outcome: nation-wide 28 basis points, regional/ODIs 46 basis points, specialized 50 basis points. ### HOUSING, JEONSE, AND PUBLIC FOOTPRINT — Key observations - Financial system size and housing: - Total assets of financial institutions ≈ 300 percent of GDP as of 2018Q4. - Since 2013 FSAP, system grew by ≈ 40 percentage points of GDP. - Jeonse leasehold market: - Jeonse deposits average about 50-70 percent of house price. - Tenants finance Jeonse deposits via Jeonse loans, which doubled to KRW 72.2tr since 2014. - Loan-to-deposit ratios for Jeonse loans increased to 42 percent in 2018. - BOK assessment: following a 20 percent fall in Jeonse deposit price (and 13 percent house price fall), about 78 percent of landlords could cover leasehold deposits with own financial assets; 22 percent would need an additional loan. - Public-sector involvement: - HF securitizes non-subsidized long-term mortgages; total MBSs issued and guaranteed amount to 116tr KRW (large portion held by insurers). - Loans, securities and guarantees issued/provided by KHFC in 2017 ≈ 7 percent of GDP. - KAMCO acquired about USD 100bn of bad loans after Asian crisis and sizable NPL purchases after subsequent crises. - House price sensitivity: - Hypothetical -40 percent house price decline combined with adverse scenario: banking system appears resilient; additional burden on capital ratios limited (mortgage portfolios LGD effects only; commercial real estate not captured). ### FINTECH — Channels of risk and recommended workstreams - Channels and risks: - Open Banking and e-money developments change liquidity dynamics in retail payments and money markets. - Mechanisms: swift deposit transfers across banks via mobile apps; e-money providers offering remuneration resembling interest; higher-frequency deposit mobility enabling “electronic deposit runs”; concentration risk if e-money providers re-deposit funds with a few banks. - Current flow constraints: Open Banking System transfer cap KRW 10 million; e-money transfers cap KRW 3 million — may not prevent gradual outflows. - Policy recommendation: - Conduct impact assessment related to Fintech developments focusing on systemic liquidity, competition (solvency impact), and system-wide operational risks. Timing: I (Immediate). ### DEMOGRAPHICS — Long-term implications and scenario-conditional forecasts - Demographic trends: - By 2050 almost 40 percent of population expected to be older than 65 (up from 13 percent now). - Working-age population will shrink significantly. - Old-age poverty almost 44 percent for those aged 66 or older (OECD average 14 percent). - Macro-financial implications: - Structural drag on natural interest rates and inflation; downward pressure on demand and net interest income for banks. - Long-term effects on housing demand and Jeonse system viability. - Pension systems face sustainability challenges; second and third pillars need strengthening. - Demographics scenario-conditional forecasts (2019–2040): - All financial sector components except pension funds expected to shrink or move side-wards in nominal asset size. - Pension funds expected to grow as rising old-age dependency dominates. - Banks expected to move sideward with small cross-scenario dispersion. - Pension fund asset growth trajectories depend most on fertility assumptions. - Policy implications: - Monetary policy frameworks should explicitly account for population dynamics. - Fiscal policy responses to demographics require long lead times and pre-emptive action. ### POLICY RECOMMENDATIONS — High-level and timing (selected) - Systemic Risk Analysis: - Conduct detailed Fintech impact assessment (Open Banking, e-money): Timing: I. - Further develop models to reflect Korea real estate structures including Jeonse to assess rollover risk and connectedness: Timing: ST. - Categorize ODIs as banks (not non-banks) for risk analyses: Timing: MT. - Solvency Analysis and Stress Testing: - Develop capacity to capture loan insurer structures (HF, SGI) and implied nonlinear system behavior: Timing: MT. - Account for banks’ continuous selling of NPLs to NPL market in solvency stress tests: Timing: MT. - Develop lifetime ECL models for K-IFRS 1109-compatible top-down stress testing: Timing: MT. - Broaden insurance stress-test scenarios and use stress-test results post-K-ICS for supervisory intensity/capital add-ons: Timing: ST/MT. - Liquidity Analysis and Stress Testing: - Improve bank liquidity stress test for domestic currency and develop FX liquidity stress test by major currencies: Timing: ST. - Expand joint BOK–FSS FX liquidity stress test methodology to include FX LCRs: Timing: MT. - Supervisory and model development priorities: - Invest in back-testing top-down model components, especially LGD models. - Develop dynamic balance sheet features and explicitly account for asset sales and write-offs. - Further develop stress test models for credit guarantee institutions and integrate guarantor stress tests with banks and other FIs. - Treat Korea’s ODI sector as part of the banking system in risk analyses (ODIs ≈ 20 percent of banking system). *Italic: IMF staff.* ### EXECUTIVE SUMMARY _________________________________________________________________________ 7 ### EXECUTIVE SUMMARY ### Overview and context - The systemic risk analysis was conducted as part of the 2019 Korea FSAP and comprises: - forward-looking solvency analysis for banks, insurers, and pension funds; - a liquidity stress test for banks; - an assessment of network and interconnectedness across financial sector entities and the real economy. - Structural characteristics informing the analysis: - Korea’s financial system has grown by 40 percentage points of GDP since 2013. - Mortgage insurance schemes are widely used and influenced the bank risk assessment approach. - The life and non-life insurance sector is large, highly concentrated, and saturated. - Fintech developments (Open Banking system and e-money providers) are accelerating. - Demographic developments are among the most adverse world-wide, implying continuous drag on demand, downward pressure on interest rates, financial firms’ income, and capitalization unless business models change. - The FSAP adverse macro-financial scenario is judged severe enough to encapsulate a COVID-19 implied fallout on economic activity; the scenario’s deep downturn spans over two full years before normalizing. ### Bank solvency analysis — key findings - The Korean banking system is broadly resilient to the FSAP severe adverse economic downturn scenario. - Performance under the downturn is heterogeneous: - Specialized banks exhibit the most pronounced drop in capital ratios. - ODIs (including Mutual Savings Banks and Credit Cooperatives) show the most dispersed capital ratio outcomes. - None of the nation-wide, regional, and specialized banks are found to fall short of regulatory capital minima. - Fintech-implied rising competition can have notable implications for: - systemic liquidity; - implied solvency across institutions; - system-wide operational risks. ### Bank liquidity analysis — key findings - Measures have been taken to strengthen banking system liquidity; banks have ample room to withstand liquidity shocks. - Overall, the banking system is likely to maintain adequate liquidity in both domestic and foreign currency following hypothetical shocks to asset prices, retail funding, and wholesale funding. - State-owned banks’ reliance on unsecured wholesale funding would lead to a fall in FX liquidity coverage from 112 percent to 85 percent under the stress calibration. ### Insurance and pension sector analysis — key findings - The top-down solvency stress test covered about 75 percent of the life and non-life insurance market. - Insurance firms’ capital would decline substantially under the stress scenario, but all firms would remain above regulatory thresholds. - Under the current accounting regime, held-to-maturity designation is widely used among life insurers, shielding a significant portion of investments from market price changes. - Life insurers would experience a significant decline in income, indicating the need to: - restructure business to restore underwriting profitability; and - shift further from guaranteed savings products toward lower guarantees and protection business. ### Network and interconnectedness — key findings - Specialized banks are as systemic as nation-wide commercial banks. - Nation-wide banks are net lenders to specialized banks and therefore more vulnerable to specialized bank stress than vice versa. - Insurers are not as systemically central but are net lenders to all other financial sub-segments and thus vulnerable to stress in the banking system. ### Policy recommendations (high-level and timing) - Systemic Risk Analysis - Conduct an impact assessment related to Fintech developments (Open Banking, e-money developments) focusing on systemic liquidity, competition (solvency impact), and system-wide operational risks. Timing: I - Further develop risk assessment models to reflect real estate market structures in Korea, including the Jeonse leasehold market system, to assess rollover risk and connectedness. Timing: ST - Categorize ODIs (including Credit Unions, Mutual Savings Banks, and Credit Cooperatives) as banks, not non-banks, to avoid misleading inferences about banking system dynamics. Timing: MT - Solvency Analysis and Stress Testing - Further develop capacity for capturing loan insurer structures (e.g., HF, SGI) and their implied nonlinear system behavior under downturns. Timing: MT - Develop solvency stress test models to account for banks’ continuous selling of NPLs to the NPL market. Timing: MT - Develop lifetime expected credit loss models to allow K-IFRS 1109-compatible top-down stress testing to gauge provision adequacy. Timing: MT - Broaden range of insurance stress test scenarios based on regularly updated medium-term risk outlook. Timing: ST - After implementation of the new insurance capital standard (K-ICS), use stress test results as an input to supervisory intensity and possible capital add-ons. Timing: MT - Liquidity Analysis and Stress Testing - Continue to improve bank liquidity stress test framework for domestic currency and develop FX liquidity stress test by major currencies. Timing: ST - Expand the methodology of the joint BOK and FSS FX liquidity stress test to include FX LCRs. Timing: MT *Source: IMF staff.