## Japan FSAP Stress Testing — cr17285 ## Source details **Canonical URL:** [Japan FSAP Stress Testing — cr17285](https://www.imf.org/-/media/files/publications/cr/2017/cr17285.pdf) ## Other formats - [Markdown version](/-/media/files/publications/cr/2017/cr17285.pdf.md) - [Structured JSON version](/-/media/files/publications/cr/2017/cr17285.pdf.json) --- ### Executive summary - Comprehensive stress tests and interconnectedness analyses assessed resilience of Japan’s financial system and identified linkages and potential vulnerabilities. - System features: large financial conglomerates, a major banking sector, a highly concentrated insurance sector, large securities markets, heterogeneous securities firms. - High-level findings: - Financial system generally resilient to short-term risks; pockets of vulnerability exist. - Under a Severe Adverse Scenario, aggregate CET1 ratio would drop to 8.3 percent at its lowest point within the five-year horizon. - Some life insurance companies and regional banks may need to strengthen capital buffers. - Spillovers within the system appear limited (analyses do not capture all potential effects during stress). ### Major macrofinancial scenarios used - Three five-year scenarios (GFM/Global Macrofinancial Model): - Baseline: based on October 2016 WEO projections; short-term interest rates close to zero; BoJ yield curve control; CPI inflation below 2 percent target. - Moderate adverse (“De-Globalization”): protectionism, global growth peak drop of 1.6 percentage points vs baseline; growth drops by 1.8 percentage points in advanced economies and by 1.0 percent in EMEs; higher money market interest rates, lower equity prices. - Severe adverse (“Accelerated U.S. Monetary Policy Normalization”): 200 basis-point U.S. policy rate increase in 2017–18; assumed JGB yield total increase about 300 bps; long-term rates increase peaking at 1.6 percentage points for advanced economies on average. - Combined solvency and liquidity scenario: banks needing to improve capital buffers under severe adverse would face higher liquidity shocks. ### FSAP stress testing approach and coverage - Banking TD solvency stress test covered 20 largest banks, representing 90 percent of banking system assets. - Insurance TD and BU stress tests covered seven life and six non-life companies, representing 73 percent and 92 percent of respective sectors. - IMF TD solvency stress test horizon: five-year projections (2017–21). - BoJ TD used three-year projections. - Data bases: September 2016 balance sheets; March 2016 income statements. ### Key FSAP policy recommendations (selected) - Continue assessing the impact of default by largest borrowers on the whole banking system (JFSA). — Time: I - Implement an in-house multi-year top-down scenario stress testing framework for banks and insurance companies (JFSA). — Time: NT - Continue improving data consistency, especially between FX maturity mismatch and LCR (JFSA). — Time: I - Continue conducting liquidity stress testing regularly with significant foreign currencies and require banks to hold sufficient counterbalancing capacity, particularly HQLA (JFSA). — Time: I - Intensify supervision on FX funding liquidity risk by i) considering caps on funding concentration for significant foreign currencies; and ii) deducting from HQLA the assets subject to ringfencing by foreign jurisdictions (JFSA). — Time: NT - Collect data and adopt analysis of interconnectedness as toolkit of financial supervision (JFSA). — Time: I - Use insurance stress test results also to validate and benchmark ORSA reports and to gain further insights into the economic solvency regime (JFSA). — Time: NT --- ### Banking sector solvency stress tests — key results and drivers - Aggregate and system metrics: - Aggregate CET1 ratio would drop to 8.3 percent under the Severe Adverse Scenario (lowest point within five-year horizon). - Expected credit losses over five years: - Severe stress scenario: ¥12 trillion (equivalent to 1.2 percent of total banking system assets). - Baseline scenario: ¥4 trillion (equivalent to 0.4 percent of total banking system assets). - Main drivers of capitalization fall: equity- and market-related losses (including bond valuation effects from higher interest rates), credit losses, lower profitability from overseas lending due to higher U.S. dollar funding costs, and changes in RWAs. - Hurdle and recapitalization thresholds used: - CET1 of 4.5 percent (applied to City- and Regional banks with international exposure). - Total capital ratio of 4 percent applied to other regional banks without international exposure. - Tier 1 ratio of 6 percent, and 8 percent total capital ratio applied as additional thresholds for City- and Regional banks with international exposure. - 3 percent leverage ratio used as potential benchmark. - Concentration tests: - No bank fails to meet minimum CAR in case of default of the single largest borrower even under 100-percent-LGD assumption. - Default of five largest borrowers → 4 banks fall below minimum CET1 CAR. - Default of 10 largest borrowers → 12 banks fall below minimum CET1 CAR. - Default of the largest borrower in the system would lead to loss of 4.5 percent of system’s total capital; no bank undercapitalized in that scenario. - Portfolio exposures (distribution of banks’ exposures by asset class, in percent): - Qualifying retail; Equity; Corporate; Retail; Residential mortgages; Sovereign; Bank exposures = 5.4; 0.5; 2.0; 39.1; 1.6; 7.9; 43.5. - Risk-weighted assets distribution (in percent): - Qualifying retail; Equity; Corporate; Retail; Residential mortgages; Sovereign; Bank exposures = 4.8; 1.2; 11.9; 68.8; 2.6; 9.1; 1.7. - PD projections (median corporate PD): - from 0.5 percent in 2015 to 2.3 percent in 2019 under the severe adverse scenario. - peak of 1.1 percent under the moderate adverse scenario. - Equity and bond valuation losses during the three worst years ≈ 1 percent of assets; credit risk related losses ≈ 0.9 percent of assets. - Bank-type sensitivities: - Regional banks: highest sensitivity to credit-related losses; three regional banks would need additional capital to meet the 8 percent total capital requirement. - Domestic banks: relatively higher overhead costs and less diversified loan portfolios; slow decline in profitability and capital even under baseline. - City banks: smaller shocks to corporate PDs; equity- and market-related losses most important. ### Banking sector liquidity stress tests — structure and outcomes - Coverage and methodology: - LCR-based analysis and currency-level cash flow maturity ladder tests for yen, U.S. dollar, euro. - FG-consolidated basis: 14 banks; bank-solo: 16 banks. - Maximum liquidity stress horizon: 1 year; market-wide stress scenarios: Mild (2 weeks), Intermediate (3 months), Severe (1 year). - LCR and counterbalancing capacity: - All-currency aggregated LCRs: above 100 percent in all banks as of end-December 2016. - System-wide LCR in U.S. dollars: stands below 60 percent. - City banks (FG-consolidated; end-December 2016): Median (131%); Average (142%). - Regional banks: Median (134%); Average (144%). - Cash flow–based foreign currency results: - No bank fails in yen liquidity stress testing. - A few internationally active regional banks would have negative cumulative funding gaps in U.S. dollars in the severe one-year scenario after exhausting unencumbered U.S. dollar liquid assets. - Euro-denominated liabilities account for less than 5 percent of total liabilities; very small negative funding gaps in two regional and two city banks in severe 1-year scenario. - Combined yen–U.S. dollar stress: since U.S. dollar funding gaps are small relative to excess yen liquidity, no bank fails under joint stress conditions. - FX funding concentration and run-off assumptions (selected exact values): - In 2-week mild scenario: unsecured wholesale funding run-off assumed 30 percent; in U.S. dollar/euro mild scenarios assumed 40 percent of FX swap contracts cannot be rolled over. - In 3-month intermediate scenario: unsecured wholesale funding run-off assumed 40 percent; assumed 60 percent of FX swaps cannot be rolled over. - In 1-year severe scenario: wholesale funding secured by level 2A and 2B HQLA run-off rates assumed 30 percent and 80 percent, respectively; FX swap market severe stress assumed 80 percent run-off of U.S. dollar-denominated derivatives and 80 percent roll-off for yen-denominated derivatives. - Integrated solvency–liquidity results: - Losses from liquidation around 0.02 percent of GDP; or 0.05 percent of GDP to recapitalize banks up to 8 percent total CAR. - Overall system-wide capitalization impact: limited due to ample liquidity in JPY; affected banks can remain liquid in JPY after liquidating U.S. dollar portfolios. - Effects on credit contraction: negligible (shortages are in U.S. dollars, not Japanese yens). ### Cash flow run-off and roll-off parameter examples (selected exact values) - 2-week scenario run-off illustrative entries (percent): - Retail sight deposits (thereof sight deposits): 0.9 1.4 0.7 0.0 0.0 0.0 0.0 0.0 0.0 - Unsecured wholesale other deposits: 25.0 25.0 25.0 0.0 0.0 0.0 0.0 0.0 0.0 - Secured wholesale funding secured by Level 2B assets — unconditionally callable: 15.2 23.2 11.6 0.0 0.0 0.0 0.0 0.0 0.0 - Roll-off for secured lending (reverse repos): unconditionally callable = 100.00 (O/N); others = 100.00 across buckets. --- ### Insurance sector solvency stress tests — key findings and sensitivities - Solvency metric and hurdle: - Solvency Margin Ratio (SMR); statutory requirement (hurdle rate) = 200 percent. - Top-Down (TD) results — Life insurers (seven companies): - Average SMR: 949 percent before stress; 708 percent in the moderate adverse scenario; 419 percent in the severe adverse scenario. - Six out of seven life companies remain above 200 percent under current regime. - Aggregated available capital of seven life insurers: ¥34.4 trillion before stress; declines by 37 percent in moderate adverse and by 58 percent in severe adverse. - TD results — Non-life insurers (six companies): - Average SMR declines from 736 percent pre-stress to 625 percent in scenario 1 and 481 percent in scenario 2; all remain above 200 percent. - Bottom-Up (BU) results: - BU SMRs on average slightly higher than TD; all companies in BU remain above 200 percent. - Life BU: average SMR 949 percent before stress; 723 percent in moderate adverse; 452 percent in severe adverse. - Non-life BU: average SMR 736 percent before stress; 624 percent in scenario 1; 479 percent in scenario 2. - Key drivers: - Interest rate shocks and equity price declines account for more than three quarters of decline in available capital in the severe scenario. - Life insurers strongly sensitive to interest rates; under current statutory regime, sensitive to rate hikes. - Economic solvency regime (ESR) caveat: recent JFSA field test ESR for life sector = 104 percent as of March 2016 (considerably lower than statutory SMRs). - Under an economic solvency regime, life insurers less vulnerable to interest rate increases but more vulnerable to prolonged low rates. - Catastrophe and biometric sensitivity (selected exact impacts): - Great Kantō–like earthquake: reduction between 2 and 6 percent of available capital for non-life companies. - Typhoon Mireille repeat: reduction between 2 and 3 percent. - Hurricane Andrew–like repeat: marginal loss around 1 percent for Japanese non-life insurers exposed to that region. - Longevity shock (20 percent permanent decline in mortality): slightly positive effect up to 2 percent for most life insurers. - Pandemic: reductions in available capital between zero and five percent (varies by company). ### Insurance sector asset composition (as of March 2016) - Life insurers: - 44 percent in mostly long-duration JGBs and public bonds; 28 percent in corporate bonds. - Fair values exceed book values by 20 percent (sample JFSA data). - Share of assets not in JPY: 21 percent; FX-denominated liabilities: 7 percent. - Non-life insurers: - 20 percent in government/semi-government bonds; 32 percent in corporate bonds; 22 percent in equities. - 73 percent of assets in JPY (exceeds JPY-denominated liabilities). --- ### Connectedness, contagion, and spillovers — patterns and model outputs - Channels of propagation: - Primarily via client and investor bases; direct exposures modest in funding shock transmission. - Strong links between financial and nonfinancial firms underpin market-based spillovers. - Cross-shareholding and common exposures to foreign markets underpin some spillovers. - Market-data (equity-return) spillovers: - Main sources: Japan’s nonfinancial firms and city banks. - Shocks to largest banks and largest nonfinancial firms propagate strongly to insurance companies. - Balance-sheet connectivity metrics (average / max by sector; selected entries preserved exactly): - Betweenness (Average / Max): - Mega Banks: 45.80 / 64.30 - City Banks: 18.60 / 68.30 - Regional Banks: 12.20 / 31.60 - Life Insurance: 6.40 / 18.20 - Funding centrality (Average / Max): - Mega Banks: 0.99 / 1.00 - City Banks: 0.73 / 0.94 - Regional Banks: 0.69 / 0.86 - Life Insurance: 0.38 / 0.78 - Balance-sheet contagion simulations (Espinosa-Vega and Solé approach): - Funding-shock result: withdrawal of funding by large city banks would cost slightly more than 2 percent of the financial system’s total capital. - Securities firms are most vulnerable to funding shocks; average capital losses ≈ 2 percent of total capital in the simulation. - Credit-shock result: default by a large city bank would absorb 12 percent of the financial system’s total capital (average losses). - Default by a smaller city bank or a securities firm → simulated losses ≈ 7 percent and 5 percent of system’s total capital, respectively. - Life insurers suffer most under credit-default scenarios because of deposits, credit, and equity exposures. - Sector-default simultaneous