* ### 1. As of 2018Q4, total assets of financial ### 1korea2020006 - 1. As of 2018Q4, total assets of financial ### Financial system size and macro-structure - Total assets of financial institutions in Korea reached about 300 percent of GDP as of 2018Q4. - Since the previous 2013 FSAP, Korea’s financial system has grown by about 40 percentage points of GDP. - Real estate is the central asset class where leverage is high. - The asset management industry and second-pillar pension funds are less developed than other segments but are becoming increasingly important amid a material demographic transition. - Growth in Fintech is adding to changing dynamics. - Growth in the onshore FX derivatives market lags cross-border investment flows; the offshore NDF market in Korean won remains large and growing. - Selected indicators of the Korean financial system structure are presented in Figure 1. ### Banking sector composition and state-owned entities - The banking sector comprises nation-wide and regional commercial banks, specialized (policy) banks and Other Depository Institutions (ODIs). - State-owned financial firms are prominent participants. - The KHFC provides mortgage insurance and issues fully guaranteed mortgage-backed securities composed of “conforming” loans (a conforming loan is a long-term, fixed-rate mortgage loan provided by the KHFC). - Loans, securities and guarantees issued and provided by KHFC in 2017 amounted to about 7 percent of GDP. - KAMCO acquired about USD 100bn of bad loans after the Asian crisis and has bought substantial amounts of NPLs after subsequent crises (credit card crisis 2003, global financial crisis 2008, household debt crisis 2013). - The National Pension Service (NPS) has USD 600bn in assets and is the largest investor in Korea. - The Korean Investment Corporation (KIC) is a sovereign wealth fund with about USD 130bn assets under management. ### Financial holding companies (FHCs) - As of end-2018, nine FHCs held about 40 percent of total financial institutions’ assets (114 percent of GDP). - Four FHCs have been identified as D-SIBs. - Financial firms belonging to bank holding companies operate across insurance, capital markets and asset management businesses. ### Banks’ business models and funding - Banks, including ODIs, are primarily funded by retail deposits. - Assets are concentrated in loans, mostly related to real estate. - Bank loans split about equally across lending to households and firms. - 80 percent of the stock of corporate bank loans outstanding is to SMEs. - Most SME loans are collateralized against real estate. - Commercial banks have a relatively diversified loan portfolio and securities’ holdings. - State-owned banks are focused on lending to SMEs. ### Insurance sector structure and role - The insurance sector comprises 24 life insurance firms and 30 non-life insurers, managing KRW 1170 trillion in total. - Insurance penetration (premiums to GDP) is one of the highest in the world, exceeded only by Taiwan Province of China and Hong Kong SAR. - Life insurance assets account for more than 20 percent of household financial assets. - Bancassurance market: 50 percent of new business in life is sold via banks. - Market concentration: top firms owned by FHCs and chaebols dominate the sector. ### Capital markets, asset management, and bonds - The equity market has a market capitalization of around USD 1.8 trillion with foreigners holding around 35 percent of the listed Korean stocks. - The largest five conglomerates account for over 50 percent of market capitalization. - Korea exhibits a “Korea discount” with notably lower price-to-book and price-to-earnings ratios relative to major advanced and emerging market economies. - The bond market is dominated by government and other public debt; ‘green’ or ESG bonds’ importance is rising. - Privately placed funds and derivatives-linked products grew at double digit rates over the past year and now amount to about KRW 500 trillion; roughly 30 percent of GDP. ### Fintech and retail payments - Fintech innovations are rapidly taking root in the retail payments service sector and altering the market structure. - Korea features great financial institution depth, a saturated credit card market, high smartphone penetration, and customer eagerness to adopt new technologies—fertile for mobile payment services. - Entry of large technology companies in payment services (stored value payment products and payments-initiation interfaces) is beginning to alter market structure and competition. - Authorities are facilitating innovation via a work plan including the introduction of an “Open Banking” system and loosening legal restrictions for electronic financial transactions and use of personal data. ### Bank asset quality, capitalization, and profitability - Banks’ aggregate capital ratio is about 16 percent of risk-weighted assets as of the post-2013 assessment period. - NPL ratios are structurally low but should be interpreted cautiously because banks sell NPLs swiftly into an active market for distressed assets. - Korean banks underperform regional peers in terms of profitability. - The share of foreign business in total assets is about 6 percent, up from 4 percent in 2013. - Nation-wide banks’ largest exposure to China amounted to 1.7 percent of total assets in 2018 (1.1 percent in 2013). - Nation-wide banks’ asset exposures are diversified; regional banks’ and mutual savings banks’ exposures are concentrated in SME lending; ODIs’ lending is concentrated in consumer credit. ### Insurer capitalization, profitability, and challenges - Median life insurers recorded a coverage of risk-based capital (RBC) at end-2018 of 239 percent (regulatory minimum 100 percent; recommended level 150 percent). - Non-life sector RBC coverage was 263 percent at end-2018—the highest level since 2012. - Sector return on assets stood at 0.3 percent in 2018. - Investment yields declined from 5.1 percent in 2012 to 3.4 percent in 2018. - Insurers have extended asset durations (including through foreign investments) and shifted toward offering more ‘protection’ instead of savings products. - Non-life sector: combined ratio hovered slightly above 100 percent in recent years; median return on assets in 2018 was 1.3 percent. - Profitability is impacted by underwriting losses, especially in health insurance where the government aims for lower cost of private coverage. - With unfavorable demographics and competition, growth prospects are limited, prompting foreign expansion and cooperation with fintech providers. *IMF staff summary based on the provided Korea financial system structure content.* ### 11. While well capitalized under the current regime, additional capital needs are expected ### 11. While well capitalized under the current regime, additional capital needs are expected ### New solvency regime and transition risks - The new solvency regime will be aligned with IFRS 17 and introduce a market-consistent valuation of assets and liabilities by 2022. - An initial quantitative impact study in 2018, based on the IAIS’ Insurance Capital Standards and Solvency II, showed a notable capital shortfall in the life insurance sector. - A more recent study in 2019 implies substantially smaller capital needs as several parameters were adjusted to local market conditions. - The transition has already triggered increased issuance of subordinated debt. - An increased re-classification of fixed-income assets from available-for-sale to held-to-maturity has been observed; these would need to be reverted when transitioning to a fully market-consistent asset valuation. - Most notably, the ultimate forward rate used in the interest rate term structure modeling for deriving the value of insurance liabilities was increased to 5.2 percent, which is significantly above the 3.9 percent used in the EU’s Solvency II regime for KRW-denominated liabilities. ### Insurance sector solvency and profitability (selected findings) - Life sector: median solvency ratio of 239 percent, above the 100 percent minimum requirement. - Non-life sector: median solvency ratio stands at 263 percent, the highest ratio since 2012. - Profitability: life insurance records weak profitability with returns on assets well below the non-life sector. - Investment returns have been steadily declining in recent years. - The combined ratio in non-life hovers above 100 percent, mainly driven by underwriting losses in health insurance. - Sample of insurers included in the stress test (end-2018, total assets and RBC Ratio): - Hanwha Life — 114.3 (KRW trillion), 212% - Kyobo Life — 101.5 (KRW trillion), 312% - NongHyup Life — 64.7 (KRW trillion), 195% - Orange Life — 32.7 (KRW trillion), 425% - Samsung Life — 262.2 (KRW trillion), 314% - Shinhan Life — 32.0 (KRW trillion), 239% - Tongyang Life — 31.9 (KRW trillion), 205% - DB Insurance — 39.8 (KRW trillion), 216% - Hyundai Marine&Fire — 43.7 (KRW trillion), 219% - KB Insurance — 33.7 (KRW trillion), 187% - Korean Re — 10.7 (KRW trillion), 211% - Meritz Fire&Marine — 20.4 (KRW trillion), 211% - Samsung Fire — 79.0 (KRW trillion), 334% ### Banking system capitalization and asset quality (high-level indicators) - Regulatory capital to risk weighted assets: 2012–2017 = 14.3, 14.5, 14.0, 13.9, 14.8, 15.2 - Regulatory Tier 1 capital to risk-weighted assets: 2012–2017 = 11.1, 11.4, 11.4, 11.7, 13.1, 13.1 - Non-performing loans net of provisions to capital: 2012–2017 = 3.1, 3.6, 2.8, 2.1, 2.1, 1.7 - Non-performing loans to total gross loans: 2012–2017 = 0.6, 0.6, 0.5, 0.5, 0.5, 0.4 - Return on assets: 2012–2017 = 0.7, 0.3, 0.5, 0.5, 1.9, 0.7 - Return on equity: 2012–2017 = 8.2, 4.0, 6.0, 6.6, 24.3, 8.5 - Liquid assets to total assets: 2012–2017 = 36.2, 33.8, 37.9, 34.4, 32.9, 29.9 - Liquid assets to short-term liabilities: 2012–2017 = 111.3, 107.9, 122.0, 105.0, 102.6, 101.2 - Net open position in FX to capital: 2012–2017 = 0.2, -0.2, 0.5, 0.0, 0.0, -1.0 - Sectoral distribution of loans (2012–2017 examples): Domestic residents 93.3, 92.5, 92.5, 91.0, 92.6, 92.6; Nonfinancial corporations 53.3, 50.7, 50.0, 48.7, 49.5, 46.1; Households 34.8, 36.5, 36.3, 36.3, 35.3, 39.9 - Additional FSIs (selected): Capital to assets (leverage ratio) 2012–2017 = 8.2, 8.3, 8.1, 8.0, 7.9, 8.0; Customer deposits to total non-interbank loans 2012–2017 = 84.1, 84.1, 81.6, 76.0, 71.2, 75.4; FX loans to total loans 2012–2017 = 13.2, 12.8, 13.4, 12.2, 11.7, 10.9 ### Macro-financial environment and systemic vulnerabilities - Korea is very open and closely integrated with international supply chains; export-oriented manufacturing accounts for about a quarter of GDP. - GDP growth was robust at around 3 percent in 2017 and 2018. - Growth is expected to remain sluggish due to trade tensions and low business confidence, despite policy easing. - The financial cycle is advanced with household debt among the highest for OECD countries. - Total non-financial private sector debt to GDP stands close to 200 percent. - Core debt (debt of the non-financial sector owed to banks) amounts to about 130 percent of GDP. - Credit-to-GDP gap is close to zero. - Household debt dynamics: - Household lending growth (primarily housing-related) has slowed but remains above nominal GDP growth. - Household debt as a ratio of disposable income stands at about 180 percent. - Demographics: - By 2050 almost 40 percent of the population is expected to be older than 65, up from 13 percent now. - The working-age population will shrink significantly. - Implications include a rising share of debt held by older households and rising DSTI for some cohorts; growing retirement savings invested abroad. - COVID-19 (assessment as of March 2020): - Rapid rise in cases since February 2020 likely implies a significant drag on economic activity. - Korean authorities proposed a supplementary budget of KRW 11.7trn (0.6 percent of GDP) to step up disease control and provide economic support. ### Housing market structure, Jeonse risks, and public sector footprint - House prices: - National nominal house and apartment price growth has moderated since the GFC; nation-wide median price-to-disposable income ratios have trended upward since 2014. - Model-based valuation points to overvaluations in Seoul (10 percent) and the larger capital region (5 percent). - Jeonse (leasehold deposit) market: - Jeonse deposits average about 50-70 percent of the house price. - Tenants often finance Jeonse deposits via Jeonse loans, which doubled to KRW 72.2trn since 2014. - Loan-to-deposit ratios for Jeonse loans increased to 42 percent in 2018. - The Bank of Korea assessment: following a 20 percent fall in the Jeonse deposit price (and a 13 percent house price fall), about 78 percent of landlords could cover total leasehold deposits with their own financial assets, while 22 percent would need an additional loan. - Public sector involvement: - Korean Housing-Finance Corporation (HF) securitizes non-subsidized long-term mortgages; total MBSs issued and guaranteed by the HF amount to 116tr KRW, a large portion held by insurance companies. - HF guarantees include loans using Jeonse deposits as collateral and loans for landlords to cover falls in Jeonse deposit prices. - HF also offers reverse mortgages to low-income pensioners. - There is lack of transparency regarding the public sector’s total contingent liabilities related to the housing market and adequacy of capital buffers. ### Supervisory scope and stress testing coverage - Banking sector stress test: - Analysis based on public and supervisory data for 24 banks, covering 95 percent of banking system assets at end-2018. Five of the 24 are sub-aggregates forming the ODI category. - Insurance sector stress test: - Seven life and six non-life insurers included. - Analysis covers 73 percent of life sector assets and 76 percent of non-life sector assets (end-2018). *Source: IMF staff compilation based on the content of the provided chapter.* ### 25. The risks to consider for the Korean economy stem primarily from escalating trade ### 25. The risks to consider for the Korean economy stem primarily from escalating trade ### Major risks and structural vulnerabilities - Primary triggers: - Escalating global trade tensions involving the US, China, the EU, and Japan. - Geopolitical risks related to North Korea despite stated commitment to denuclearization. - Structural macro-financial features increasing vulnerability: - Korea’s export orientation and limited export diversification. - Strongly intertwined network of interdependencies across firm conglomerates. - Rising competition from new financial firm types (Fintech), including the Open Banking initiative launched in December 2019 and growing e-money providers. - Adverse demographic developments. - Fintech-specific structural implications: - Changing liquidity dynamics in retail payments and money markets. - Implications for solvency through competitive pressure and for system-wide operational risks. ### Adverse macro-financial scenario (trade-tension-driven) — narrative and channels - Trigger and market reaction: - Scenario rooted in escalation of global trade tensions leading to a broad-based, world-wide sell-off in equity markets and a general fall in investors’ risk appetite. - Korea would face significant capital outflows and strong currency depreciation due to close trade ties with China. - Real economy transmission: - Falling export and import flows severely pressure export-oriented Korean firms. - Adverse spillovers to domestic demand via higher unemployment, drop in income, depressed confidence, and investment and consumption contraction (expectation channel). - Large firm conglomerates would be significantly disrupted due to cross-border supply-chain interconnectedness. - Monetary and financial channels: - Interest rate-based monetary policy would have less room to stimulate demand because interest rates are structurally lower at the outset than before the GFC. - Capital outflows would distress banks’ liquidity, particularly in FX, through funding and liquid asset valuation shocks and indirectly via NFCs export revenues. - FX liquidity risk also determined by channels through which Korean banks and NFCs access FX, including the interbank swap market and high interconnectedness of banks. ### Scenario calibration, horizon, and key quantitative projections - Scenario configuration: - Adverse scenario assumed to start unfolding from 2019Q1 onward. - Scenario horizon covers 20 quarters from 2019Q1-2023Q4. - Adverse scenario calibrated to match a Growth-at-Risk (GaR) estimate: at a 5 percent tail probability GaR ≈ -3.1 percent year-on-year for the year 2019. - Real GDP shock severity corresponds to an approximate 2.1x standard deviation multiple (2-year cumulative) in historical perspective. - Comparison with IMF WEO (April 2020): - IMF WEO April 2020 baseline for 2020: real GDP growth = -1.2 percent year-on-year. - FSAP adverse scenario for 2020: real GDP growth = -3.1 percent year-on-year. - Unemployment: WEO April 2020 expects 4.5 percent; FSAP adverse expects 4.8 percent. - Judgment: FSAP adverse scenario considered sufficiently conservative as of end-April 2020 to encompass adverse consequences of COVID-19. - Inflation and asset price projections under the adverse scenario: - Measures of inflation (deflator- and CPI-based) move up visibly in the first year due to import-cost pass-through from currency depreciation, then level off and temporarily move into negative territory before recovery in the second half of the horizon. - Nation-wide nominal house prices: cumulative drop of -16 percent between 2018Q4 and 2021Q2. - Equity prices: drop of -39 percent from 2018Q4 to 2020Q2 (trough), before starting to recover. ### Historical context and GDP path comparisons - FSAP 2019 baseline is the weakest compared to AFC and GFC pre-crisis trends and the FSAP 2013 baseline. - FSAP scenarios (2013 and 2019) imply a more protracted path compared to AFC and GFC (more frontloading and faster recovery in AFC/GFC). - Cumulative real GDP deviations under FSAP adverse are less negative than FSAP 2013 deviations; justified by weaker baseline and GaR benchmark estimates at a 5 percent probability level. - Figures and temporal origins: - Scenario horizon charts and comparisons reference AFC = 1997/98, GFC = 2007-09. Origins “t0” denote [1997Q4, 2008Q3, 2012Q4, 2018Q4] for [AFC, GFC, FSAP 2013, FSAP 2019]. ### Fintech developments: channels of risk and features - Three dimensions of implications from Open Banking and e-money developments: - Liquidity dynamics in retail payments and money market have changed, with implications for bank solvency and system-wide operational risks. - Mechanisms of competitive pressure: - Open Banking may enable swift deposit transfers across banks via mobile applications displaying competitive pricing, pressuring banks’ interest expense. - Customers may seek more competitive loan offers, pressuring banks’ interest income. - Rising financial literacy may amplify customer mobility and sensitivity to rates. - E-money providers’ potential impacts: - May compete with commercial bank deposits by offering remuneration resembling interest (enabled by gradual de-regulation). - Banks may offer higher deposit rates to e-money providers to attract re-deposited customer funds, raising banks’ interest expense. - Higher-frequency liquidity dynamics and concentration risks: - Reduced stickiness of bank deposits could enable “electronic deposit runs” that exacerbate liquidity squeezes beyond fundamentals. - E-money providers re-depositing funds in one or few banks can create concentration risk (system-wide operational risk and structural shortage of liquid funds for banks not holding those funds). - Regulatory observations and current flow constraints: - E-money providers currently mitigate redemption risk by holding funds underlying e-money on a 1:1 basis. - Flow constraints: Open Banking System does not allow more than KRW 10 million to be transferred; no more than KRW 3 million through e-money service providers. Such constraints may not prevent gradual outflows over a sequence of days. - Policy recommendation: - The FSAP recommends Korean oversight institutions conduct a detailed Fintech and competition-related impact assessment for the banking sector and other components of the financial system, focusing on medium-term solvency implications, systemic liquidity, and system-wide operational risk implications. ### Forward-looking bank solvency analysis — methodology overview - Core features of the Solvency Analysis Tool Suite: - i) an account for IFRS 9 loan loss provision principles; - ii) a nonlinear solvency-funding cost feedback mechanism; - iii) dynamic balance sheets; - iv) a “Fintech Overlay” capturing competitive and liquidity dynamics from Fintech developments. - P&L structure and modeling approach: - The solvency analysis uses modular connected modules summarized in Figures 15–17 and adopts specific model approaches for each P&L line. - Credit Risk (CR) Module specifics: - Structure: “hybrid” PD-satellite-Z-factor methodology. - Default rate satellite models estimated using confidential bank data provided by the FSS and structured via a Bayesian Model Averaging (BMA) Methodology with sign constraints on long-run multipliers. - Default rate models developed for four portfolio segments: large nonfinancial corporates (NFCs), SMEs, household mortgage loans (including Jeonse leasehold loans), and consumer credit. - Default rate models used to derive forecasts conditional on baseline and adverse scenarios. - PDs at outset and maxima under adverse scenario are presented (annual horizon PDs; distribution statistics shown across banks with 90th/10th percentiles, maxima/minima, medians). - Household mortgage portfolio PDs were found very reactive to the adverse scenario in terms of multiples to their starting points (starting PDs being very small in comparison to other portfolios as of 2018); levels under the adverse scenario deemed sufficiently conservative in historical perspective. - Modeling and robustness: - A regime-switching structural vector autoregressive model framework used to quantify external demand shock deviations for Korea; base variant contains real GDP growth, GDP deflator inflation, house prices, interest rates, and the USD-KRW exchange rates. - Numerous model variants were considered (inclusion/exclusion of variables, different lag structures) to assess robustness. - Banking system-wide portfolio-specific default rate scenario paths attached to banks’ portfolio-specific starting points using a distance-to-default transformation scheme. - Stress assessment practice: - The solvency position of financial institutions is assessed along the scenario horizon (not only at end), with emphasis on the “low-point” (minimum capital ratios) for each bank along the horizon. *Source: IMF staff report (Korea FSAP chapter content provided).* ### 47. Historical transition data in accordance with IFRS 9 classification criteria have been ### 1korea2020006 - 47. Historical transition data in accordance with IFRS 9 classification criteria have been ### Historical transition data and mapping to IFRS 9 stages - Historical transition data were compiled by the FSS from Korean banks under FSAP guidance. - Mapping from Korean prudential asset classification to IFRS 9 stages: - Prudential “normal” → Stage 1. - Prudential “precautionary” → Stage 2. - Prudential “substandard”, “doubtful” and “loss” → Stage 3 (jointly). - Historical flows included additional flow categories: new business, repayment, write-offs, and asset sales. - Data were collected at bank-level for three portfolio segments: nonfinancial corporate (large firms and SMEs combined), household mortgages, and consumer credit. - Quarterly historical transition flow data availability: 2010Q1-2018Q4 for the majority of banks and portfolios. ### Z-factor estimation and methodology (Box 1) - Historical Z-factors and related parameters were estimated based on historical bank-portfolio-specific transition flow data (reference: Figure 17, Block C). - Z-score methodology summary: - Assumes probability density X is a function of idiosyncratic driver Y and systematic driver Z; both are independent unit normal random variables. - Parameter 휌 captures correlation between Z and X, with Z explaining a fraction 휌 of the variance of X. - Representative relation (equation numbering preserved): - (1) 푋푡 = √1−휌 푌푡 + √휌 푍푡 - Fitted transition probabilities expressed as: - (2) ∆푡(푥푔+1퐺,푥푔퐺,푍푡,휌) = Φ(푥푔+1퐺 − √휌푍푡 / √1−휌) − Φ(푥푔퐺 − √휌푍푡 / √1−휌) - Φ is a standard normal cumulative distribution function. - 푥푔퐺 are “bin boundaries” computed from a long-term average transition matrix. - Historical deviation between observed and fitted transition matrix minimized by: - (3) min푍푡 ∑∑ 푤푡푔 (푃푡(퐺,푔)obs. transition rates − ∆(푥푔+1퐺,푥푔퐺,푍푡,휌)fitted transition rates)2 - Constant 휌 and time series 푍푡 were estimated jointly subject to the constraint that the variance of 푍푡 equals one, for each bank-portfolio. - Interpretation: a +1/-1 value for Z denotes a “1-standard deviation from normal (long-run average)” conditions. - Estimated bank-portfolio-specific 