failures generally have mild implications for other sectors when considering direct balance-sheet exposures only. - Determinants of spillover sensitivity (econometric/PPML gravity & panel results): - Firms with higher z-Score and higher ROA are less sensitive to shocks. - Larger firms are more important sources of spillovers. - Among banks, reliance less on wholesale funding and higher loan shares → less sensitivity. - Higher shares of institutional ownership → associated with lower sensitivity to spillovers. - Greater geographic proximity → larger spillovers. ### Model and data caveats for contagion and spillover analysis - Balance-sheet model excludes derivatives exposures. - Exposures do not capture common-client exposures, shareholder links, or market-sentiment feedback loops (simulated contagion losses are lower-bound estimates). - PDs for sovereign exposure historically extremely low (below 0.03 percent in most cases). - Stress tests calibrated to extreme but plausible tail events; they do not capture all possible unanticipated shocks or interactions. --- ### Additional quantitative inputs and illustrative parameters (selected) - Total financial assets held by the financial sector: about 620 percent of GDP in 2016. - Twenty financial conglomerates ≈ 170 percent of GDP in assets; four largest financial groups account for 136 percent of GDP. - Life insurance total financial assets ≈ 75 percent of GDP; life accounts for about 90 percent of insurance sector. - Equity market capitalization ≈ 100 percent of GDP. - Average maturity of public debt: about 7.7 years. - Recovery rate for one-year default probability for Japan sovereign used: 40 percent. - Statutory tax rate used in modeling: 30 percent. - Behavioral cap on fee income response: growth should not decline more than 10 percent (y-o-y) in any scenario. - Reported model R^2 examples: 99%, 97%, 93%, 81%. - Banking sample composition shares (percent, excludes Japan Post Bank): 86.10; 8.30; 5.60 (City banks; Regional banks; Domestic banks). --- _Italic: Source — EXECUTIVE SUMMARY and selected chapters/extracts from cr17285 (IMF staff and JFSA materials)._ ### EXECUTIVE SUMMARY __________________________________________________________________________ 7 ### EXECUTIVE SUMMARY ### Overview and context - A comprehensive set of stress tests and interconnectedness analyses were conducted to assess the resilience of Japan’s financial system and to identify linkages and potential vulnerabilities. - Japan’s financial system features large financial conglomerates, a major banking sector, a highly concentrated insurance sector, large securities markets, and a heterogeneous set of securities firms. - Various quantitative tools and models were used to examine short- and medium-term macrofinancial shocks on banks and insurers, and to assess connectedness risks within and outside the financial sector. ### Key high-level findings - The Japanese financial system appears generally resilient to short-term risks, but pockets of vulnerability exist. - Overall, banks have sufficient capital and liquidity buffers to cope with a scenario of severe recession due to disruptions in global trade accompanied by a sharp increase in interest rates and risk premiums, and a decline in equity prices. - Spillovers within the system appear limited (though analyses do not capture all potential effects during stress times). - Some life insurance companies and regional banks may need to strengthen their capital buffers. ### Banking sector solvency stress tests - Top-down (TD) banking sector stress test covered up to 20 of the largest banks, representing 90 percent of banking system assets. - Under the Severe Adverse Scenario: - Aggregate Common Equity Tier 1 (CET1) ratio would drop to 8.3 percent at its lowest point within the five-year stress horizon. - Main drivers of the fall in capitalization: equity- and market-related losses (including bond valuation effects from higher interest rates), credit losses, lower profitability from overseas lending due to higher U.S. dollar funding costs, and changes in risk-weighted assets. - Regional banks: - Credit-related losses represent the highest risk for regional banks. - Regional banks without international exposure have relatively higher overhead costs and less diversified loan portfolios, showing a slow decline in profitability and capital even under the baseline scenario. - Three regional banks would need additional capital to meet the 8 percent total capital requirement (the number is very small in terms of ratio to GDP). - Loan concentrations and exposures: - Loans to companies are the most important exposure class after sovereign exposure. - Loan portfolio concentration tests show no bank fails to meet minimum CAR in case of default of the single largest borrower, even under extreme loss-given-default assumptions. - Many large exposures are connected across banks (multiple banks exposed to common large borrowers). ### Banking sector liquidity stress tests - Internationally active banks rely heavily on wholesale sources for foreign currency funding; U.S. dollar and euro funding mainly through unsecured wholesale funding, repos, and FX swaps. - Banks have been shifting toward more stable and longer funding sources to reinforce resilience in FX markets. - Liquidity test results: - No bank fails in the yen liquidity stress testing. - A few internationally active regional banks would experience negative cumulative funding gaps in U.S. dollars in the severe stress scenario. - These U.S. dollar funding gaps are relatively small compared to excess yen liquidity; no bank fails in the combined liquidity stress testing of both yens and U.S. dollars. - All-currency liquidity coverage ratios (LCRs) are above 100 percent in all banks in the sample. - System-wide LCR in U.S. dollars stands below 60 percent, and the ratio for regional banks is lower on average. ### Insurance sector solvency stress tests - TD and bottom-up (BU) stress tests were conducted for seven life and six non-life companies covering 73 percent and 92 percent of the respective sectors. - Scenarios were the FSAP two macrofinancial scenarios with front-loaded financial market shocks for the insurance tests. - Impact measured by instantaneous changes in the solvency margin ratio (SMR); statutory requirement (hurdle rate) is 200 percent. - Life insurers: - TD results: average SMR drops from 949 percent before stress to 708 percent in the moderate adverse scenario and to 419 percent in the severe adverse scenario. - Six out of seven life companies remain above the statutory 200 percent requirement under the current regime. - Life companies are strongly sensitive to interest rates. - Non-life insurers: - Less sensitive to higher interest rates and more resilient. - Analyses indicate resilience of the non-life sector toward large natural catastrophes if modeled as single events; domestic perils like earthquakes and typhoons are most relevant. - BU stress tests broadly confirm TD results; BU SMRs on average slightly higher and all companies remain above the 200 percent hurdle. - Valuation and solvency regime caveat: - Economic solvency ratios (ESR) in a recent JFSA field test are considerably lower than statutory SMRs—on average, the life sector had an ESR of only 104 percent as of March 2016. - An economic solvency regime makes life insurers less vulnerable to an interest rate increase but more vulnerable to a prolonged period of low rates. ### Connectedness, contagion, and spillovers - Propagation of financial shocks occurs primarily through client and investor bases; direct exposures play a more modest role in transmission of funding shocks. - Strong links between financial and nonfinancial firms underpin market-based spillovers. - Japan’s nonfinancial firms and city banks are the main source of financial spillovers. - Cross-shareholding and common exposures to foreign markets underpin some spillovers. - Empirical drivers of resilience to spillovers: - Financial firms with stronger balance sheets are less sensitive to financial spillovers. - Among banks, institutions that rely less on wholesale funding or for which credit accounts for a larger fraction of assets are less vulnerable. - Higher shares of institutional ownership (a proxy for better corporate governance) are associated with lower sensitivity to spillovers. - Comparative contagion effects: - Losses to the financial system from a credit shock would be significantly larger than those from a funding shock. - Among funding shocks, those emanating from large city banks have relatively more important consequences, especially for securities firms. - In the case of a default by a large city bank, average capital losses would be six times larger than under a funding shock, and life insurers would suffer most. - High levels of capital held by many Japanese financial institutions help dampen credit and funding contagion. ### Interpretation and caveats - Stress test scenarios are calibrated to extreme but plausible “tail events” based on historical events or expert judgment; crises can involve unanticipated shocks and interrelationships not fully captured by history. - Stress test and quantitative risk assessment results should be interpreted with caution. ### FSAP main recommendations on stress testing and systemic risk assessment (selected from Table 1) - Continue assessing the impact of default by largest borrowers on the whole banking system (JFSA). — Time: I - Implement an in-house multi-year top-down scenario stress testing framework for banks and insurance companies (JFSA). — Time: NT - Continue improving data consistency, especially between FX maturity mismatch and LCR (JFSA). — Time: I - Continue conducting liquidity stress testing regularly with significant foreign currencies and require banks to hold sufficient counterbalancing capacity, particularly HQLA (JFSA). — Time: I - Intensify supervision on FX funding liquidity risk by i) considering caps on funding concentration for significant foreign currencies; and ii) deducting from HQLA the assets subject to ringfencing by foreign jurisdictions (JFSA). — Time: NT - Collect data and adopt analysis of interconnectedness as toolkit of financial supervision (JFSA). — Time: I - The collection of data about exposures between financial firms should continue on a regular basis (JFSA). — Time: I - Use insurance stress test results also to validate and benchmark ORSA reports and to gain further insights into the economic solvency regime (JFSA). — Time: NT *Source: EXECUTIVE SUMMARY, JAPAN, INTERNATIONAL MONETARY FUND.* ### INTRODUCTION ### INTRODUCTION ### A. Financial System Structure - Japan has one of the largest and most sophisticated financial systems in the world. As of September 2016, total financial assets held by the financial sector reached about 620 percent of GDP in 2016, compared to 675 percent in the Euro Area and 462 percent in the United States (U.S.). - Financial conglomerates: - The twenty conglomerates make up about one third of total financial assets or about 170 percent of GDP. - The four largest financial groups (FGs) hold banks, trust banks, and securities firms, and account for 136 percent of GDP. - Twelve regional bank holding companies have combined assets of 21 percent of GDP. - Other than Japan Post Insurance, none of the large insurers is part of a group that also includes significant banking activities. - Asset distribution: - More than half of total financial assets are held by commercial banks. - Insurance companies: 15 percent of total assets. - Pension funds: 8 percent. - Securities firms: 5 percent. - Investment trusts: 6 percent. - Banking sector structure: - Sector consists of city banks (three mega banks classified as G-SIBs), trust banks, regional banks, Shinkin banks (credit unions), credit associations, and credit cooperatives. - The three mega banks account for about 18 percent of assets. - Regional banks: 14 percent. - Shinkin banks: 5 percent. - Credit associations and credit cooperatives: about 5 percent. - Japan Post Bank accounts for about 8 percent. - Deposits represent the largest source of funding for banks. - Foreign banks have a very small market share and are mostly involved in investment- and private banking and financial derivatives trading. - Insurance sector: - Japan’s insurance sector is the world’s second largest after the U.S. - Life insurance accounts for about 90 percent of the sector, with total financial assets of about 75 percent of GDP. - The five largest life insurers—three mutual—account for almost 70 percent of life insurance sector assets. - The four largest non-life insurers represent near 90 percent of non-life insurance sector assets. - Japan Post Insurance is the largest life insurer, with 22 percent market share by assets. - As of March 2016, in the life insurance sector new business premium composition: annuities 28 percent and medical insurance 23 percent. - More than 80 percent of annuities are fixed-term annuities, entailing little longevity risk for life insurers. - In the non-life sector: motor insurance is 43 percent of total premium income; fire insurance contributes 17 percent. - Securities markets: - Equity market capitalization of about 100 percent of GDP. - Japan Exchange Group’s (JPX) Tokyo Stock Exchange (TSE) is the third largest exchange in the world after the U.S. New York Stock Exchange and NASDAQ. - Derivatives trading is less significant; Osaka Exchange ranked number 16 in the world in 2015 with about a tenth of the trading volume of CME Group. - Corporate bond market is 17 percent of GDP. - Stock investment trust increased from 3 percent of GDP in 2013 to 7 percent in 2017. - Securities firms: - Heterogeneous group; largest five firms—three subsidiaries of the megabanks—are major global players. - Many commercial banks have securities subsidiaries, but market share of these firms is small. ### B. Stress Testing