휌 values for Korean banks are reported (reference: Figure B1.2). ### Hybrid PD–satellite–Z-factor methodology (Box 2) - A “hybrid” PD-satellite-Z-factor methodology was developed for Korea (Figure 17, Blocks A-G). - Rationale for hybrid approach: - Could use Z-factor time series to derive transition matrices and implied default rates directly, but historical transition matrix time series (starting in 2010) deemed too short to capture a full economic cycle. - Default-rate series were longer (starting end of the 1990s) for most portfolios; therefore, PD-satellite models were used in combination with Z-factor methods. - Method design: - Bank-portfolio-specific Z path for a 20-quarter forward horizon was set so that exposure-weighted PD would match the path implied by PD satellite models. - PD paths from satellite models attached to bank-portfolio-specific PD starting points using a distance-to-default transformation. - Exposure-weighted PDs based on transition matrices computed period by period via: - (6) PDt+1|t PiT = (S1t × TR13t+1|t + S2t × TR23t+1|t) / (S1t + S2t) - Translation from Z-score forecast path back to transition matrix uses same functional form as eq. (2), now with scenario-dependent Z denoted 푍̂푡: - (4) ∆t(xg+1G,xgG,Ẑt,ρ) = Φ(xg+1G − √ρ Ẑt / √1−ρ) − Φ(xgG − √ρ Ẑt / √1−ρ) - Bin boundaries and 휌 are fixed from Box 1 estimates; only Z varies across scenarios. ### Dynamic balance sheets, stock equations, and model inputs (Box 2, related text) - Stock evolution formulas (preserved as presented): - (5) S2t = S2t−1 + TRt12 S1t−1 + TRt32 S3t−1 [Inflows to S2] − TRt21 S2t−1 − TRt23 S2t−1 − MtS2 S2t−1 [Outflows away from S2] - S3t = S3t−1 + TRt13 S1t−1 + TRt23 S2t−1 [Inflows to S3] − TRt31 S3t−1 − TRt32 S3t−1 − WROt S3t−1 [Outflows away from S3] - St = S1t + S2t + S3t = (1 + gt) St−1 - S1t = max(0, St − S2t − S3t) - Portfolio-specific gross loan growth gt is scenario-determined and equal across banks. - Repayment percentage for S2 exposures and write-off/asset-sale parameters (MtS2 and WROt) held constant at observed end-sample positions for all portfolios and banks. - Write-offs and asset sales allowed to be positive; important due to Korean banks’ active use of NPL management firms. ### Perfect foresight assumption and scenario treatment - A “perfect foresight” assumption was employed (Figure 17, Block E): - IFRS 9’s multiple probability-weighted scenario requirement for accounting provisioning was ignored for solvency stress testing. - Instead, two scenarios considered separately with 100 percent weight each: baseline and adverse. ### Time horizons and treatment beyond 5-year scenario window - ECL provisioning under IFRS 9 requires expectations over the lifetime of financial assets; this may exceed the FSAP scenario 5-year horizon for portfolios with residual maturities larger than 5 years (notably mortgage portfolios). - Baseline risk parameters (PDs, LGDs) as of 2023Q4 (last quarter of year 5) were held constant thereafter. - Adverse scenario parameters at 2023Q4 assumed to decay back to baseline over an 8-year period (reference: Figure 19). - A sensitivity analysis varied this mean reversion period between a 5-12-year range; solvency stress test results were not very sensitive to this range. ### LGD modeling and overlays - A simple structural LGD model linked housing-collateralized portfolios’ LGDs to house price paths (applied for residential real estate portfolios including Jeonse leasehold loans). - For all other portfolios, LGDs: - Kept constant in the baseline. - Scaled by a factor of 1.15 under the adverse scenario. - Conservative overlay applied for all portfolios in scope of the CR Module: take the maximum of model-implied LGDs (structural model or simple factor) and bank-reported regulatory downturn LGDs. - Resulting distributions of LGDs under baseline and adverse scenarios are shown (reference: Figure 20). ### Risk-weighted assets (RWs), PD smoothing, and parameters - RWs for STA exposures kept constant at end-2018 levels. - RWs for IRB exposures allowed to move dynamically as a function of underlying changes in regulatory TTC-PDs. - Downturn LGDs kept constant. - TTC PDs at bank-portfolio level made a smooth function of PiT-PDs via: - PDt+1|tTTC = Φ−1(Φ(PDt|t−1TTC) + α × ∆Φ(PDt+1|tPiT)) - Smoothing parameter α set to 0.5 for base results; a sensitivity analysis related to α reported in results section. - RWA evolution in volume terms reflects allowance for dynamic balance sheets. ### Interest income, expense, and solvency–funding feedback - Interest income and expense module based on cross-bank panel econometric models using banks’ historical effective interest income and interest expense rates (Figure 22). - Panel models estimated separately for: nation-wide banks, regional banks, and specialized banks. - Regional banks’ models used for ODIs (credit unions, credit cooperatives and mutual savings banks). - Interest expense model contains feedback from banks’ solvency position to funding cost (Appendix V), creating a sequential iterative link from end-period capital ratios to subsequent quarter funding cost. - Effective interest rates as of end-2018 were sourced at bank-portfolio level and held constant for discounting ECLs (reference: Figure 21). ### Lifetime ECL, provision stock and flow calculations (Box 3) - Lifetime ECL formula used for S2 exposures: - (6) ECLtLT,S2 = ∑ TRs2−3,* × LGDs × S2s−1 / (1 + r)s, with M denoting average residual maturity of a portfolio (measured in monetary units). - Incremental PD used in numerator: - (7) TRs2−3,* = TRs2−3 × ∏s−1 t=1 (1 − TRt2−3) (PiT survival probability = cumulative survival probability) - TRs2−3 is unconditional transition probability for S2 stocks from the transition matrix forecast path; incremental PD measures probability of default in period s conditional on survival up to s−1 and approaches zero over time. - Exposure S2 projected using a linear principal repayment scheme. - For S1 exposures (12-month ECL): - (8) PROVt,S1 = ECLt,S1 = TRt+1|t13 × LGDt+H|t × S1t - TRt+1|t13 is expected default rate for S1 exposures conditional on end-period-t information for the following year. - LGDt+H|t denotes forward-looking LGD beyond 1 year if expected time until collateral sale exceeds 1 year. - For S3 exposures: - (9) PROVt,S3 = ECLt,S3 = LGDt+H|t × S3t - Total provision stock: - (10) PROVt = PROVt,S1 + PROVt,S2 + PROVt,S3 - Loan loss provision flow: - (11) PROVFLOWt = ∆PROVt + WROt × LGDt × S3t−1 - Adjustment for write-offs/asset sales assumes LGD estimate just prior to write-off/sale equals realized LGD upon collateral sale for exposures written off or sold. ### Miscellaneous model design notes and sensitivities - The hybrid design used a 20-quarter forward Z path tied to PD satellite models. - Perfect foresight used instead of multi-scenario probability-weighting for solvency stress testing. - Sensitivity checks included: - Mean reversion period for adverse-to-baseline parameter decay tested in a 5-12-year range with limited sensitivity in results. - Smoothing parameter α around 0.5 included in sensitivity analysis. - Capital floors for banks with IRB portfolios (as per Detailed Regulations on Supervision of Banking Business, Appendix 3) were not accounted for in this stress test (relevant for four banks in the sample). *Italic source attribution: Republic of Korea — IMF Financial Sector Assessment content unit 1korea2020006 (extract).* ### 59. In addition to being included as such, the solvency-funding cost feedback was allowed ### 59. In addition to being included as such, the solvency-funding cost feedback was allowed to be nonlinear ### Nonlinear solvency-to-funding-cost feedback - A notable nonlinearity was estimated in the solvency-to-funding-cost relation (Figure 23). - Interpretation example for regional banks: a 1p.p. shift to the capital ratio starting at 1 percent (4 percent) would induce a -0.25 p.p. (-0.1 p.p.) response of their cost of debt. - The estimated feedback from solvency to the banks’ cost of debt is significantly negative on average over the capital ratio range observed for the banks. - The light grey (yellow) shaded area in the referenced figure indicates the 5th-95th percentile range of banks’ shareholder equity/TA ratios under the baseline (adverse) scenario across banks and the scenario horizon. - Solid lines (dashed lines) depict the mean (10th and 90th percentiles) of the derivative of the cost of debt to the initial capital ratio. ### Interest income, credit risk linkages, and additional feedbacks - Credit risk materialization is linked structurally to interest income through the CR Module: initially performing exposures (Stage 1 and 2) moving to nonperforming status (Stage 3) imply no interest income by assumption in Stage 3. - Loan interest rates depend on banks’ funding cost; therefore the solvency → funding-cost feedback can effectively influence lending rates at the bank level (Appendix V). - Even a rather linear solvency-to-funding relationship would imply a nonlinear feedback effect on lending; this is amplified if the relationship itself is nonlinear. ### Market Risk (MR) Module—valuation approaches - Sovereign, nonfinancial and financial bond exposures under the FVOCI and FVPL categories were revalued using a modified-duration formula as a function of reported residual duration and scenario bond yield assumptions. - Domestic and foreign equity exposures were revalued via assumed equity index paths: KOSPI for domestic and MSCI World for foreign equity exposures under FVPL. - Domestic NFC exposures in FVOCI were subject to both MR and CR Modules: CR revaluation component reflected in P&L (tax non-neutral); FVOCI differential beyond CR impact reflected through OCI (tax neutral). - Sovereign and financial bonds under AC were also subject to MR due to missing Year 0 transition matrices; Korean banks’ foreign bond exposures were excluded from MR as they were deemed insignificantly sized. ### Fintech Overlay — methodology (Box 4) (medium-term solvency impact, no business model change) - Anchored in bank-level panel econometric models for interest income rate (IIR) and interest expense rate (IER) with bank-specific intercepts. - A Fintech Overlay “strength” parameter α ranges between zero and one. - Notation and construction: - Share of household loans in total loans: ℎ_b^L; share of household deposits in total liabilities: ℎ_b^D. - Initial bank-specific intercepts: c_b^L (IIR) and c_b^D (IER). - Cross-bank percentiles: ĉ_L = perc((c_b=1^L,...,c_b=B^L),1−α) and ĉ_D = perc((c_b=1^D,...,c_b=B^D),α). - Adjusted intercepts: - c̃_b^L = min(c_b^L, ĉ_L(α) + (1−ℎ_b^L)(c_b^L − ĉ_L(α))) - c̃_b^D = max(c_b^D, ĉ_D(α) + (1−ℎ_b^D)(c_b^D − ĉ_D(α))) - Interpretative assumptions: - Downward adjustment of income-intercept scaled by household loan share; upward adjustment of expense-intercept scaled by household deposit share. - Banks with no household deposits face no funding-cost change by assumption. - min/max operators imply intercepts already below/above target percentile remain unchanged. - Two embedded assumptions: “no business model change” (fees and other income not modified) and no caps on transaction fees (Open Banking caps not considered due to insufficient data). - The methodology is simplistic but useful for scenario-conditional approximations and can be further developed. ### Regulatory capital ratio thresholds - Inclusive of the CCB (2.5 percent), thresholds for non-DSIBs: - CET1/RWA: 7 percent - Tier1/RWA: 8.5 percent - Total capital (Tier 1 + Tier 2)/RWA: 10.5 percent - For DSIBs, these three thresholds are 1 percentage point higher. - Note for ODIs: capital thresholds were used only indicatively due to insufficient data quality; ODI capital is expressed in accounting terms net of loan loss provision stocks. ### Headline bank solvency stress-test results (adverse