Under the FSAP Program - Objective: - Assess the resilience of the whole financial sector rather than the capital adequacy of individual institutions. - Macroprudential focus: resilience of the broader financial system to adverse macrofinancial conditions. - Ensure consistency in macroeconomic scenarios and metrics across firms. - Identify key sources of systemic risk and inform macroprudential policies. ### C. Stress Testing Approach for the Japan FSAP - Scope and methods: - Wide range of stress tests assessing systemic solvency, liquidity, and contagion risks in banking and insurance sectors. - Macroprudential approach with both bottom-up (BU) and top-down (TD) analyses by IMF and BoJ. - Included both banking and insurance sectors. - Banking sector TD stress test: - Covered 20 large banks, comprising 90 percent of banking system assets. - IMF TD solvency stress test: five-year projections of earnings, costs, balance sheets, and impact on (risk-weighted) assets and capital. - BoJ conducted a separate TD stress test using its own models, following IMF scenarios; BoJ TD results are based on three-year projections. - Liquidity and contagion analyses used IMF TD approaches and models; supervisory, market, and firms’ proprietary data employed. - Additional simulation integrated liquidity stress test results into solvency stress testing. - Insurance sector tests: - TD and BU stress tests for seven life and six non-life companies, covering 73 percent and 92 percent of the respective sectors. - Impact measured as instantaneous changes in the solvency margin ratio (SMR); statutory requirement of a 200 percent SMR is the hurdle rate. ### Banking Sector Solvency Stress Tests — A. Macroeconomic Scenarios - Framework: - Risk Assessment Matrix (RAM) is the basis for stress testing scenarios. - Three five-year macroeconomic scenarios: one baseline, one moderate adverse (“De-Globalization”), and one severe adverse (“Accelerated U.S. Monetary Policy Normalization”). - Scenarios differ in risk factors, transmission channels, and magnitudes of shocks. - Stress testing scenarios simulated using the Global Macrofinancial Model—a DSGE model disaggregated into forty national economies. - Scenario descriptions: - Baseline: - Largely based on projections from the October 2016 World Economic Outlook (WEO). - Reflects continued weak growth supported by the recently announced fiscal stimulus package and private consumption. - Global growth uncertainty and yen appreciation expected to pose a drag in the near term. - Growth will fall below potential in 2020 as the fiscal package ends and the consumption tax hike is implemented in October 2019. - Inflation will gradually pick up over the medium term, but remain below the BoJ’s inflation target of 2 percent. - Short-term interest rates are expected to remain close to zero, and long-term rates will only marginally rise over the medium term, reflecting the BoJ’s yield curve control. - Moderate adverse (“De-Globalization”): - Features protectionism and economic isolationism, reduced global and regional policy collaboration, a significant credit slowdown in emerging market economies (EMEs), and an initial depreciation followed by a tightening in domestic financial conditions. - Compared to baseline: higher money market interest rates, lower equity prices, and a correction in the domestic housing market. - The sharp decline in global growth (which at the peak drops by 1.6 percentage points globally on average compared to the baseline) increases banks’ credit risks in overseas and domestic loan portfolios. - Growth drops by 1.8 percentage points in advanced economies and by 1.0 percent in emerging market economies. - Severe adverse (“Accelerated U.S. Monetary Policy Normalization”): - Features a 200 basis-point policy interest rate increase in the U.S. during 2017–18, leading to capital outflows and tightening of financial conditions in Japan, including higher JGB yields, higher money market interest rates, and lower equity prices. - Assumes the initial JGB yield spike triggers a loss of confidence, resulting in a further increase in JGB yields; the total increase in the JGB yield in this scenario is about 300 bps. - Higher bond yields lead to significant valuation losses on JGB holdings. - Correction in domestic housing market is also assumed. - Credit risks at home and abroad rise due to the increase in interest rates (peaking at 1.4 percentage points for long maturities globally on average). - Combined solvency and liquidity scenario: - Assumes banks needing to improve capital buffers under the severe adverse scenario would face higher liquidity shocks (details in source pages 38−46). *IMF staff and Japan Financial Services Agency materials as presented in the INTRODUCTION.* ### 12. In the moderate and severe adverse scenarios, the real GDP growth in Japan follows a ### 12. In the moderate and severe adverse scenarios, the real GDP growth in Japan follows a ### Scenario impacts on real GDP and historical comparison - Real GDP growth follows a U shape in the moderate adverse scenario and a V shape in the severe adverse scenario. - In the moderate adverse scenario: real GDP growth falls less and takes longer to recover after the initial shocks. - In the severe adverse scenario: real GDP falls more and recovers relatively quicker. - Japan’s real GDP declines more in the severe adverse scenario than in the moderate adverse scenario, primarily due to the higher JGB yields triggered by the accelerated U.S. monetary policy normalization. - The initial declines in real GDP levels in the moderate and severe adverse scenarios are comparable to those during the Asian financial crisis (1997–98) and the global financial crisis (2007–09). - The cumulative losses in the two adverse scenarios over the five-year horizon are larger than those in the two historical crises, when the recovery in real GDP growth was relatively faster than in the scenarios. - The increase in long-term rates peaks at 1.6 percentage points for advanced economies on average, and at 0.8percentage points for emerging market economies. ### Macroeconomic and financial projections and model capabilities - The Global Macrofinancial Model (GFM) captures macrofinancial feedback effects between the real economy and financial sector (particularly banking sector and capital markets). - Interactions between credit cycle and business cycle lead to deleveraging by banks and hence lower credit growth in both adverse scenarios. - Projected variables shown include: real GDP, real GDP growth, CPI inflation, 10-year government bond yield, exchange rate, unemployment rate, equity price, and real house price. - Note: The projections for GDP were made before the National Accounts revision. - Note: A few key financial variables that are not projected by the WEO database including equity price and real house price are assumed to remain constant in the baseline scenario. ### Coverage of banks and sample composition - Banks were grouped into three categories to highlight differences: city (seven banks), regional (nine banks), and domestic. - Japan Post Bank was treated as a separate entity due to its size (almost 18 percent of banking system assets), business model, and analysis used only public data. - Sample composition shares (in percent, excludes Japan Post Bank): 86.10; 8.30; 5.60 (City banks; Regional banks; Domestic banks). ### Modeling approach and timing of data - An array of satellite models projected banks’ income, expenses, expected losses, and balance sheet adjustments; outputs fed into the solvency template. - Data basis: September 2016 balance sheets, and March 2016 for income statements. Historical data went back as far as 2002 and included the Global Financial Crisis (2008–09). - IMF’s proprietary Solvency stress testing tool projected banks’ balance sheet, losses, RWAs, and capital evolution for 2017–21. ### Key methodological assumptions for balance sheet and profit projections - Growth of banks’ balance sheets: - Projected to grow in line with macroeconomic scenario projections, albeit with a cap on deleveraging. - Cap prevents banks from meeting capital requirements simply by excessively shrinking balance sheets or reducing RWAs. - In scenarios that incorporate liquidity and solvency risks, assumptions were made about asset liquidation strategies and respective shrinkage of the balance sheet. - Maximum limit of deleveraging is capped at negative GDP (real or nominal, whichever is lower) growth rate. - Composition of balance sheet assets and liabilities: - Composition may change due to different growth rates between Japanese banks’ domestic and foreign investments. - Growth rate linked to credit growth rate (same for each bank) and decomposed based on relative shares of foreign assets. - Focus on two key markets: Japan (domestic) and U.S. (foreign). Foreign exposures change with yen/U.S. dollar exchange rate. - Liabilities grow in line with assets, assuming non-interest bearing assets and liabilities remain constant. - Projection of risk-weighted assets: - Largest banks operate under Basel II/III Internal Rating-Based approach; risk weights projected using Basel II formulas depending on PD, LGD, maturity, and asset correlation. - Loan portfolio loss projections: - Use shocked exposure weighted average TTC PDs, shocked downturn LGDs, and adjusted EADs to project expected losses. - Evolution of profits: - Income statement items and non-interest profit items (operational and administrative expenses, net fee and commission income) projected via satellite models. - Distribution of dividends: - Undercapitalized banks in any year were not allowed to distribute dividends. - Banks that pass all minimum requirements distribute dividends according to a 30 percent rule (average dividends rate calculated from BU stress testing submissions). - Hurdle rates used to calculate potential recapitalization needs: - CET 1 of 4.5 percent (applied to City- and Regional banks with international exposure). - Total capital ratio of 4 percent applied to other regional banks without international exposure. - Tier 1 ratio of 6 percent, as well as 8 percent of total capital ratio, applied as additional thresholds for City- and Regional banks with international exposure. - A 3 percent leverage ratio (Tier I capital to total assets) was used as a potential benchmark (proxy used since off-balance sheet items, derivatives etc., were not included). ### Credit risk: composition, exposures, and modeling - Balance of banking sector assets: - Loans represent around half of total banking sector assets. - Debt securities represent 20 percent (most are marked to market). - Exposures are still predominantly domestic, but loan activity in foreign markets (U.S., EU, Emerging Asia) has increased. - IMF TD stress testing focused on aggregated loan portfolio data without separating exposures by region due to consolidation and data availability; BoJ’s TD model and BU STs included all key regions. - Use of IRB approach: - IRB approach dominant; own sovereign risk exempted with zero risk weight treatment. - Zero risk weight treatment leads to very low capital requirements for some smaller banks with large JGB investments, high cash/liquid assets, and low regulatory requirements (4.5 percent). - Retail mortgages are a relatively small item; associated capital requirements are low due to low PDs and LGDs. - Capital requirements associated with equity risk are relatively high; almost all banks use IRB approach with fixed 90 percent LGD. - Banks commonly use a 50 percent drop in equity prices scenario to calculate stressed losses and capital needs. - Distribution of banks’ exposures by asset class (In percent): - 5.4; 0.5; 2.0; 39.1; 1.6; 7.9; 43.5 (order corresponds to Qualifying retail; Equity; Corporate; Retail; Residential mortgages; Sovereign; Bank exposures as presented). - Distribution of banks’ exposures by Risk-Weighted Assets (In percent): - 4.8; 1.2; 11.9; 68.8; 2.6; 9.1; 1.7 (order corresponds to Qualifying retail; Equity; Corporate; Retail; Residential mortgages; Sovereign; Bank exposures as presented). ### Credit risk satellite models, PDs and special treatments - Satellite models estimated transmission of macro shocks to PDs and loan loss provisions for six broad exposure classes; time series started in 2007 and covered one credit cycle (defaults increased above means in 2011–13). - PDs used for satellite models were Through-The-Cycle (TTC) without defaulted exposures; time series also included PDs with defaulted exposures which are more volatile. - TTC PDs in practice represent “hybrid” PDs with around 90 percent weight on historical estimates and 10 percent on annual update for Japanese banks. - PDs for low-default portfolios (foreign sovereigns and institutions) were challenging to model with macro variables; BuDA methodology used for institution asset class to project one-year PDs up to five years using macro scenarios. - In baseline scenario, PDs assumed to remain constant. - BuDA PDs behave similarly to EDFs; regressors include equity prices and 3-month Japan Treasury bill rates. - Scenario drivers for PD increases: - Moderate adverse: increase in PDs driven by decline in equity prices and negative risk free rates (reducing banks’ profitability). - Severe adverse: increase in PDs driven by decline in equity prices, decline in GDP, and a flattening of the yield curve (reducing banks’ profits). - For sovereign exposure class, PDs were extremely low historically; in most cases below 0.03 percent (Basel II/III minimum PD for certain exposure classes). - A panel model showed some sensitivity of foreign sovereign PDs to long-term interest rates (10-year JGB yields) and was used to capture increases in risk for the foreign sovereign exposure class. *Source: IMF staff estimates and IMF’s WEO and Global Assumptions databases (projections for GDP made before the National Accounts revision).* ### 24. Corporate PDs are sensitive to real GDP, equity prices and interest rate developments, ### 24. Corporate PDs are sensitive to real GDP, equity prices and interest rate developments, while retail portfolios—to unemployment, GDP and interest rates. ### PD and LGD sensitivity and modeling approach - Separate satellite models estimated for corporate, residential mortgages, retail, equities. - Estimation results: the low interest rate environment supports historically low corporate and retail PDs, and that an abrupt rise in interest rates would lead to significant increase in defaults among corporates. - Mortgage loans remain less sensitive to a macroeconomic downturn, mostly because of relatively sticky unemployment and little impact from a decline in real estate prices. - LGD models used historical time series of downturn LGDs for corporate and residential mortgage exposure classes. LGDs for some exposure classes, like equities and retail loans are very high, and the FSAP team assumed no further shocks on them in adverse scenarios. - Average LGDs and estimated shocked LGDs are provided in Figure 10 (source: IMF staff estimates). - PD estimation methodology: - Bayesian Model Averaging (BMA) used to select significant macrofinancial variables and lags (posterior probability of inclusion higher than prior probability of inclusion; prior probability calculated by limiting maximum size of the model to three explanatory variables). - Variables significant under BMA were used to construct separate equations estimated using panel data OLS and GMM methods. - PD/LGD forecasts made for each bank; estimates across multiple models were averaged to obtain average PD/LGDs for each bank. - Median PDs/LGDs were calculated and respective multipliers constructed; PD/LGD multipliers were used to project evolution of risk factors by multiplying the last observed PD/LGD value from each bank by obtained multiplier. - PCA model used to simulate full yield curve behavior (30-years, 5-years, 2-years, 1-year JGB yields; 12-month, 3-month, 1-month TIBOR; long-term and short-term prime lending rates). PCA principal components: first = parallel movement, second = slope, third = curvature. ### Credit risk and expected losses (key figures and projections) - IMF TD stress test results: banks likely to experience significant increases in PDs under the severe stress scenario; PDs remain almost flat under the baseline scenario. - Median corporate PD: - from 0.5 percent in 2015 to 2.3 percent in 2019 under the severe adverse scenario. - peak of 1.1 percent under the moderate adverse stress scenario. - Expected credit losses over five-year horizon in the loan book: - Severe stress scenario: ¥12 trillion, equivalent to 1.2 percent of total banking system assets. - Baseline scenario: ¥4 trillion, equivalent to 0.4 percent of total banking system assets. - In the baseline scenario, new provisions (¥4 trillion) are more than offset by net income before losses (1.6 percent). - PD behavior by portfolio: - Corporate portfolio: expected losses in the adverse scenario are mostly driven by losses in the corporate portfolio. - Residential mortgage loans: PDs do not exhibit a sharp increase (partly because banks typically have full recourse on these loans). ### Market risk, portfolios and valuation impacts (key figures) - Debt and equity securities portfolios covered in trading book (HFT) and available for sale book (AFS). - Debt securities portfolio among the nineteen major banks: - around ¥165 trillion or 18 percent of total assets. - 76 percent domestic bonds (Japan government bonds dominate); remainder foreign securities predominantly U.S. government debt. - Average duration: 2.5 years (varies across banks; regional banks may have much longer duration). - Equity securities: - value in banks’ balance sheets: ¥16 trillion or 1.8 percent of total assets. - share of equity securities to total assets among banks ranges from 0.2 percent to almost 4 percent. - Market shocks application: - Effect of increase in domestic and foreign interest rates and JGB yields applied to domestic and foreign debt securities. - TOPIX and NIKKEI indices used as benchmarks to shock equity portfolios; market shock applied as instantaneous shock to all equity positions for each year of the horizon. - Relative magnitudes: - Equity- and bond valuation effects during the three worst years in the stress equal to losses of around 1 percent of assets. - Credit risk related losses equal to 0.9 percent of assets. - Market risk from commodities is negligible; market risk is mostly due to term premium shocks across debt markets and sharp decline in equity prices. ### Solvency stress test results and vulnerabilities - Aggregate solvency results: banking system broadly resilient, but some banks are particularly vulnerable to extreme shocks. - Under the severe adverse scenario: - Aggregate Common Equity Tier 1 (CET1) ratio would drop to 8.3 percent. - Small capital shortfall remains manageable at the macro level. - Three regional banks would need additional capital to meet 8 percent total capital requirement. - Equity and other market risk losses represent the most important risk factor for the large banks included in the stress tests. Credit losses, lower profitability from overseas lending (higher U.S. dollar funding costs), and change in RWAs are also important. - Stress drivers and dependencies: - Banking system vulnerable to correlated market- (equity) and credit risk shocks. - Results reflect large investments into equities, interest rate risk from large holdings of debt securities, and reliance on wholesale funding in U.S. dollars. - Regional/domestic bank sensitivities: - Regional banks maintain highest sensitivity towards credit risk related losses; domestic banks have relatively higher overhead costs and less diversified loan portfolios and exhibit slow decline in profitability and capital even under baseline. - City banks experience smaller shocks to corporate PDs; regional banks experience larger shocks. - Recovery assumptions: estimated recapitalization needs are conditional on a recovery in interest margins and a gradual increase in equity prices embedded in the severe adverse scenario. - System-wide CAR dynamics: while system-wide CAR ratio drops by almost 50 percent, high average CAR buffers imply total potential recapitalization needs are small because banks needing additional capital have small market share and would not fall below CET1 minimal capital requirement. ### Concentration and counterparty sensitivity - Loan portfolio concentration tests: - No bank fails to meet minimum CAR in case of default of the single largest borrower even under 100-percent-LGD assumption. - Default of five largest borrowers leads to 4 banks falling below minimum CET1 CAR. - Default of 10 largest borrowers leads to 12 banks falling below minimum CET1 CAR. - Many large exposures are connected (multiple banks have the same large borrower). - Default of the largest borrower in the system would lead to loss of 4.5 percent of system’s total capital, though no bank would be undercapitalized. - BU sensitivity tests (four banks): - Residential and commercial real estate price drop by 20 percent would not lead to undercapitalization of these banks (impact on their capital would be 0.14 percent of combined CET1). - Default of the two counterparties with the lowest credit rating would lead to small losses for the four banks (less than 0.01 percent of combined capital). - A 50 percent equity price drop would lead to capital drop by almost two percentage points. *Source: IMF staff estimates and IMF FSAP team analysis as presented in the referenced chapter.* ### 41. Results from the BoJ’s TD stress testing results also confirm that the banking system ### Results from the BoJ’s TD stress testing results also confirm that the banking system ### Solvency stress testing — top-down (TD) key findings - System-wide vulnerability is concentrated in market (equity) and credit risk shocks. - Losses stemming from cross-shareholdings of equity securities of borrowing companies are the largest driver of the system-wide decline in capital adequacy ratios. - Sovereign risk from Japanese and foreign sovereign valuations is a smaller driver in relative terms. - While all banks included in BoJ’s TD exercise remain above the minimum CET1 or core capital requirements, the U.S. monetary policy normalization scenario leads to the highest losses in the system. - Domestic banks CAR dropping from 13 percent in 2016 to 7.7 percent in 2018 under the U.S. monetary policy normalization scenario. - Small regional banks are more vulnerable to credit risk (decline in GDP growth) than their large peers. - Note on equity losses: equity losses are twofold—an increase in probability of default leads to higher expected losses, RWAs and deductions from other comprehensive income. Only impairment of equities is directly included into profit and loss statement. - Group definitions used in TD results: - Group 1 stands for city banks. - Group 2 for internationally active regional banks. - Group 3 regional banks without international exposure. ### Bottom-up (BU) stress testing — findings and comparisons - BU stress test results reveal similar outcomes: most vulnerability lies in market and credit risk related losses, with a relatively higher impact from market risk in BU tests. - Banks’ BU estimates of market risk related losses are more granular than IMF and BoJ TD models: - For equity risk, banks used TOPIX and NIKKEI equity indices as benchmarks to simulate drop of price for each individual equity security. - Losses from debt securities were calculated using duration gap analysis; banks and BoJ estimated the precise impact on each type of security in banks’ portfolios (more granular than IMF TD approach). - Comparison across tests: - Overall only small differences among TD (IMF), TD (BoJ), and BU results. - City banks forecast a higher impact of the severe adverse scenario on CET1 capital in BU results. - Losses from credit risk were higher in the IMF model—attributed in part to granularity differences: IMF used weighted average TTC PDs for each exposure class, while banks calculated ELs in a more granular manner. - Scenario design note: Scenario assumptions restricting changes in portfolio composition worsened the BU ST results. Actual bank behavior might rebalance portfolios to minimize shock impacts, but rebalancing may be difficult during systemic stress; thus behavioral constraints in BU design might not overestimate losses. ### Liquidity stress testing — funding structure and concentration - Banks rely heavily on wholesale sources for foreign currency funding (U.S. dollars and euros), unlike yen funding which has greater retail components. Much FX funding is unsecured wholesale funding, repos, and FX swaps that could be difficult to obtain in stress. - Some FX funding has maturities longer than one year (medium- to long-term FX and cross-currency swaps), but a large share (CDs, repos, short-term FX swaps) are typically rolled over and repriced every three to six months. - System-wide share of FX funding: - U.S. dollar and euro funding account for less than 25 percent and 5 percent of total funding in all currencies, respectively. - Funding concentration indicators and findings: - Two indicators used: (i) the Herfindahl index; and (ii) the top-five largest funding sources as a share of total funding. - Indicators calculated by currency (yen, U.S. dollar, euro) and aggregated by bank type (city banks, internationally active regional banks). - Findings: - Both types of banks are more concentrated in FX funding, particularly euro funding (though inclusion of CCPs in secured funding may affect comparability). - Regional banks are less concentrated in yen funding than city banks, but are highly concentrated in euro funding. - Recommendation: The FSA could consider raising the intensity of its supervision over banks’ funding concentration in significant foreign currencies. - Note: A foreign currency is considered “significant” if aggregate liabilities denominated in that currency amount to 5 percent or more of the bank’s total liabilities. ### Liquidity stress testing — methodology, data coverage, and scenarios - Two types of liquidity stress tests used: - LCR-based analysis (standardized 30-day stress embedded). - Fully fledged maturity ladder cash flow stress tests (currency-level, longer horizons). - Data coverage: - Supervisory data included bank-solo basis and FG-consolidated basis. - Bank-solo basis: 16 banks (seven city banks and nine regional banks); contractual cash outflows/inflows broken down by maturity buckets and counterbalancing capacity in yen, U.S. dollar, euro. - FG-consolidated basis: 14 banks (five city banks—including three FGs—and the nine regional banks); similar variables as bank-solo. FG-consolidated findings are presented because FGs typically manage liquidity at group level. - Maturity ladder and scenario structure: - Cash flow approach applied at currency level (yen, U.S. dollar, euro). - Market-wide stress scenarios per currency consist of three severity levels and durations: - Mild stress: 2 weeks. - Medium (intermediate) stress: 3 months. - Severe stress: 1 year. - Two levels of BoJ support considered in yen analysis: no support (pure market) and support through BoJ’s discount window (lower asset haircut ratios than market). - Market liquidity shocks to equities considered to estimate haircut ratios (haircuts are dynamic and depend on total equities sold; methodology in Appendix V). - For U.S. dollar analysis, assumed increases in run-off rates on outflows from FX swaps across severity levels; same rates used as roll-off rates for yen cash inflows from FX swaps (other leg of U.S. dollar/yen swaps). - Combined USD–yen analysis assumes banks may use yen-denominated counterbalancing capacity to offset U.S. dollar funding gaps. - For unconditionally