scenario) - The Korean banking system appears resilient under the FSAP adverse macro-financial scenario. - Distribution of losses: - A few regional, specialized banks and ODIs face the most sizable capital losses across banks. - Credit risk losses and diminished net interest income are the most pronounced sources of pressure on banks’ capital ratios. - Specialized banks would experience notable changes in capital ratios in both absolute and relative terms; cross-bank heterogeneity is most visible in the ODI category. - Aggregate outcomes: - System-wide capital depletion from the starting point amounts to about 22 percent of initial capital levels. - This represents about 2.9 percent of GDP as of 2018. - All banks’ capital ratios would stay above regulatory minima when allowing consumption of the capital conservation buffer (CCB). - A subset of specialized banks would consume between 0.3 and 1.4 percentage points of their CCB under the adverse scenario. - Portfolio sensitivities: - Consumer credit losses roughly align with portfolio shares. - Nonfinancial corporate loss shares are higher than their portfolio shares. - Household mortgage-related loss shares (including Jeonse loans) are lower relative to portfolio shares—reflecting widespread use of mortgage insurance and its effect on risk weight densities. - Insured portion of mortgage portfolios (exposure-weighted medians at end-2018): nation-wide 26 percent, regional 45 percent, specialized 27 percent (assumed constant along horizon). ### Fintech Overlay—results and implications - Fintech-induced competition could intensify pressure on capital ratios, particularly regional banks, ODIs and selected specialized banks. - Baseline scenario impacts under Fintech Overlay: - Regional banks’ capital ratios may fall by 0.6 to 1.3 percentage points by end of the five-year horizon. - For nation-wide and specialized banks, effects are nonlinear—only high overlay strength yields notable impacts. - Activating the Fintech Overlay raises system-wide capital depletion to 4 percent, from 3 percent without the overlay. - Aggregate consumption of CCBs when overlay activated: nation-wide 0.4 percentage points, regional 0.7 percentage points, specialized 2.2 percentage points. - Main channel: NIM compression—Figure 29 shows differences in NIMs compared to baseline, on average per annum, in percentage points. - MSBs: may be less affected because their loan spreads over funding cost are structurally higher due to riskier customer base, so competition may not reduce spreads as much. ### Solvency–liquidity (funding-cost) feedback quantified - The solvency-liquidity cost feedback is notable under the adverse scenario. - For eight of the banks, the additional feedback explains between -0.3 and -0.7 percentage points in the fall of their regulatory capital ratios (Figure 30). ### House price sensitivity analysis - Rationale: Korea’s special housing market—leasehold rent system and use of insurance for leasehold and mortgage loans by state-run institutions. - Historical reference: past endogenous peak-to-trough response during past episodes reflected about -16 percent. - Hypothetical severe shock assessed: a -40 percent house price decline, coupled with the adverse scenario. - The Korean banking system appears resilient to this combined shock; additional burden on banks’ capital ratios appears limited (Figure 31). - Caveat: analysis affects only mortgage portfolios (LGDs) and does not capture commercial real estate portfolios. - Assumption: house price movements in isolation would not materially affect mortgage PDs in Korea’s full-recourse environment. ### Mortgage insurance (shielding effect) - Mortgage insurance schemes are used to a significant extent in Korea and provide protection to banks’ capital and relief in risk weight densities. - Insured portion of mortgage portfolios (medians at end-2018): nation-wide 25 percent, regional 45 percent, specialized 27 percent. - These shares were assumed constant under both baseline and adverse scenarios. *Source: IMF staff analysis and supervisory data from the FSS, as presented in the referenced IMF document.* ### 76. To quantify the impact of switching the insurance off, three effects need to be ### 1korea2020006 - 76. To quantify the impact of switching the insurance off, three effects need to be ### Impact of switching off mortgage insurance - Three effects accounted for: - accounting provision stock difference at the end-sample position, - provision flow difference conditional on the scenario along the 5-year scenario horizon (focus on the adverse scenario), - relief in terms of risk weights. - Auxiliary regression used to capture effect on RWAs: - Equation structure: 푅푊퐷 푏 =훼+훽푃퐷 푏 푇푇퐶 +훾 ( 1−푤 푏 퐼 ) 퐿퐺퐷 푏 퐼 . - Definitions preserved from source: - 푅푊퐷 푏 : bank-specific risk weight density for the IRB mortgage portfolio (for banks with non-zero share). - 푃퐷 푏 푇푇퐶 : IRB mortgage portfolios’ TTC-PD. - 푤 푏 퐼 : share of the mortgage portfolio that is insured. - 퐿퐺퐷 푏 퐼 : regulatory downturn LGD estimated for the insured portion, computed as 퐿퐺퐷 푏 퐼 =퐿퐺퐷 푏 /(1−푤 푏 퐼 ), where 퐿퐺퐷 푏 is the weighted average DT-LGD for the portfolio. - Assumptions: TTC-PD same for insured and uninsured portion; LGD for insured portion is zero percent. - All variables measured as of 2018. - Regression performance: R-square of 89 percent; 훽 and 훾 economically significant positive coefficients. - Counterfactual results (insurance share set to zero): - Weighted average impact per bank group on CET1/RWA from combined drivers: nation-wide banks 21 basis points, regional banks 28 basis points, specialized banks 7 basis points. - About 70 percent of these impacts stem from the risk weighted asset effect. - For selected banks in the three categories, overall impacts reach 26, 46, and 32 basis points, respectively. - Regional banks most sensitive due to highest insurance shares in the banking system. - Note: ODIs excluded from counterfactual analysis because mortgage insurance shares were not available for them; the average across the regional banks was employed for them. ### Static vs. Dynamic bank balance sheets - Key findings: - Assuming static instead of dynamic balance sheets makes bank-solvency results more benign. - Disallowing write-offs and asset sales (while otherwise operating under dynamic balance sheet) yields higher capital ratios than full dynamic balance sheet assumption. - Move to full static balance sheet (zero gross loan growth, no write-offs and asset sales) renders solvency ratios even more benign. - Median differences to full dynamic balance sheet (DBS) outcome: - nation-wide banks: 28 basis points, - regional banks and ODIs: 46 basis points, - specialized banks: 50 basis points. - For some selected nation-wide banks, the difference would equal 90 basis points. ### Regulatory loan loss provision overlay - Regulatory provision coverage ratios: - Tend to exceed top-down accounting model-implied starting point coverage ratios based on the adverse scenario. - Regulatory coverage ratios largely exceed the Year 0 adverse scenario-based accounting model-implied coverage ratios (focus on orange dots for the adverse scenario). - Differential would concern largely Stage 1 and Stage 2 stocks; Stage 3 exposures are provisioned based on net-of-expected-recovery exposures and effectively similar to accounting provisions. - Capital depletion if regulatory provision scheme switched off: - Headline impact would rise from 2.9 percent of GDP to 3.7 percent of GDP. ### RWA pass-through strength and sensitivity - Pass-through parameter links changes in PiT-PDs to TTC-PDs to obtain RW changes for IRB portfolios. - Baseline ("headline") pass-through strength used: 0.5. - Sensitivity analysis range: from zero to one. - Under strongest pass-through of 100 percent: - All nation-wide, regional and specialized banks would not fall short of their regulatory minima. - Two nation-wide banks would start consuming their CCB under this conservative assumption. - Higher sensitivity of nation-wide banks due to higher IRB portfolio shares. - Impact of pass-through relates to IRB portfolio shares (analysis based on subset of 19 banks who do or may employ the IRB approach). ### Recommendations (selected) - FSS and BOK top-down stress test frameworks are well developed and instrumental for bank-by-bank analyses. - Fintech developments (Open Banking, e-money providers) require thorough risk analysis, emphasizing systemic liquidity implications and system-wide operational risks. - Develop a lifetime ECL and IFRS 9-compatible credit risk model to improve top-down scenario analysis and assessment of portfolio-specific provision stock coverage. - Invest resources in back-testing top-down model components, particularly structural components such as LGD models. - Invest resources into developing a dynamic balance sheet feature and explicitly account for asset sales and write-offs in models; acknowledge FSS’ willingness to modify stress test model to account for write-offs. - Further develop stress test models for credit guarantee institutions (HF, SGI, HUG, others), link to a common scenario, develop satellite model systems for guarantors, and integrate guarantor stress tests with banks and other non-bank FIs. - Treat Korea’s ODI sector as part of the banking system in risk analyses (ODI sector constitutes about 20 percent of the banking system). ### Forward-looking bank liquidity analysis — Overview and scope - Top-down liquidity stress test (LST) conducted based on 2018Q4 regulatory data. - Sample: 18 banks (six nation-wide commercial banks, two internet-only banks (KRW exposures only), six regional banks, four specialized banks) plus foreign bank branches (KRW only). - Coverage: approximately 90 percent of the banking system as of 2018Q4 in terms of total assets. - Analysis included: liability characteristics, liquidity buffers, asset-liability-matching. ### Current liquidity conditions and bank liquidity profiles - Liquidity Coverage Ratios (LCR): - Stand above recently introduced requirement of 100 percent. - LCR requirements introduced to levels of 100 percent for all currency (60 percent for foreign bank branches) and 80 percent for FX in 2019. - Most banks already met these requirements in 2018. - Net Stable Funding Ratios (NSFR): - All banks subject to the NSFR met the 100 percent minimum regulatory threshold in 2019Q1. - Funding mix: - Diversified; FX funding structure has become more stable. - Short-term FX funding mostly comprises FX deposits by domestic nonfinancial corporations. - Retail deposits represent most KRW funding. - NSFR funding structure appears more diversified across retail and SME deposits, corporates, other financial institutions and sovereigns. - Maturity matching: - Contractual maturities of assets and liabilities appear well matched overall. - Some short-term maturity mismatch for selected regional banks and internet-only banks in KRW: - Maturity gaps close to -8 percent for both bank categories for assets and liabilities maturing within 6 months. - FX funding gap positive for all bank categories. ### Liquidity stress test methodology - LST assesses ability to withstand net-cash outflows amid shocks to asset prices, retail funding, and wholesale funding. - Four scenarios: one for each of the three shocks and an extreme combined scenario. - Assessment conducted for KRW only, FX only, and both combined. - Hurdle rate set to 100 percent for all cases. - Liquidity shocks calibrated more severely than historical experience of Korea due to rarity of liquidity crises and Fintech/Open Banking-induced deposit mobility. - Example: LCR assumes 40 percent runoff on unsecured non-operating deposits, whereas LST assumes 50 percent. - Liquidity shortfalls calculated as percentage of banks’ counterbalancing capacity over 20 scenarios with increasingly severe shocks for retail and wholesale funding. ### Liquidity stress test results - Asset price shocks: - Liquidity coverage decreases from 121 percent to 120 percent under asset price shocks consistent with bank solvency stress test. - HQLA composition: central bank reserves withdrawable in stress (15 percent) and securities with 0 percent risk-weight (73 percent). - Retail funding shocks: - Banks appear resilient overall. - Average liquidity coverage of nation-wide banks falls to 98 percent (marginal liquidity shortfall) for KRW; FX liquidity coverage remains above 100 percent. - Regional and specialized banks would carry sufficient FX liquidity to cover implied shortfalls in KRW. - Internet-only banks and foreign bank branches: KRW liquidity buffers would cover deposit run-off. - Internet-only banks coverage ratio falls from 284 percent to 140 percent. - Wholesale funding shocks: - FX liquidity is sensitive to wholesale funding shocks. - Banking system likely to maintain liquidity following wholesale funding shock, but specialized banks’ reliance on unsecured wholesale funding leads to FX liquidity coverage falling from 112 percent to 85 percent. *Source: Supervisory data from the FSS, publicly available data for banks, and IMF staff calculations (excerpts from the provided content).