callable funding (stocks rather than flows), run-off rates (in percent of stocks) assumed to broadly follow two-lag autoregressive (AR(2)) processes. - Example assumption for cumulative share of total cash outflows over 1 year used in run-off illustration: - 5 percent in the 2-week mild scenario. - 7 percent in the three-month intermediate scenario. - 10 percent in the 1-year severe scenario. - Scenario parameter assumptions (selected exact values): - 2-week mild scenario: - Run-off rates for yen-denominated retail and term deposits calibrated from recent historical cases; largest run-off rate in range of 3–5 percent used for yen retail and term deposits. - Run-off rates for unsecured wholesale funding (including corporate deposits) assumed 30 percent. - Run-off rates for secured wholesale funding assumed same as LCR run-off rates; run-off for secured wholesale funding from domestic central bank assumed 0 percent. - Run-off rates for secured wholesale funding from non-central bank sources assumed 0 percent, 15 percent, and 50 percent for those backed by level 1, level 2A, and level 2B HQLA, respectively. - In mild U.S. dollar and euro scenarios, assumed 40 percent of FX swap contracts cannot be rolled over. - 3-month intermediate scenario: - Run-off rates on unsecured wholesale funding assumed 40 percent. - Run-off rates on retail deposits assumed 5–10 percent. - Cash outflows from secured wholesale funding assumed same run-off rates as 2-week scenario (or LCR run-off rates). - Assumed 60 percent of FX swaps cannot be rolled over. - 1-year severe scenario: - Run-off rates for wholesale funding secured by level 2A and 2B HQLA assumed 30 percent and 80 percent, respectively. - In severe U.S. dollar and euro market scenarios, FX swap market assumed under severe stress: - 80 percent run-off rate of cash outflows from U.S. dollar-denominated derivatives. - 80 percent roll-off rate for cash inflows from yen-denominated derivatives. - Run-off (or roll-off) rates for derivatives increased to 100 percent in market and idiosyncratic liquidity scenario for banks whose capital ratios fall below regulatory levels after the first year in solvency stress testing. - Maturity buckets covered in cash flow data: overnight; 2 days to 1 week; 1 week to 2 weeks; 2 weeks to 1 month; 1 month to 2 months; 2 months to 3 months; 3 months to 6 months; 6 months to 9 months; 9 months to 1 year; 1 year to 2 years; 2 years to 3 years; 3 years to 5 years; and over 5 years. Maximum horizon of liquidity stress scenarios is 1 year. - An idiosyncratic stress scenario on top of the severe stress scenario is considered for banks whose capital ratios fall below regulatory levels after the first year in solvency stress testing. *International Monetary Fund staff estimates from the Japan FSAP stress testing chapter.* ### 53. LCRs analysis shows that banks have ample liquidity in yen but not in foreign ### cr17285 - 53. LCRs analysis shows that banks have ample liquidity in yen but not in foreign ### Liquidity Coverage Ratios (LCRs) — all currencies vs. U.S. dollar - All-currency aggregated LCRs: ratios above 100 percent in all banks in the sample as of end-December 2016. - System-wide LCR in U.S. dollars: stands below 60 percent. - Bank-type differences: the ratio in regional banks is lower on average than that in city banks. - Distributional statistics reported (FG-consolidated; as of end-December 2016): - City banks: Median (131%); Average (142%). - Regional banks: Median (134%); Average (144%). - All-currency LCRs show minimum, maximum, 1st quartile, 2nd quartile measures in Figure 18. ### Cash flow–based liquidity stress tests — foreign currency vulnerabilities - Yen liquidity: no bank fails in the yen liquidity stress testing. - U.S. dollar liquidity (cash flow-based): - A few internationally active regional banks would experience negative cumulative funding gaps in U.S. dollars after exhausting unencumbered U.S. dollar liquid assets in the severe one-year scenario. - Number of banks with negative cumulative funding gap in U.S. dollar shown for 2-week, 3-month, 1-year scenarios (FG-consolidated; as of end-December 2016); Figure 19 indicates up to several banks in the 1-year scenario. - Euro liquidity: - Euro-denominated liabilities account for less than 5 percent of total liabilities for the whole banking sector. - Negative yet very small funding gaps (relative to total funding needs) in two regional and two city banks in the severe 1-year scenario. - Scenario robustness: - Results using end-March and end-September 2016 FG-consolidated data are comparable with those using end-December 2016 data in the 1-year severe scenario. ### Cross-currency offsetting and combined stress tests - Combined yen and U.S. dollar liquidity stress testing assumption: banks can convert yen-denominated liquid assets to U.S. dollars in the market. - Result: since U.S. dollar funding gaps are relatively small compared to excess yen liquidity, no bank fails under the joint stress conditions in both yen and U.S. dollar funding markets. - City banks: all city banks can survive the 2-week mild stress at the FG-consolidated level. ### Funding sensitivity and unsecured wholesale funding - Regional banks are more sensitive to cash outflows from unsecured wholesale funding. - Unsecured wholesale funding is typically more sensitive to negative market developments and may dry out completely in extreme cases. - Applying higher run-off rates on outflows from unsecured wholesale funding leads to more failures of regional banks than applying the same run-off rates on outflows from secured wholesale funding. ### Policy implications and supervisory recommendations on FX funding liquidity risks - Intensify banking supervision on FX funding liquidity risks. - Improve consistency between data from the FX maturity mismatch template (maturity ladder data) and the LCR data. - JFSA should monitor the level of HQLAs which are not ringfenced or “trapped” in other jurisdictions (e.g., by the regulation W in the U.S.) through ordinary supervision. - Conduct cash flow–based liquidity stress testing on a regular basis using the maturity ladder data. - Require banks to hold sufficient counterbalancing capacity (particularly HQLAs) to survive for a certain period (e.g., three months). ### Integrated solvency and liquidity stress testing — approach and key assumptions - Integration rationale: banks needing additional capital because of credit losses would face higher liquidity outflows due to investor concerns about solvency. - Timing and triggers: the analysis followed an annual sequence; solvency shock happens in 2017; liquidity stress spans one year. - Asset liquidation strategy choices for banks: - Liquidate assets in Japanese yen or U.S. dollars; choose between liquid and less liquid assets. - If less liquid assets are liquidated (equities, corporate bonds, commercial papers), additional haircuts applied using Amihud (2002) measure of stock market liquidity. - Assumed behavior after exhausting unencumbered U.S. dollar assets: - Banks liquidate JGBs or use yen-denominated cash or central bank reserves to offset U.S. dollar funding gaps. - Losses mainly come from fire sales of unencumbered non-HQLA assets denominated in U.S. dollars (non-HQLA CPs, agency bonds, corporate bonds). - Increased run-off rate assumptions for “failed” banks (as specified): - Stable retail deposits: increased from 5 percent to 10 percent. - Less stable retail deposits (or term deposits): increased from 10 percent to 20 percent. - Wholesale corporate funding outflows: increased from 40 percent to 80 percent. - Debt securities outflows: increased from 70 percent to 100 percent. - Wholesale funding secured by level 2A assets: increased from 30 percent to 50 percent. - Wholesale funding secured by level 2B assets: increased from 80 percent to 100 percent. - Cash outflows from derivatives: increased from 80 percent to 100 percent (to reflect closure of FX swap market for weak banks). ### Integrated stress test results — impacts and magnitudes - Overall system-wide capitalization impact: limited due to ample liquidity in JPY. - Affected banks face higher capital shortages compared to solvency scenarios alone. - Several banks failing U.S. dollar liquidity stress tests need to liquidate less liquid non-HQLA assets. - Assumed liquidation strategy: banks choose to liquidate foreign (U.S. dollar) assets rather than use swap market to exchange yen to U.S. dollars. - Losses from liquidation: - Around 0.02 percent in terms of GDP. - Or 0.05 percent of GDP to recapitalize banks up to 8 percent total CAR. - Overall impact on CAR ratios: limited; affected banks can remain liquid in JPY after liquidating U.S. dollar portfolios. - Effects on credit contraction: negligible (shortages are in U.S. dollars, not Japanese yens). - Rationale: banks facing U.S. dollar liquidity shortages hold relatively liquid U.S. dollar assets compared to loans or other non-marketable assets. ### Additional notes on integration assumptions and limitations - Integration started from a solvency trigger because Japanese banks enjoy ample JPY liquidity. - The test explicitly assumes liquidation of foreign assets rather than reliance on the swap market; scenario may be extreme for idiosyncratic shocks but plausible in market-wide U.S. dollar shortages. - Technical cutoff: analysis ended in 2018 due to limited further impact on banks’ CAR and liquidity positions. *Source: IMF staff analysis and Japan Financial Services Agency data as presented in the referenced content unit.* ### 67. The stress test was performed by applying the current solvency regime and J-GAAP ### cr17285 - 67. The stress test was performed by applying the current solvency regime and J-GAAP ### Stress test methodology and regulatory context - The stress test applied the current solvency regime and J-GAAP and focused on the change in available capital and the resulting coverage of the required solvency margin. - The Japanese solvency regime defines a 200 percent solvency margin ratio (SMR) as the first trigger for early remedial actions. - SMR is the ratio of solvency margin (net assets and special reserves based on accounting balance sheet, excluding items such as intangible assets) to half of risk requirements calculated based on standard factors stipulated in subsidiary legislation. - A 200 percent SMR means that the “margin” exactly covers the “risks.” - Management actions were allowed only to the extent they relate to non-discretionary rules already in place at the reference date. - Japanese insurance companies hedge interest rate risks and currency risks to different degrees. ### Investment exposures and asset characteristics (as of March 2016) - Life insurers: - Invested 44 percent of their assets in mostly long-duration JGBs as well as bonds of municipalities and public sector entities. - Invested 28 percent in corporate bonds. - Fair values exceed book values by 20 percent (JFSA data for the life insurers in the stress test sample). - Unrealized gains composition: 65 percent attributable to domestic bonds, 20 percent to domestic stocks, and 14 percent to foreign securities. - Share of assets not denominated in JPY: 21 percent; share of FX-denominated liabilities: 7 percent. - Most companies hedge currency risk largely via forwards and options with domestic and foreign banks as counterparties. - Non-life insurers: - Invested 20 percent in government and semi-government bonds, 32 percent in corporate bonds, and 22 percent in equities. - 73 percent of assets are invested in Japanese yen, which exceeds the value of Japanese yen-denominated liabilities. - Asset valuation is more in line with fair values because hold-to-maturity classification is of lesser relevance. ### Top-Down (TD) stress test results - Life sector: - Average SMR: 949 percent before stress; 708 percent in the moderate adverse scenario (scenario 1); 419 percent in the severe adverse scenario (scenario 2). - Six out of seven companies remain above the statutory requirement under the current regime. - After the severe adverse scenario, half of the life companies would have an SMR in the range between 241 percent and 442 percent. - Aggregated available capital of the seven life insurers: ¥34.4 trillion before stress; declines by 37 percent in the moderate adverse scenario and by 58 percent in the severe adverse scenario. - Required capital declines by 14 percent in the moderate adverse scenario and by 12 percent in the severe adverse scenario (calculated on the basis of a smaller investment portfolio). - Non-life sector: - Average SMR declines from 736 percent pre-stress to 625 percent in scenario 1 and 481 percent in scenario 2. - All non-life companies remain above the 200 percent threshold. - Main drivers of changes in available capital: - Interest rate shocks (domestic and foreign) and decline in equity prices account for more than three quarters of the decline in available capital in the severe adverse scenario. - Assumed default of the largest banking and nonbanking counterparty has a considerable impact. ### Interpretation: solvency regime implications - The large sensitivity to interest rate hikes is typical under the current Japanese solvency regime. - Under an economic solvency regime: - Life insurers would be less vulnerable to interest rate increases because liabilities would decline in sync with the value of fixed-income assets (discount rate for liabilities would be higher). - A prolonged period of low rates would be a more relevant vulnerability, as the duration of liabilities usually exceeds the duration of assets for many life insurers in Japan, causing liabilities to increase more than assets when interest rates decline. ### Bottom-Up (BU) stress test results - Life insurers (BU): - Average SMR: 949 percent before stress; 723 percent in the moderate adverse scenario (scenario 1); 452 percent in the severe adverse scenario (scenario 2). - All life insurers remain above the statutory hurdle rate of 200 percent. - Non-life insurers (BU): - Average SMR: 736 percent before stress; 624 percent in scenario 1; 479 percent in scenario 2. - None of the six non-life insurers in the sample falls below the 200 percent hurdle rate in either scenario. - Recovery dynamics: - Insurers are able to recover from the severe adverse scenario mainly due to higher expected investment returns after the interest rate hike. - Solvency ratios recover as insurers retain earnings; recovery is quicker and more uniform in the non-life than in the life sector. - Net surplus projections for life companies are rather flat and slightly declining in the baseline, improving gradually in the severe adverse scenario. - The spread between investment returns and guaranteed interest rates widens substantially in the severe adverse scenario, boosting life company profits. - Some non-life firms see an immediate profit impact in the first year, mainly due to less reversal of price fluctuation reserves, but subsequent years are slightly more profitable than baseline. ### Anticipated adjustments in asset allocation - Life insurers indicated they would invest in newly issued JGBs with higher coupons following a higher-yield environment; quantification and timing are uncertain. - Some companies plan gradual strategies—directing regular premium income into JGB investments; a few would consider divestments of non-investment grade corporate bonds and equity. - Life companies may consider reducing their duration gap by increasing their asset duration in a higher-yield scenario. ### Sensitivity analysis: biometric and catastrophic shocks - Single-factor shocks tested separately (not aggregated with macro scenarios); covered biometric risks and catastrophic events; only direct claims impact reported. - Biometric shocks for life business (ST participants): - Longevity shock: modeled as a permanent 20 percent decline in mortality rates. - Pandemic event: temporary 35 percent increase in disability and morbidity rates over one year, and temporary 10 percent increase in mortality rates over one year. - Catastrophe scenarios for non-life business (based on historic events and current exposures): - Great Kantō earthquake (1923) modeled for Japan domestic exposure. - Typhoon Mireille (1991) modeled for Japan domestic exposure. - Hurricane Andrew (U.S., 1992) modeled for non-Japanese exposures. - Reinsurance recoveries from each company’s five largest reinsurers were included (on a group basis). - Impacts on available capital: - Great Kantō–like earthquake: largest reduction between 2 and 6 percent of available capital for non-life companies. - Typhoon Mireille repeat: reduction between 2 and 3 percent. - Hurricane Andrew–like repeat: marginal loss of around 1 percent for Japanese non-life insurers exposed to that region. - Longevity shock (permanent decrease in mortality): slightly positive effect on available capital of up to 2 percent for most life insurers. - Pandemic: reductions in available capital ranging between zero and five percent, varying by company due to heterogeneity in medical insurance products. - Observations: - Natural disasters within Japan have an impact on non-life available capital but none of tested events in isolation causes solvency concerns. - Sensitivity to catastrophic risks outside Japan is limited but could grow with overseas expansion of Japanese non-life insurers. - For non-life insurers, it is essential to model subsequent financial market reactions in addition to direct insurance impacts. ### Box: Natural disaster risks in Japan (summary) - Japan is vulnerable to earthquakes and typhoons; the Great East Japan Earthquake (March 11, 2011) caused insured losses of around US$40 billion (in 2011 prices). - After the Great East Japan Earthquake (2011): - Non-life combined ratio temporarily rose to nearly 117 percent. - Return on equity turned negative to -6.3 percent. - SMR dropped from 629 to 477 percent but the sector recovered and was profitable again in 2012 and subsequently. - Financial markets recovered quickly: the Nikkei 225 dropped nearly 21 percent from March 11 to March 15, but recouped more than half of its losses within two weeks. - Japan Earthquake Reinsurance (JER) system: - Centers on Japan Earthquake Reinsurance Ltd., set up in 1966 and owned by Japanese non-life companies. - Insurable: losses to residential property due to earthquakes (including tsunamis) and volcanic activity. - As of April 1, 2017: beyond a loss amount of around ¥224 billion per event, the government pays 99.8 percent of any losses up to a threshold of ¥11.3 trillion. - After the Great East Japan Earthquake, nearly ¥1.3 trillion were paid out, of which around three quarters within 75 days. - Modeling challenges: - Catastrophe modeling (especially typhoons) is becoming more challenging due to climate change, which impacts frequency and severity and poses structural risks of mispricing and under-reserving. - Earthquake models may underestimate tsunami impacts; companies rely on third-party vendor models but consider some models sub-optimal. *Source: IMF staff calculations and company submissions as presented in the referenced IMF chapter.* ### 85. Measuring financial interconnectedness sheds light on the transmission of financial ### Measuring financial interconnectedness sheds light on the transmission of financial stress across financial institutions ### Channels and patterns of interconnectedness - Cross-ownership: - On average, cross-ownership accounts for 18 percent of the institutions equity capital. - Cross-ownership is pervasive among FGs of the three mega banks (Mitsubishi UFJ, Mizuho, and Sumitomo Mitsui), Nomura Holdings, and several major insurance companies, and there is a dense network of cross-ownership among regional banks. - Funding and counterparty linkages: - Banks drawing on large deposit bases provide funding to the rest of the financial system through equity investment or lending and thus can propagate liquidity shocks via withdrawal of funding. - Expansion of investments overseas by many Japanese financial institutions has introduced foreign-currency market access as an additional channel for funding-shock propagation. - Common exposures: - Only mega banks and insurance companies have significant foreign exposures; regional banks largely rely on domestic customers. - This implies foreign-exposure channels are particularly relevant for linkages between city banks and insurance companies, while domestic common exposures (including via cross-shareholdings) pose risks of concentrated domestic shocks. ### Market-data (equity return) spillover findings - Main sources of spillovers (equity-return based network using Diebold and Yilmaz (2014)): - Japan’s nonfinancial firms and city banks are the main source of financial spillovers. - Shocks to Japan’s largest banks and largest nonfinancial firms propagate strongly to other sectors, especially to insurance companies. - Links among internationally exposed and domestically-oriented regional banks are also strong. - The only significant spillovers from abroad emanate from foreign nonfinancial firms. - Revenue-geography patterns summarized in text: - Mega banks, on average, source almost one third of their income from abroad. - Regional banks have almost no exposure to foreign markets. - Securities firms source almost 90 percent of their income from domestic clients. - Except for the three largest city banks and nonlife insurance companies, foreign income accounts for a small fraction of total operating income. ### Firm-level determinants of spillover sensitivity (econometric results) - Firms with stronger financial statements are less sensitive to shocks: - Higher z-Score and higher returns on assets (ROA) → less sensitivity to shocks from other firms (statistically significant). - Size effects: - Larger firms are a more important source of financial spillovers (spillovers from j to i are larger when j's size is larger than i's). - Business-model and funding structure: - Among Japanese banks, those that rely less on wholesale funding or have loans-driven business models (higher share of loans as percentage of assets) are less sensitive to shocks from other banks. - Corporate governance: - Higher shares of institutional ownership (proxy for better corporate governance) → firms are less sensitive to financial spillovers. - Geography: - Greater geographic proximity (shorter distance between headquarters) → larger spillovers. ### Balance-sheet (bilateral-claims) network analysis: connectivity metrics - Centrality and connectivity (average and max values by sector, based on confidential balance-sheet bilateral exposures): - Betweenness (Average / Max): - Mega Banks: 45.80 / 64.30 - City Banks: 18.60 / 68.30 - Regional Banks: 12.20 / 31.60 - Nonlife Insurance: 20.60 / 66.00 - Life Insurance: 6.40 / 18.20 - Securities Firms: 20.60 / 34.10 - Closeness (Average / Max): - Mega Banks: 0.97 / 1.00 - City Banks: 0.82 / 0.95 - Regional Banks: 0.79 / 0.84 - Nonlife Insurance: 0.74 / 0.93 - Life Insurance: 0.71 / 0.84 - Securities Firms: 0.84 / 0.95 - Borrowing centrality (Average / Max): - Mega Banks: 0.77 / 0.89 - City Banks: 0.56 / 0.84 - Regional Banks: 0.69 / 0.89 - Nonlife Insurance: 0.56 / 0.82 - Life Insurance: 0.60 / 0.93 - Securities Firms: 0.76 / 1.00 - Funding centrality (Average / Max): - Mega Banks: 0.99 / 1.00 - City Banks: 0.73 / 0.94 - Regional Banks: 0.69 / 0.86 - Nonlife Insurance: 0.46 / 0.90 - Life Insurance: 0.38 / 0.78 - Securities Firms: 0.67 / 0.90 - Interpretation: - The largest city (mega) banks have the highest levels of connectivity and are by far the most connected providers of funding to the system (highest funding centrality). - Insurance companies generally show the lowest levels of centrality. - Other city banks show greater relevance as providers of funding than as borrowers. ### Balance-sheet contagion model: scenarios and calibration - Model scope and assumptions: - Uses Espinosa-Vega and Solé’s (2011) network approach applied to confidential bilateral claims among six financial sectors: large city banks, other city banks, regional banks, life insurance companies, nonlife insurance companies, and securities firms. - Exposures do not include derivatives. - An institution is assumed to be under stress if it loses its capital buffer, defined as capital amount higher than 2 percent of assets. Total capital buffer equals sum across institutions of (capital − 2 percent of assets). - In all scenarios institutions cannot raise additional capital and all losses are absorbed by capital. - The model does not account for exposures to common clients/shareholders or changes in market sentiment, but can partially accommodate such effects via calibration of parameters (fire-sale discount, LGD). - Scenarios and calibration specifics: - Funding shock: - Institutions cannot rollover a given percentage of their funding, calibrated on the share of short-term liabilities in total liabilities vis-à-vis network firms. - To cover shortfall, institutions sell assets at full price for liquid assets and at fire-sale prices for illiquid assets; asset-price loss calibrated using share of illiquid investment assets in total investment assets. - Credit shock: - Each financial institution fails on its liabilities vis-à-vis other institutions, causing credit loss. - Exposures are net of collateral and other risk mitigation; LGD is assumed to be 100 percent. - Credit and funding shock: - Combination of the two shocks above. - Outcome metrics: - Propensity to propagate a shock: capital buffer losses induced in all other financial institutions (expressed as percent of the system’s total buffer capital in the simulation). - Vulnerability of an institution: total capital buffer lost when other institutions fail or withdraw funding. ### Simulation results: magnitude and distribution of contagion - Funding-shock results: - Withdrawal of funding by large city banks would cost slightly more than 2 percent of the financial system’s total capital. - Securities firms are the most vulnerable to funding shocks (regardless of origin) with average capital losses equivalent to 2 percent of the total capital in the simulation. - Credit-shock results: - Default by a large city bank would absorb 12 percent of the financial system’s total capital (average losses). - Simulated losses from default by a smaller city bank or a securities firm would be around 7 percent and 5 percent of the financial system’s total capital, respectively. - Life insurance companies would suffer most under a credit-default scenario because of their deposits, credit, and equity exposures. - The combination of credit and funding shocks has somewhat higher effects but of similar nature. - High levels of capital held by many Japanese financial institutions help contain credit and funding contagion throughout the financial system. - Sector-default (simultaneous failure of all institutions in a sector) results: - Scenarios of sector-specific simultaneous failures generally have mild implications for other sectors when considering only observed direct balance-sheet exposures. - Failures by city banks cause a fair number of insurance and securities companies to become stressed in the simulation. - Under all other sector-default scenarios, the resulting number of stressed institutions is small. - The measured effects should be considered a lower bound because the model does not fully capture cascading effects from changes in market sentiment or exposures to the nonfinancial sector. ### Key methodological and interpretative caveats - The balance-sheet model: - Measures interconnectedness risk from direct counterparty exposures only. - Does not include exposures through derivatives. - Assumes bank subsidiaries are ring-fenced and does not consider intra-conglomerate capitalization strategies. - Does not capture indirect channels such as common-client exposures, shareholder links, or market-sentiment driven feedback loops, though some effects can be proxied by calibration choices (LGD, fire-sale discounts). - Result interpretation: - Simulated contagion losses and numbers of stressed institutions should be considered lower bounds due to omitted channels like market sentiment and nonfinancial-sector feedbacks. *Source: IMF staff analysis in “Measuring financial interconnectedness sheds light on the transmission of financial stress across financial institutions.”