* ### 101. The liquidity shortfall is largest for nation-wide banks. Using gradually more severe ### 1korea2020006 - 101. The liquidity shortfall is largest for nation-wide banks. Using gradually more severe ### Liquidity stress-test findings (banking system) - Overall system shortfall in KRW would reach 2 percent of the initial counterbalancing capacity. - Foreign bank branches would have excess liquidity of approximately 8 percent of the banking systems counterbalancing capacity. - Nation-wide banks would face a shortfall of close to 9 percent of counterbalancing capacity. - The banking system’s FX liquidity shortfall resulting from FX funding shocks would reach 4 percent of FX counterbalancing capacity. - Nation-wide banks would have an 8 percent FX liquidity shortfall. - Regional and specialized banks would have combined liquidity buffers of 4 percent. - In an extreme scenario where banks with liquidity shortfalls had no access to liquidity from banks with surpluses: - Aggregate shortfalls would reach KRW 43tr in domestic currency. - Aggregate FX shortfalls would reach USD 3.5bn, equivalent to less than 1 percent of the BOK’s FX reserves. ### Bank liquidity structure and vulnerabilities (summary of stress-test diagnostics) - KRW funding: - Domestic banks’ KRW funding is mainly from retail deposits. - Nationwide banks are most sensitive to run-offs in KRW retail deposits. - Regional and internet banks have negative KRW funding gaps over the next 6 months in some scenarios. - FX funding: - Most banks have FX LCRs above the 80 percent minimum requirement introduced in 2019. - Korean banks largely rely on the deposits of domestic corporates for FX funding, making them sensitive to FX wholesale funding shocks. - The maturity of FX assets and liabilities are well matched for the sector overall, but non-D-SIBs are more sensitive to FX wholesale funding shocks. - LCR/NSFR: - Banks’ LCRs remain above the recently introduced requirement of 100 percent in KRW. - The NSFR suggests banks will have buffers to meet funding requirements for the next 12 months; the NSFR funding structure is well diversified. - Stress-scenario sensitivity: - Retail and wholesale funding shocks have similar effects on banking liquidity. - Liquidity shortfalls in KRW are sensitive to retail deposit shocks. - Reliance on FX deposits of domestic corporates magnifies sensitivity to FX wholesale funding shocks. - D-SIBs are resilient to liquidity shocks and have ample FX liquidity buffers; non-D-SIBs show greater sensitivity. ### Central bank (BOK) capacity and policy space - The BOK has ample space to provide liquidity in extreme stress scenarios. - Both KRW and FX liquidity could be provided through standard facilities or unconventional policy measures. - The FX shortfall estimate (USD 3.5bn) is less than 1 percent of the BOK’s FX reserves. ### Recommendations on liquidity stress testing and monitoring - Continue to improve the bank LST for domestic currency and develop the FX LST by major currencies. - Expand the methodology of the joint BoK and FSS FX liquidity stress test to: - include FX LCRs; - add an NDF stress scenario in different types of stress tests; - periodically survey the resident NDF FX market participants for any potential macroprudential effects as global shifts take place in currency and interest rate markets. - Authorities are monitoring FX liquidity closely with both the BOK and the FSS carrying out FX liquidity stress tests separately, and maintain a joint working group responsible for monitoring liquidity. - Efforts to stress test KRW liquidity should be accelerated to keep pace with recent developments in FX liquidity stress testing. ### Insurance sector: stress-test design and results - Scope: - Top-down solvency stress test for seven life and six non-life insurance firms covering about 75 percent of the sector’s assets, using public and supervisory data as of end-2018. - Adverse scenario (comparable to bank stress test; more focused on market shocks): - Included shocks to sovereign bond yields (up to +250 bps), equity prices (up to -39 percent), corporate credit spreads and defaults, FX rates. - Shock applied to mutual funds: -12.5 percent. - No explicit shock used for real estate (direct holdings allowed only for own use). - All shocks assumed to occur instantaneously and modelled as haircuts to asset values. - Accounting and valuation effects: - Held-to-maturity designation is widely used; around 23 percent of the sample’s investment assets were shielded from market price changes. - Main output: RBC coverage impact - Required capital assumed to decline marginally by 2 percent in the stress model. - Life insurers: - Starting median RBC coverage: 239 percent. - Median life company drops to 169 percent under stress. - Two insurers would see coverage slightly below the recommended level of 150 percent, but remain above the 100 percent regulatory threshold. - Overall reduction in available capital: KRW 27.4 trillion, corresponding to a 38.7 decline, or 4.3 percent of total assets. - Non-life insurers: - Median RBC coverage declines from 214 to 158 percent. - Available capital reduced by KRW 8.9 trillion (-27.8 percent, or -3.9 percent of assets). - Only one company in the sample falls below the recommended level. - Contribution of shocks to available capital decline: - Life insurers: - Equity shock: 35 percent of the KRW 27.4 trn reduction. - Corporate bond shock: 30 percent. - Sovereign yield shock: 18 percent. - Mutual fund shock: 15 percent. - Non-life insurers: - Corporate bond shock contributes 50 percent of the reduction. - FX shock affects only a minority of companies due to widespread hedging using currency swaps and forwards. - Profitability implications for life insurers: - Underwriting results have been under pressure and turned negative in 2018. - Under the instantaneous stress with no recovery, investment income in 2019 would be 12 percent (KRW 2.2 trillion) lower than 2018. - Net income before tax would drop to 2.3 trn (-42 percent) in 2019, the first year of the projection horizon, recovering afterwards primarily from positive investment results. - Projected underwriting results remain muted, highlighting urgency for life companies to restructure business toward lower guarantees and protection business. ### Pension fund sector: key risks and NPS profile - Systemic challenges: - Korean pension system faces heterogeneous challenges across all three pillars. - First pillar currently provides decent returns, but long-term sustainability depends on reforms to contribution and benefit parameters. - Strengthening second and third pillars is urgently needed. - National Pension Service (NPS) specific metrics and projections: - NPS current assets: KRW 700 trn (current). - Projected growth to KRW 1,778 trn in 2041 when benefit payments will start exceeding contributions. - With parameters unchanged, fund depletion projected by 2057. - Options to address long-term sustainability: reduce benefits, increase contribution rate, increase retirement age, increase target for investment returns. - Operational challenges: - Recent staff fluctuation, especially among investment and risk managers, increased after moving the headquarter out of Seoul. - NPS faces competition with the private sector for talent. - Investment portfolio and risks: - More than 60 percent invested in domestic assets (NPS still mostly invested in domestic assets (60 percent)). - In domestic markets: - NPS holds more than 7 percent of the stock market’s capitalization. - NPS holds 18 percent of Korean sovereign bonds. - Domestic equity portfolio heavily weighted towards electrical and electronic equipment (30 percent). - Alternative investments: 12 percent of total assets, split: real estate 37 percent, private equity 32 percent, infrastructure 30 percent, hedge funds 2 percent. - Past performance: - Return was negative in 2018 at -0.9 percent. - Since inception in 1988, average annual performance was 5.6 percent. - 2019 performance noted as for the first nine months only; after a negative 2018, NPS recorded an almost double-digit return in the first three quarters of 2019. - Risks from increasing overseas allocation: - Greater exposure to currency risk as foreign assets are currently largely unhedged. - Requires development of risk management capacity. - Historical tendency: KRW/USD exchange rate tended to co-vary negatively with foreign asset prices, especially in the long term. *Source: IMF staff and FSS calculations as presented in the supplied content.* ### 115. In the second pillar, risks to pension scheme members and beneficiaries are largely ### 1korea2020006 - 115. In the second pillar, risks to pension scheme members and beneficiaries are largely ### Second pillar (retirement pensions): structure, performance, and reforms - Products offered: bank deposits, guaranteed investment contracts (financial products with principal and interest guaranteed), or funds (financial products whose earnings depend on their performance). - Providers: around 50 pension service providers, including banks, insurers and securities firms. - Assets: Assets managed by these pension service providers reached KRW 190 trillion at end-2018. - Banks managed 51 percent. - Life insurers managed 23 percent. - Investment firms managed 19 percent. - Non-life insurers managed 6 percent. - Investment composition: Around 90 percent of retirement pension funds were being invested in principal-protected products, and the remainder in funds. - Returns in 2018: - Average return: 1.0 percent. - Principal-protected product yield: 1.6 percent. - Performance-based type: -3.8 percent. - Market developments and constraints: - Pension participants and sponsors have recently shown more interest in diversified fund vehicles and products without guarantees. - Retirement pensions cannot exploit their full potential yet because necessary reforms are stuck in the parliamentary process. - Multi-employer (pooled) pension funds have been discussed for several years; such schemes require careful monitoring to ensure fair treatment of all cohorts and scheme members. ### Third pillar (private pensions): composition and performance limitations - Product types: savings-based annuities (largest proportion with 78 percent), pension savings trust, and pension savings funds. - Tax treatment: All third-pillar products offer tax deduction benefits to investors. - Performance issues: - Pension savings trusts offered by banks have traditionally shown relatively weak performance. - These trusts can typically