* ### 99. The analysis of spillovers based on market and balance sheet data suggest that the ### cr17285 - 99. The analysis of spillovers based on market and balance sheet data suggest that the ### Spillovers and Transmission Mechanisms - Propagation of financial shocks occurs primarily through the client and investor bases. - Direct exposures play only a modest role in the transmission of funding shocks. - Strong links between financial and nonfinancial firms underpin market-based spillovers. - Changes in market sentiment associated with credit and/or funding shocks may magnify transmission, although market participants believe the strong overall health of the Japanese financial system makes swift changes in market sentiment unlikely. - The safe-haven status of the Japanese economy and financial system implies net inflows from abroad should increase during stress periods; ensuing currency appreciation or other asset price changes could, over the medium term, decrease profitability of nonfinancial firms and weaken balance sheets of financial intermediaries. ### Quantitative Indicators and Observed Model Outputs (figures and reported statistics preserved as in source) - Average maturity of public debt: about 7.7 years. - Recovery rate used for one-year default probability for Japan sovereign: 40 percent. - Statutory tax rate used in modeling: 30 percent. - Behavioral caps on fee income response: growth should not decline more than 10 percent (y-o-y) in any scenario (assuming that bank would increase fees in response to decline in interest income). - Reported R^2 statistics in sample regressions: 96%, 96%, 96%, 70% (examples shown). - Observations in regressions: 190, 160, 170, 170 (examples shown). - Reported model R^2 for other regressions: 99%, 97%, 93%, 81% (examples shown). - Posterior probabilities and inclusion statistics shown in Bayesian Model Averaging outputs (examples in tables). ### Risk Assessment Matrix — Key Risks, Likelihoods, and Impacts - Retreat from cross-border integration - Likelihood (Over next 1–5 years): High - Impact: Medium/High - Transmission channels: reduced trade, capital and labor flows; yen appreciation due to safe-haven effects; drag on equity prices and exports; increased credit losses on overseas investments; higher credit risk for domestic companies. - Sharp growth slowdown and financial risks in China and other emerging markets over the medium term - Likelihood: Medium - Impact: High - Transmission channels: stall in export recovery, safe-haven appreciation, stock market corrections, reductions in export revenue hitting companies and increasing banks’ credit losses. - Abrupt normalization of U.S. monetary policy - Likelihood: Medium/High - Impact: High - Transmission channels: higher Fed policy rates, asset price adjustment, decompression of credit spreads, FX liquidity strains, higher funding costs for Japanese banks relying on FX and cross-currency swaps. - Bond market stress from a reassessment of sovereign risk in Japan - Likelihood: Medium - Impact: High - Transmission channels: sharp increase in long-term JGB yield, tightening in domestic financial conditions, possible distress in the financial sector with knock-on effects on debt (risk assessed via solvency and market risk analysis). - Tightening in domestic financial conditions - Likelihood: Medium - Impact: High - Transmission channels: higher domestic risk premia, higher money market interest rates, lower equity prices, yen depreciation, potential house price declines. ### IMF Credit Risk Model — Structure and Scenario Treatments - Scenarios covered: Baseline, Moderate Adverse, Severe Adverse. - PDs: Satellite models using TTC PDs estimated at individual bank and exposure-class level; logit transformation applied to PDs to ensure predictions between 0 and 1. - PD model specification (logit-transformed PD Yit): - Yit modeled as linear function of exogenous macrofinancial variables Xt, lagged dependent variable Yit-1, with parameters α, β, δ and iid error εit (equations described in source). - Use of Bayesian Model Averaging (BMA) for panel regressions: - Purpose: test for significant explanatory variables and their lags; simulate multiple combinations of regressors; help select lag structure and avoid specification bias. - Final PD under stress computed using inverse logit: - PDstress,type,t = 100 * [exp(α + β'Xt,stress + ... ) / (1 + exp(α + β'Xt,stress + ... ))] (formula structure preserved as in source). - PD multipliers: - Median PDs from banks in the sample used as benchmark to calculate PDs multiplier: PDstress,t / PD0,t (formula provided). - LGDs: - LGD models for corporate and mortgage exposure classes obtained using similar methodology; LGDs were not converted into logit form. ### Modeling Inputs and Behavioral Assumptions - Growth rates of interest-bearing assets, non-interest bearing assets, open positions: in line with macro forecasts (scenario-specific). - AFS/HFT own sovereign spread: No changes in some scenarios; in one scenario a one-year default probability for Japan based on Japan CDS spread data and 40 percent recovery rate was used for provisioning only (no application to RWAs). - AFS/HFT foreign sovereign spread: PDs/LGDs approach applied. - Funding risk shock: Projected separately for JPY and USD liabilities; regime switching and panel GMM models used. - Interest income growth: Repricing gap model for shock to risk-free interest rates; effective interest rate model for loans and other interest earning assets; shock to Euribor minus shock to risk-free interest rate on loans. - Net fee and commission income growth: Satellite model with fees and commissions income as a share of assets; key explanatory variables include one period lagged dependent variable and real GDP growth rate. - Trading income: Satellite model with trading income ratio to total assets as dependent variable; explanatory variables include equity indices (Topix), exchange rate, lagged GDP, lagged dependent variable. Results compared with valuation changes of HFT assets to avoid double counting. - Other non-interest income growth: Assumed as 0 growth in scenarios. - Non-interest expenses (OPEX): Modeled using satellite model with OPEX ratio to total assets as dependent variable and lagged loan portfolio growth as explanatory. - Tax rate consistently assumed: Actual statutory tax rate (30 percent). ### PD and LGD Model Findings (high-level) - Corporate PDs: Mostly sensitive to lagged real GDP growth rate, equity indices, and interest rates. Elasticities are small, reflected in relatively small multipliers. - Mortgage PDs: Sensitive to GDP, unemployment, and interest rates. - PDs for equity exposures: Depend on equity prices, interest rates, and exchange rates. - Other retail loans: Depend on interest rates, GDP, and unemployment. - Estimation approaches used: OLS, GMM, Robust Least Squares, SUR; BMA used to verify macro significance and select specifications. _Italic: Source — cr17285 (excerpt provided)._ ### Appendix Table 3. Japan: Examples of Equations Obtained Using BMA Estimation ### cr17285 - Appendix Table 3. Japan: Examples of Equations Obtained Using BMA Estimation ### BMA: Corporate Loans — Example equations and estimation metrics - Models reported: Model1, Model2, Model3, Model4, Model5 (two separate panels shown). - Selected estimated intercepts (C): - Panel 1: C = -9.03, -9.44, -9.43, -8.63, -9.51 (standard errors shown in parentheses: (0.27)(0.20)(0.21)(0.22)(0.21)). - Panel 2: C = -9.45, -9.69, -9.44, -9.60, -9.45 (standard errors shown in parentheses: (0.03)(0.23)(0.03)(0.21)(0.03)). - Selected coefficients (with lags and standard errors where reported): - RGDP_G: -11.29 (panel 1; (4.31)). - RGDP_G(-1): 9.27 (panel 1; (4.20)). - RGDP_G(-4): -2.61 (panel 2; (3.40)). - TOPIX and TOPIX(-1) coefficients reported as 0.00 or -0.00 in panel 1 (standard errors (0.00)). - TIBOR_3M(-2): 0.86 (panel 1; (0.17)). - LT_PRIME(-1): -0.73 (panel 1; (0.09)). - LT_PRIME(-2): 0.49, -0.63, -0.70 (panel 1; standard errors (0.09)(0.08)(0.09)). - UNEMP and UNEMP(-1): UNEMP coefficient in Model4 panel 2 = 0.04 (standard error (0.05)); UNEMP(-1) in Model2 panel 2 = 0.06 (standard error (0.05)). - Model fit and selection statistics (examples): - No of Variables: 3, 4, 3, 3, 4 (panel 1); 1, 2, 2, 2, 2 (panel 2). - R-Squared: 0.364, 0.373, 0.352, 0.352, 0.370 (panel 1); 0.000, 0.008, 0.004, 0.004, 0.002 (panel 2). - Bayesian Info Criterion: -66.564, -63.882, -63.443, -63.380, -63.233 (panel 1); 0.000, 3.818, 4.412, 4.462, 4.689 (panel 2). - Posterior Probability: 0.283, 0.074, 0.060, 0.058, 0.054 (panel 1); 0.507, 0.075, 0.056, 0.054, 0.049 (panel 2). ### BMA: Equity Portfolio — Example equations and estimation metrics - Intercepts (C) and standard errors for Models 1–5: - C = 3.32 (4.01), -5.31 (1.62), -14.36 (0.38), -6.66 (1.54), -13.88 (0.37). - Selected coefficients: - NEER: 0.09, 0.03, 0.04, 0.03, 0.04 (standard errors (0.02)(0.005)(0.004)(0.005)(0.005)). - DUSDJPY: -0.02 (Model2; standard error (0.01)). - REER: -0.08 (Model1; standard error (0.03)). - JGBY_10Y: 0.52 (Model4; (0.11)); JGBY_30Y: 0.53 (Model3; (0.11)). - Model fit and selection statistics (examples): - No of Variables: 4, 4, 3, 3, 3. - R-Squared: 0.438, 0.433, 0.417, 0.415, 0.412. - Bayesian Info Criterion: -88.060, -86.708, -86.672, -86.003, -85.260. - Posterior Probability: 0.187, 0.095, 0.094, 0.067, 0.046. --- ### Appendix III — Profit and Funding Cost Projection: Methods and key assumptions - Projection of bank income: - Bank’s income growth projected using the loan interest income ratio to total loans (period 2002–16) and the difference between 10-year JGB yield and three-month Japan Government Treasury notes yield. - Projected interest income from overseas converted using changes in USD/JPY rate. - Assumed income from extraordinary items did not recur during the 2017–21 period in the baseline and the adverse scenarios. - Non-performing loans assumed not to provide any accrued income. - Funding costs model structure: - Liabilities and assets split into funding costs in U.S. dollar and in Japanese yen. - Total funding costs = interest expenses / interest rate sensitive liabilities. - Estimation method: panel GMM model with fixed effects (annual data over 2002–16) estimating Japanese banks’ average annual interest expense to total funding cost ratio. - Explanatory variables included: - lagged dependent variable (funding cost t-1), - Libor U.S. dollar three-month rate (USD 3m LIBOR), - average rate of ordinary deposits (Rate_t; proxy for funding costs in Japanese yen), - each bank’s capital gap (CapGap_i,t) = difference between actual and minimum required CAR. - Estimated equation (text form): - Funding cost_{i,t} = C_i + α * funding cost_{i,t−1} + β * USD 3m Libor_t + γ * Rate_t − CapGap_{i,t}. - Ordinary deposit rate projection: - Projected using a Markov regime-switching model with Tibor 6-month rate as switching regressor, similar to the model used for USD LIBOR 3-month. ### U.S. dollar funding cost and liquidity regime-switching model - Rationale: - Banks’ U.S. dollar funding relies heavily on FX swaps; stress in the USD/JPY swap market affects U.S. dollar funding cost. - Liquidity risk premium changes quickly (regime-switching behavior), so a Markov regime-switching model is used. - Change-in-U.S-dollar-funding-cost model (equation (1) in source): - ∆FC_t^USD = β0^k + β1 * ∆LIBOR_t + β2^k * BAS_t + ε_t^k - where k denotes liquidity regime; ∆LIBOR_t = change in 3-month USD LIBOR; BAS_t = bid-ask spread of 3-month USD/JPY swap contract (measure of market liquidity). - The constant β0^k is also allowed to vary across regimes. - Estimation findings and regime interpretation: - Two regimes identified: a “stress regime” (high liquidity risk premium) and a “tranquil regime” (low liquidity risk premium). - Bid-ask spread has a significant impact on U.S. dollar funding cost only in the stress regime; coefficient significant in one regime, insignificant in the other. - The estimated coefficient in the stress regime is used as the liquidity risk premium for projections. ### Scenario assumptions for liquidity risk premium and funding cost projections - Baseline and moderate adverse scenarios: - Assumed no deterioration in market liquidity in the USD/JPY swap market; main driver of projected U.S. dollar funding cost is 3-month USD LIBOR. - Liquidity risk premium assumed zero (estimated coefficient insignificant). - Severe adverse scenario: - Assumed liquidity of the USD/JPY swap market (bid-ask spread) deteriorates by two standard deviations and fully recovers by end of the five-year horizon. - It is assumed the liquidity risk premium switches from tranquil to stress regime due to more rapid tightening in U.S. monetary policy; estimated stress-regime coefficient applied to project USD funding cost. - U.S. dollar funding cost calculation: - U.S. dollar funding cost is calculated from the 3-month USD/JPY swap cost. ### Projection outcomes and impacts on banks - Funding cost projections by scenario (qualitative summary from estimation results): - Baseline: funding costs would increase only moderately due to small changes in short term interest rates and higher U.S. dollar Libor rate. - Moderate adverse: funding costs will remain flat. - Severe adverse: biggest increase of funding costs projected—driven by faster-than-expected monetary policy