be redeemed annually without full market value adjustments, effectively locking in guaranteed amounts every year rather than providing the guarantee at retirement age. - Investments are heavily in sovereign bonds and highly rated corporate bonds. - This design limits long-term performance and reduces incentives for potential customers to join. ### Recommendations: insurance stress testing - The FSS should further develop its insurance stress tests and include adverse macro-financial scenarios based on a medium-term risk outlook. - Current practice: FSS uses single-factor sensitivity analyses which potentially underestimate the effect of a widespread financial crisis on balance sheets and solvency ratios of insurers. - Suggested approach: Combine risk factors in a consistent adverse scenario that considers dependence across risk factors; use for both top-down and bottom-up stress tests. - Post-K-ICS implementation: Stress test results could inform the need for potential capital add-ons and, at a minimum, determine the level of supervisory intensity. ### Network and contagion analysis — analysis - Methodology and scope: - Network approach based on cross-exposures for 64 large financial institutions across banks, insurance, investment firms, and credit-specialized firms. - Network visualization based on raw exposure data and on the outcome of a default simulation confirm significant balance sheet interlinkages. - Default simulation follows the methodology developed by Espinosa-Vega and Solé (2010); assumed LGD equal 100 percent. - Key findings: - Specialized banks are about as systemic as nationwide banks; confirmed by network centrality metrics. - Nationwide banks are net lenders to specialized banks and therefore more vulnerable to specialized banks than vice versa. - Life and non-life insurance firms are not as systemic but are vulnerable to stress in specialized banks and nationwide banks. - Insurers are net lenders to all other subsegments of the Korean financial system. - Banks and insurance firms that face the hardest hit on capital under the adverse solvency stress test are not tightly linked with the rest of the financial system; spillover potential from such firms appears sufficiently contained at the current juncture. - Caution: confidence effects can imply spillovers beyond the direct financial linkages captured by the default simulation methodology. - Data/sample notes: - Sample includes 21 banks, 25 life insurers, 29 non-life insurers, foreign bank branches as a block (38 underlying entities), Mutual Savings Banks as a block (79 entities), credit-specialized firms (102 entities), and investment firms (485 firms). - Firm-level impact and vulnerability results were developed and discussed with Korean authorities but are confidential. ### Network and contagion analysis — recommendations - Maintain and expand bilateral exposure data set compiled by the FSS for network and contagion analysis. - Collect exposure data on a regular and frequent basis (for example, quarterly) to support ongoing network and contagion analysis. - Develop dynamic network model methodologies to analyze time-varying exposure data. ### Demographic developments: trends and macro-financial implications - Broad trends: - Demographic developments in Korea have shifted adversely over past decades and are expected to keep rotating adversely in terms of stagnating population growth and rising dependency ratios. - The demographic rotation in Korea is one of the most adverse world-wide. - Old-age poverty: almost 44 percent of those aged 66 or older in Korea; OECD average is 14 percent. - Macroeconomic effects: - Demographic shifts imply a longer-term structural drag on natural interest rates and inflation, reflecting continuous drag in demand. - Demographics are one factor related to “secular stagnation.” - The impact on demand is uncertain and industry-specific: some industries may see slower investment, while others (e.g., health care, medical science) may benefit. - Elderly may spend savings, supporting velocity of money and contribution to GDP flows. - Housing market: - Falling demand for housing implies downward pressure on house and rent prices, exacerbated by the leasehold rent system (Jeonse) premised on rising house prices. - Supply-side and policy responses: - Public and private investment in technology is instrumental to sustain productivity growth amid a downward trend in potential output growth. - Korea’s Ministry of Science’s “Fourth Industrial Revolution” plan aims to boost productivity and innovation, foster female participation, and devote larger budgets to childcare and elderly care. - Productivity boosts may counteract upward pressure on real wages from a shrinking working-age population. ### Impact on pensions, banks, and insurers - Pensions: - Public and private pension funds will experience demographic consequences directly and require special attention. - The pension system will need to balance a shrinking contributing population with rising pension outflows. - Contribution share per contributor will need to rise in absolute and relative terms, reducing disposable income of younger cohorts and further dragging demand. - Long-term rising fund outflows may pull back funds from corporate and sovereign bond markets, lowering bond prices (raising yields) and exerting upward pressure on corporate and sovereign funding costs. - The sovereign may consider borrowing from banks instead, while banks’ credit business to the private sector could shrink. - National Pension Service (NPS): - NPS is in accumulation phase and expected to double its assets until 2040 when payouts will start exceeding inflows. - NPS manages close to KRW 700 trillion. - Membership: 22.1 million members and 4.7 million beneficiaries. - Second pillar (retirement pensions) sector size: around KRW 190 trillion as of 2018; only half of employees covered; reforms stalled in the parliamentary process for more than four year already. - Banks: - Aging channels affecting banks likely point to consolidation, business model restructuring, and rotation to non-interest income sources. - Mechanisms: lower interest rates and a flatter yield curve → structural downward pressure on net interest income → lower potential to retain profits and accumulate capital. - Falling demand for long-duration loans (e.g., mortgages) keeps interest income low. - Potential bank responses include: - Shrinking branch structure to reduce staff costs. - Shifting to non-interest income (fees, commissions). - Fostering innovation and conducting non-bank intermediary business (e.g., peer-to-peer lending). - Dis-saving by the elderly may leave banks’ deposit bases relatively unchanged. - Life insurers: - Move into protection products and health insurance to compensate for lower demand for traditional savings products. - Health insurance covers dread diseases including cancer and dementia; life insurers aim to provide health management services. - Longevity risks have not yet materially reduced underwriting profits; longevity risks currently contained. - Many savings products lapse before contractual maturity; benefits often paid as lump-sums or term annuities (usually 10 years). - As financial literacy improves and concern about old-age poverty rises, demand for whole-life annuities may increase, exposing life companies to heightened longevity risks. ### Quantitative assessment and long-term scenarios - Modeling approach: - A long-run equilibrium vector model relates total asset dynamics for nine components of the financial system to macroeconomic variables, capturing dependence on cyclical factors, population growth, and dependency ratios. - Model variables include GDP per capita, equity price levels, short- and long-term interest rates, GDP deflator levels, USD-KRW exchange rate, population and dependency ratio variables (population and dependency ratio are exogenous; others endogenous). - Scenario base data: - Long-term scenarios regarding population growth and dependency ratios sourced from KOSTAT (“Population Projections for Korea (2017-2067)”, KOSTAT, 2019). - Scenario set: 27 scenarios based on different assumptions for fertility, life expectancy, and migration. - Forecasts: - Scenario-conditional forecasts for financial system assets derived until the year 2040, conditional on the 27 population scenarios. - Historical note: Insurance sector and pension fund sector have yet grown significantly in Korea. - Data sources for figures: - UNDESA Population Division, World Population Prospects, 2019 for demographic charts. - Bank of Korea, Flow of Funds Statistics for financial system asset dynamics. - FSS supervisory exposure data and IMF staff calculations for network analyses and simulations. *Italic: IMF staff summary of chapter content (compiled from the provided source).* ### 133. The demographics scenario-conditional forecasts of the nominal size of different ### 133. The demographics scenario-conditional forecasts of the nominal size of different ### Demographics scenario-conditional asset dynamics — key findings - The scenario-conditional paths for average per annum total asset growth from 2019-40 vary across financial sector segments. - All financial sector components except pension funds are expected to shrink or move side-wards in nominal asset size. - Pension funds are expected to grow, as the rising old age dependency ratio appears to dominate. - Banks, including nation-wide, regional and specialized banks, are expected to move sideward in terms of nominal asset size, without much dependence on the demographic scenarios (reflected by the small range of the cross-scenario bounds). - Percentages in Figure 53 are natural log percent. The bars depict the median of the average per annum growth rates across the 27 scenarios; the lower and upper bounds mark their 10th and 90th percentiles. ### Drivers and sensitivities - Estimated elasticities indicate: - All financial system components depend positively on population growth. - The pension fund sector would benefit from a rising dependency ratio. - For pension funds, the two opposing effects (slowing population growth, rising dependency ratios) mean the net effect on asset growth is not pre-determined a priori. - The analysis suggests: - Non-life insurers and pension funds are most dependent on the demographic scenario assumptions (reflected by wider bounds around median estimates in Figure 53). - Pension fund asset growth trajectories depend most on assumptions about fertility. - All other financial sector components depend more on assumptions about migration. ### Policy implications — monetary policy - Monetary policy ought to be conducted with particular care at times of slow-moving, yet significant, demographic change. - Models supporting monetary policy should take explicit account of population dynamics (per capita thinking is warranted) to delineate cyclical and structural (demographic) factors. ### Policy implications — fiscal policy - Fiscal policies to counteract adverse demographic developments come with sizable implementation lags. - Modifying fiscal measures nowadays (for example, in relation to retirement age and policies to foster family creation) takes a long time to reveal effects. - Demographics-related analyses and policies should be very forward-looking and pre-emptive. ### Risk assessment and systemic perspective - The materiality of concerns from demographic change in Korea warrants great attention going forward. - The banking system does not appear to be at particular high risk of a sudden disruption due to structurally slow-moving demographic change. *International Monetary Fund — Republic of Korea Financial System Assessment (excerpt)* ### 1. Institutional ### 1. Institutional ### Institutional perimeter - Institutions included: - Seven life insurers, and six non-life insurers. - Market share (balance sheet assets): - Life: 73 percent. - Non-life: 76 percent. - Data: Regulatory reporting. - Reference date: December 31, 2018. ### Channels of risk propagation - Methodology: - Investment assets: market value changes of available-for-sale securities after price shocks, increase in the default rate for corporate and mortgage loan exposures. - Insurance liabilities: unaffected by change in interest rates as discount rates are based on historic cost accounting. - Sensitivity analysis: effect on available capital and solvency position. - Time horizon: Instantaneous shock. ### Tail shocks (Scenario analysis) - Adverse scenario parameters: - Short-term KRW sovereign bond yield -20 basis points, long-term KRW sovereign bond yield +250 basis points. - Stock prices -39.0 percent (Korea), -35.0 percent (other advanced economies), mutual funds -12.5 percent. - Korean corporate bond spreads between +120 basis points (AAA) and +400 basis points (BB and lower). - Foreign corporate bond spreads between +70 basis points (AAA) and +300 basis points (BB and lower). - Haircut on unsecured loans -1.5 percent, haircut on sub-standard loans -30 percent. - 37 percent appreciation of KRW against USD. - Sensitivity analysis: Variations of short- and long-term sovereign bond yields. ### Risks and buffers - Risks/factors assessed: - Market risks: interest rates, share prices, property prices, credit spreads. - Summation of risks, no diversification effects. - Buffers: - Accounting designation (23 percent of assets in the general account held to maturity). - Behavioral adjustments: None. ### Regulatory standards and parameters - Regulatory/accounting standards: - Korean Risk-Based Capital (RBC). - National GAAP. ### Reporting format for results - Output presentation: - Impact on solvency ratios. - Contribution of individual shocks. - Dispersion measures of solvency ratios and net income. --- ### Appendix III. Liquidity Stress Test – Calibration Details ### A. Eligibility of liquid assets - LCR treatment of asset categories (selection of key entries; percentages are LCR haircuts/weights by scenario): - Level 1 Assets: 100% (S1, S2, S3, S4, extreme). - Level 2a Assets: 85% (LCR), 70% (S1), 85% (S2), 85% (S3), 70% (S4). - Level 2b Assets examples: - Qualifying Mortgage Backed Securities: 75% (LCR), 60% (S1), 75% (S2), 75% (S3), 60% (S4). - Qualifying corporate debt securities rated between A+ and BBB-: 50% (LCR), 45% (S1), 50% (S2), 50% (S3), 45% (S4). - Qualifying common equity shares: 50% (LCR), 45% (S1), 50% (S2), 50% (S3), 45% (S4). ### B. Haircuts on inflows of liquid assets (over 30 days) - Haircuts by asset type and scenario (LCR, S1, S2, S3, S4): - Level 1 assets: 0% 0% 0% 0% 0%. - Level 2a assets: 15% 30% 15% 15% 30%. - Eligible RMBS: 25% 40% 25% 25% 40%. - Other Level 2b: 50% 55% 50% 50% 55%. - Margin lending backed by all other collateral: 50% 55% 50% 50% 55%. - All other assets: 100% 100% 100% 100% 100%. - Credit or liquidity facilities: 0% 0% 0% 0% 0%. - Operational deposits held at other financial institutions: 0% 0% 0% 0% 0%. - Other inflows, by counterparty: - Retail counterparties: 50% 50% 50% 50% 50%. - Nonfinancial wholesale counterparties, transactions not listed above: 50% 50% 50% 50% 50%. - Financial institutions and central banks, transactions not listed above: 100% 100% 100% 100% 100%. - Net derivative cash inflows: 100% 100% 100% 100% 100%. - Other (contractual) cash inflows: 100% 100% 100% 100% 100%. ### C. Outflows of liquid assets (over 30 days) - Retail Deposits (selected entries): - Stable deposits (demand): 5% (LCR), 5% (S1), 10% (S2), 5% (S3), 10% (S4). - Less stable retail deposits (demand): 10% 10% 20% 10% 20%. - Term deposits, residual maturity > 30d: 0% 0% 0% 0% 0%. - Unsecured Wholesale Funding (selected entries): - Small business stable deposits: 5% 5% 5% 10% 10%. - Small business less stable deposits: 10% 10% 10% 20% 20%. - Operational deposits generated by clearing, custody, and cash management activities: 25% 25% 25% 50% 50%. - Portion covered by deposit insurance: 5% 5% 5% 5% 5%. - Cooperative banks in an institutional network: 25% 25% 25% 25% 25%. - Nonfinancial corporates etc. fully covered by deposit insurance: 20% 20% 20% 20% 20%. - Not fully covered by deposit insurance: 40% 40% 40% 50% 50%. - Other legal entity customers: 100% 100% 100% 100% 100%. - Secured Funding: - Secured funding with a central bank, or backed by Level 1 assets: 0% 0% 0% 0% 0%. - Secured funding backed by Level 2A assets: 15% 15% 15% 15% 15%. - Secured funding backed by non-Level 1 or non-Level 2a asset, with domestic sovereign, multilat dev banks, or domestic PSEs as a counterparty: 25% 25% 25% 25% 25%. - Funding backed by RMBS eligible for Level 2B: 25% 25% 25% 25% 25%. - Funding backed by other Level 2B assets: 50% 50% 50% 50% 50%. - Other secured funding transactions: 100% 100% 100% 100% 100%. - Additional requirements (selected entries): - Valuation changes on non-Level 1 posted collateral securing derivatives: 20% 20% 20% 20% 20%. - Excess collateral held by bank related to derivate transactions that could be called anytime: 100% 100% 100% 100% 100%. - Liquidity needs related to collateral contractually due on derivatives transactions: 100% 100% 100% 100% 100%. - Increased liquidity needs related to derivative transactions allowing collateral substitution: 100% 100% 100% 100% 100%. - ABCP, SIVs, conduits, SPVs, or similar: - Liabilities from maturing: 100% 100% 100% 100% 100%. - Asset backed securities: 100% 100% 100% 100% 100%. - Undrawn but committed credit and liquidity facilities: - Retail and small business: 5% 5% 5% 5% 5%. - Nonfinancial corporates, sovereigns, central banks, multilat dev. banks, PSEs: - Credit facilities: 10% 10% 10% 10% 10%. - Liquidity facilities: 30% 30% 30% 30% 30%. - Supervised banks: 40% 40% 40% 40% 40%. - Other financial institutions: - Credit facilities: 40% 40% 40% 40% 40%. - Liquidity facilities: 100% 100% 100% 100% 100%. - Other legal entity customers, credit and liquidity facilities: 100% 100% 100% 100% 100%. - Other contingent funding liabilities: - Trade finance: 5% 5% 5% 5% 5%. - Customer short positions covered by customers' collateral: 50% 50% 50% 50% 50%. - Additional contractual outflows: 100% 100% 100% 100% 100%. - Net derivative cash outflows: 100% 100% 100% 100% 100%. - Any other contractual cash outflows (not listed above): 100% 100% 100% 100% 100%. --- ### Appendix IV. Solvency Stress Test – Mapping MR and CR ### Methodology summary - Mapping of CR and MR methodologies to reporting template line items depends on accounting treatment. - Sovereign bonds under the AC approach were subject to the MR methodology because no historical “year 0” transition matrices were available; sovereign bond exposure share amounts to about 4-6 percent across nation-wide, regional and specialized banks. - Sovereign and financial corporate bond holdings under the FVOCI accounting category were subject to only the MR methodology (no historical transition matrices available). Treatment under only the MR methodology is not expected to imply any distortion because net income of the banks under the adverse scenario would be negative, implying zero tax expense by design of the stress test methodology. - Nonfinancial corporate bonds under the FVOCI accounting category were treated under both the MR and CR methodology because historical transition matrix data was available. --- ### Appendix V. Solvency Stress Test – Credit and Interest Income and Expense Models ### Credit risk – PD modeling - A BMA methodology was employed to link bank portfolio PDs to macro-financial variables. - Left-hand-side variables: historical default rates for large corporates, SME loans, household mortgages and consumer credit. - Potential right-hand-side predictor variables: - Real GDP growth and a term spread for all models. - Unemployment rate for the mortgage and consumer credit portfolio. - USD-KRW exchange rate in the large corporate loan portfolio model. - Model structure features: - LRMs reflect sum of coefficients on a predictor for contemporaneous and lagged terms. - Models contain lags of predictor variables beyond contemporaneous inclusion, and no autoregressive lags of the LHS default rate variables. - Examples: large corporate loan portfolio included two lags of the exchange rate; SME default rate had an additional lag of real GDP growth; consumer credit portfolio had an additional lag of the unemployment rate. - Historical default rate data by the Korean banks is classified as confidential; historical data and fit not presented. ### Interest income and expense models — definitions - Interest income rate (IIR) and interest expense rate (IER) definitions: - IIR = 4× II / (SEC+loans gross of LLA ×(1 −SBL ratio)) - IER = 4× IE / (TA net of LLA−E) - Where: - II denotes quarterly interest income flows in nominal terms. - IE denotes quarterly total interest expense flows. - SEC (securities) measured as banks’ total financial assets minus their loan stock. - LLA the loan loss allowance stock (accounting). - SBL the substandard and below ratio. - Historical data coverage: sample of 19 banks (all but ODIs from Table 5), covering the 2000Q4-2018Q4 period (73 time series obs.), sourced from the FSS’ statistical data warehouse. ### IIR model (Appendix V Table 1) - Panel model variables and selected coefficients (nation-wide, regional, specialized banks): - IIR (-1): coef. 0.725 (p-val. 0.000); 0.501 (0.000); 0.618 (0.000). - IER: coef. 0.352 (0.000); 0.691 (0.000); 0.348 (0.000). - D(LOG(RGDP))*100: coef. -0.089 (0.000); -0.034 (0.099); 0.015 (0.657). - D(LOG(RGDP(-1)))*100: coef. 0.141 (0.000); 0.106 (0.000); 0.048 (0.153). - D(LOG(RGDP(-2)))*100: coef. 0.043 (0.038); 0.027 (0.186); 0.007 (0.820). - D(LOG(RGDP(-3)))*100: coef. -0.019 (0.368); -0.013 (0.514); 0.036 (0.243). - Model diagnostics (nation-wide / regional / specialized): - Obs. combined T and N: 320 / 414 / 177. - R2: 0.96 / 0.95 / 0.95. - SE of residuals: 0.28 / 0.32 / 0.28. - SE of data: 1.41 / 1.34 / 1.22. - Mean of data: 5.09 / 5.81 / 4.25. ### IER model (Appendix V Table 2) - Penal model variables and selected coefficients (nation-wide, regional, specialized banks): - STN: coef. 0.326 (0.000); 0.241 (0.000); 0.364 (0.000). - LTN: coef. 0.356 (0.000); 0.430 (0.000); 0.338 (0.000). - E/TA(-1): coef. -0.181 (0.000); -0.300 (0.002); -0.229 (0.026). - (E/TA(-1))^2: coef. 0.005 (0.001); 0.023 (0.004); 0.007 (0.119). - Model diagnostics (nation-wide / regional / specialized): - Obs. combined T and N: 353 / 432 / 218. - R2: 0.92 / 0.87 / 0.88. - SE of residuals: 0.31 / 0.38 / 0.37. - SE of data: 1.07 / 1.03 / 1.07. - Mean of data: 2.46 / 2.73 / 2.64. ### Interest model features and rationale - IIR model contains a simple distributed lag structure for real GDP growth and includes the bank-specific IER variable to capture pass-through of funding cost to loan prices. - IER model includes short- and long-term market interest rates and a bank-specific level and squared capital ratio term (E/TA), lagged by one quarter. - Inclusion of the quadratic capital ratio term captures suspected nonlinearity implied by trade-off theory of capital: marginal benefit of more debt vanishes with higher levels of debt, leading to a U-shaped relation between capital ratios and cost of debt. - Endogenization/feedback: - IER and IIR models are endogenized within the overall stress test machinery to reflect feedback from banks’ solvency to funding costs. - Steady state adjustment: - Intercepts (fixed effects) in the IIR and IER panel models were adjusted to match average left hand-side interest rates observed in 2018Q1-2018Q4, conditional on realized values of right hand-side variables. *Source: 1korea2020006 - 1. Institutional (PDF chapter).* --- _Source: https://www.imf.org/-/media/files/publications/cr/2020/english/1korea2020006.pdf_