normalization (short-term interest rates and term premiums) and decline in CAR from market and credit risk losses. - Pass-through and capitalization effects: - Higher funding costs affect banks’ profit and loss and may reduce capitalization if pass-through of funding cost increases to lending rates < 100 percent. - Factors affecting pass-through: - negotiating power of clients and competition (retail pass-through might be close to 100 percent; wholesale less), - dominant type of interest rate setting in loan contracts (majority of loans in Japan tied to short-term rates such as 3- or 6-month Tibor, implying pass-through of market rate increases close to 100 percent). - Pass-through of increases in individual banks’ funding risk premiums may be below 100 percent due to dominance of corporate loans. - Dynamic updating of funding costs and capital adequacy: - Individual banks’ funding risk premiums partially captured via dynamic updates of funding costs (year-by-year) to reflect previous year’s capitalization. - Result: the decline in capital adequacy ratios (CAR) is 0.7 p.p. higher than without inclusion of dynamic funding-cost adjustment in the Adverse Severe scenario. - Comparative severity: - Estimated severity of funding costs in the adverse scenario is comparable to levels observed during 2008–09 crisis. --- ### Appendix IV — Estimation of Equity Price Haircuts - Liquidity measure: - Amihud (2002) measure used: daily absolute return divided by daily transaction volume, computed for each stock in Nikkei 225; the average across Nikkei 225 used as indicator of historical price impact of equity sales. - Regime-switching for Amihud: - A simple Markov regime-switching model estimated for the average Amihud measure: - Amihud_t = β0^k + ε_t^k - Two regimes identified: low-liquidity regime and high-liquidity regime. - The estimated constant in the low-liquidity regime is used as the average impact of equity sales on equity prices. - Haircut calculation: - Average price impact used to calculate haircut ratios dynamically: change in equity price depends on total equities sold by all banks simultaneously; haircut ratio derived from this price change. --- ### Appendix V — Liquidity Stress Testing: Cash Flow-Based Run-Off Rates (2-Week Scenario) - Table excerpt: Run-off rates (2-Week Scenario) reported in percent across time buckets (O/N, 2day～1W, 1～2W, 2W～1M, 1M～2M, 2M～3M, 3M～6M, 6M～9M, 9M～1Y). - Representative run-off rates from the table (selected items preserved exactly as in source): - Retail funding due (including small business customers but excluding SMEs): - thereof: sight deposits: 0.9 1.4 0.7 0.0 0.0 0.0 0.0 0.0 0.0 - thereof: others: 3.0 3.0 3.0 0.0 0.0 0.0 0.0 0.0 0.0 - thereof: less stable deposits: - sight deposits: 2.1 3.3 1.6 0.0 0.0 0.0 0.0 0.0 0.0 - others: 7.0 7.0 7.0 0.0 0.0 0.0 0.0 0.0 0.0 - thereof: term deposits: 5.0 5.0 5.0 0.0 0.0 0.0 0.0 0.0 0.0 - Unsecured wholesale funding due: - thereof: total operational deposits: - sight deposits: 7.6 11.6 5.8 0.0 0.0 0.0 0.0 0.0 0.0 - other deposits: 25.0 25.0 25.0 0.0 0.0 0.0 0.0 0.0 0.0 - thereof: certificate of deposits: - unconditionally callable: 7.6 11.6 5.8 0.0 0.0 0.0 0.0 0.0 0.0 - other certificate of deposits: 25.0 25.0 25.0 0.0 0.0 0.0 0.0 0.0 0.0 - thereof: debt securities (excluding certificate of deposits): - unconditionally callable: 15.2 23.2 11.6 0.0 0.0 0.0 0.0 0.0 0.0 - other debt securities (excluding certificate of deposits): 50.0 50.0 50.0 0.0 0.0 0.0 0.0 0.0 0.0 - deposits from non-financial corporates, sovereigns, central banks, MDBs and PSEs: - sight deposits: 9.1 13.9 7.0 0.0 0.0 0.0 0.0 0.0 0.0 - other deposits: 30.0 30.0 30.0 0.0 0.0 0.0 0.0 0.0 0.0 - funding from individual group entities and central branch (consolidated foreign/domestic non-financial legal entities and financial legal entities; head office and branch accounts for foreign banks) — representative entries: - unconditionally callable: 9.1 13.9 7.0 0.0 0.0 0.0 0.0 0.0 0.0 (for consolidated foreign/domestic non-financial legal entities) - others: 30.0 30.0 30.0 0.0 0.0 0.0 0.0 0.0 0.0 - consolidated foreign financial legal entities (unconditionally callable): 18.2 27.9 13.9 0.0 0.0 0.0 0.0 0.0 0.0 - consolidated foreign financial legal entities (others): 60.0 60.0 60.0 0.0 0.0 0.0 0.0 0.0 0.0 - similar 18.2 27.9 13.9 and 60.0 60.0 60.0 patterns repeated for consolidated domestic financial legal entities and head office and branch accounts (only for foreign banks). *Source: IMF staff calculations.* ### Appendix Table 1. Japan: Cash Flow-Based Liquidity Analysis: Run-Off Rates (2-Week Scenario) (concluded) ### Appendix Table 1. Japan: Cash Flow-Based Liquidity Analysis: Run-Off Rates (2-Week Scenario) (concluded) ### Cash outflows: selected run-off rate entries (2-Week Scenario) - Thereof: non-consolidated financial institutions — unconditionally callable: 18.2 (O/N), 27.9 (2day～1W), 13.9 (1～2W); all subsequent buckets: 0.0. - Thereof: others (related grouping) — others: 60.0 (O/N), 60.0 (2day～1W), 60.0 (1～2W); subsequent buckets: 0.0. - Thereof: non-consolidated financial institutions — unconditionally callable (alternate grouping): 30.4 (O/N), 46.4 (2day～1W), 23.2 (1～2W); subsequent buckets: 0.0. - Thereof: others (alternate grouping) — others: 100.0 (O/N), 100.0 (2day～1W), 100.0 (1～2W); subsequent buckets: 0.0. - Secured wholesale funding due — conducted with domestic central bank: all buckets: 0.0. - Secured wholesale funding due — not conducted with domestic central bank: - Secured by Level 1 assets — all buckets: 0.0. - Secured by Level 2A assets — unconditionally callable: 4.6 (O/N), 7.0 (2day～1W), 3.5 (1～2W); others: 15.0 (O/N), 15.0 (2day～1W), 15.0 (1～2W); later buckets: 0.0. - Secured by Level 2B assets — unconditionally callable: 15.2 (O/N), 23.2 (2day～1W), 11.6 (1～2W); others: 50.0 (O/N), 50.0 (2day～1W), 50.0 (1～2W); later buckets: 0.0. - Secured by other assets — unconditionally callable: 30.4 (O/N), 46.4 (2day～1W), 23.2 (1～2W); others: 100.0 (O/N), 100.0 (2day～1W), 100.0 (1～2W); later buckets: 0.0. - Outflows from derivatives — unconditionally callable: 12.1 (O/N), 18.6 (2day～1W), 9.3 (1～2W); others: 40.0 (O/N), 40.0 (2day～1W), 40.0 (1～2W); later buckets: 0.0. - Undrawn volume of committed credit/liquidity lines — unconditionally callable and others: all listed buckets: 0.0. - Others — unconditionally callable and others: all listed buckets: 0.0. *Source: IMF staff calculations.* --- ### Appendix Table 2. Japan: Cash Flow-Based Liquidity Analysis: Roll-Off Rates (2-Week Scenario) (For U.S. dollars, euros) ### Cash inflows: secured lending, contractual inflows, derivatives (roll-off = percent) - Secured lending (reverse repos) — secured by Level 1 assets: - Thereof: unconditionally callable: 100.00 (O/N); 0.00 for 2day～1W and all later buckets. - Thereof: others: 100.00 across O/N, 2day～1W, 1～2W, 2W～1M, 1M～2M, 2M～3M, 3M～6M, 6M～9M, 9M～1Y. - Secured lending — secured by Level 2A assets: - Thereof: unconditionally callable: 100.00 (O/N); 0.00 thereafter. - Thereof: others: 100.00 across all buckets listed. - Secured lending — secured by Level 2B assets: - Thereof: unconditionally callable: 100.00 (O/N); 0.00 thereafter. - Thereof: others: 100.00 across all buckets listed. - Secured lending — secured by other assets: - Thereof: unconditionally callable: 100.00 (O/N); 0.00 thereafter. - Thereof: others: 100.00 across all buckets listed. - Contractual inflows from fully performing loans — multiple borrower groupings (non-consolidated financial entities; consolidated foreign/domestic non-financial/financial legal entities; head office and branch accounts; loans to central banks; loans to other entities): - Thereof: unconditionally callable: 100.00 (O/N); 0.00 for 2day～1W and all later buckets. - Thereof: others: 100.00 across all buckets listed. - Contractual inflows from securities — unconditionally callable: 100.0 (O/N); 0.0 thereafter; other contractual inflows from securities: 100.0 across all buckets. - Inflows from derivatives — unconditionally callable: 100.0 (O/N); 0.0 thereafter; other inflows from derivatives: 100.0 across all buckets. - Others — unconditionally callable: 100.0 (O/N); 0.0 thereafter; others: 100.0 across all buckets. *Source: IMF staff calculations.* --- ### Appendix Table 3. Japan: Cash Flow-Based Liquidity Analysis: Haircuts and Sales of Assets (2-Week Scenario) (For U.S. dollars, euros) ### HQLA and asset haircut ratios (percent) - HQLA — Cash and banknotes, Central bank reserves, securities issued or guaranteed by sovereigns/central banks/PSEs: haircut ratios 0.01 (for each listed jurisdiction: Japan, USA, EU, UK, others) across O/N through 9M～1Y buckets (entries appear as 0.0100... repeated). - CP and bonds not included anywhere above — Japan/USA/EU/UK/others: - O/N: 15.0 - 2day～1W and onward: 0.0 or presented as 100.0 in some table formatting lines; (table shows "15.0" and then long sequence of "100.0" for later buckets in other rows—preserve exact table values as printed). - Non-financial common equity shares — Japan/USA/EU/UK/others: - O/N: 50.0 - 2day～1W: 50.0 - 1～2W and later: 100.0 across 1～2W through 9M～1Y. - CP and bonds not included anywhere above (alternate rows): - Japan/USA/EU/UK/others: 25.0 (O/N), 50.0 (2day～1W), 100.0 (1～2W and later). - Equity shares not included anywhere above — Japan/USA/EU/UK/others: - 50.0 (O/N), 50.0 (2day～1W), 100.0 (1～2W and later). - Others — Japan/USA/EU/UK/others: - 50.0 (O/N), 50.0 (2day～1W), 100.0 (1～2W and later). *Source: IMF staff calculations.* --- ### Appendix Table 4. Japan: Cash Flow-Based Liquidity Analysis: Haircut Ratios (2-Week Scenario) (For Japanese yen) ### BoJ haircut ratios (through discount window and for USD-funds supplying operations) - HQLA — Cash and banknotes / Central bank reserves (by jurisdiction: Japan, USA, EU, UK, others): - BoJ haircut ratios through discount window: 0.0 (O/N), 1.4 (2day～1W), 0.0 (1～2W and all later buckets listed). - BoJ haircut ratios for USD-funds supplying operations: 0.0 (O/N), 1.4 (2day～1W), 0.0 thereafter. - Securities issued or guaranteed by sovereigns/central banks/PSEs — Japan/USA/EU/UK/others: - BoJ haircut ratios through discount window: 1.0 across O/N, 2day～1W, 1～2W, 2W～1M, 1M～2M; 0.0 for 2M～3M and later. - BoJ haircut ratios for USD-funds supplying operations: identical pattern: 1.0 across O/N through 1M～2M; 0.0 thereafter. - CP and bonds not included anywhere above — Japan/USA/EU/UK/others: - BoJ haircut ratios through discount window: 4.0 (O/N), 4.0 (2day～1W), 15.0 (1～2W), 15.0 (2W～1M), 15.0 (1M～2M); 0.0 for 2M～3M and later. - BoJ haircut ratios for USD-funds supplying operations: same sequence: 4.0, 4.0, 15.0, 15.0, 15.0, then 0.0. - Non-financial common equity shares (market haircut ratios / estimated price impact) — Japan/USA/EU/UK/others: - BoJ haircut ratios through discount window: 20.7 (O/N), 20.7 (2day～1W), 20.7 (1～2W); 0.0 for 2W～1M and later. - BoJ haircut ratios for USD-funds supplying operations: 20.7, 20.7, 20.7; 0.0 thereafter. - CP and bonds not included anywhere above (alternate section): - CP and bonds — Japan/USA/EU/UK/others: 25.0 (O/N), 25.0 (2day～1W), 25.0 (1～2W); 0.0 thereafter. - Equity shares not included anywhere above (alternate rows): - Japan/USA/EU/UK/others: 20.7, 20.7, 20.7 for O/N, 2day～1W, 1～2W; 0.0 thereafter. - Others — Japan/USA/EU/UK/others: - 50.0 (O/N), 50.0 (2day～1W), 50.0 (1～2W); 0.0 thereafter. *Source: IMF staff calculations.* *1/ The market haircut ratios on domestic equities are based on the estimated price impact of equity sales (see Appendix IV for details).* --- ### Appendix Table 5. Japan: Cash Flow-Based Liquidity Analysis: Run-Off Rates (3-Month Scenario) ### Retail funding due (selected run-off rates, in percent) - Stable deposits — sight deposits: 0.4 (O/N), 0.7 (2day～1W), 0.9 (1～2W), 1.0 (2W～1M), 1.1 (1M～2M), 0.8 (2M～3M); later buckets: 0.0. - Stable deposits — others: 5.0 across O/N, 2day～1W, 1～2W, 2W～1M, 1M～2M, 2M～3M; later buckets: 0.0. - Less stable deposits — sight deposits: 0.9 (O/N), 1.5 (2day～1W), 1.9 (1～2W), 2.1 (2W～1M), 2.1 (1M～2M), 1.6 (2M～3M); later buckets: 0.0. - Less stable deposits — others: 10.0 across O/N through 2M～3M; later buckets: 0.0. - Term deposits: 10.0 across O/N through 2M～3M; later buckets: 0.0. ### Unsecured wholesale funding due (selected run-off rates) - Total operational deposits — sight deposits: 2.2 (O/N), 3.7 (2day～1W), 4.7 (1～2W), 5.2 (2W～1M), 5.3 (1M～2M), 3.9 (2M～3M); later buckets: 0.0. - Total operational deposits — other deposits: 25.0 across O/N through 2M～3M; later buckets: 0.0. - Certificate of deposits — unconditionally callable: 2.2 (O/N), 3.7 (2day～1W), 4.7 (1～2W), 5.2 (2W～1M), 5.3 (1M～2M), 3.9 (2M～3M); other CDs: 25.0 across O/N through 2M～3M. - Debt securities (excluding CDs) — unconditionally callable: 6.1 (O/N), 10.4 (2day～1W), 13.2 (1～2W), 14.5 (2W～1M), 14.7 (1M～2M), 11.1 (2M～3M); other debt securities: 70.0 across O/N through 2M～3M. - Deposits from non-financial corporates, sovereigns, central banks, MDBs and PSEs — sight deposits: 3.5 (O/N), 6.0 (2day～1W), 7.5 (1～2W), 8.3 (2W～1M), 8.4 (1M～2M), 6.3 (2M～3M); other deposits: 40.0 across O/N through 2M～3M. ### Funding from individual group entities and central branch (selected) - Consolidated foreign non-financial legal entities — unconditionally callable: 3.5 (O/N), 6.0 (2day～1W), 7.5 (1～2W), 8.3 (2W～1M), 8.4 (1M～2M), 6.3 (2M～3M); others: 40.0 across O/N through 2M～3M. - Consolidated foreign financial legal entities — unconditionally callable: 8.7 (O/N), 14.9 (2day～1W), 18.8 (1～2W), 20.8 (2W～1M), 21.1 (1M～2M), 15.8 (2M～3M); others: 100.0 across O/N through 2M～3M. - Consolidated domestic financial legal entities and head office and branch accounts (only for foreign banks) show the same figures as consolidated foreign financial legal entities. ### Secured wholesale funding due (3-Month Scenario) - Conducted with domestic central bank: all buckets: 0.0. - Not conducted with domestic central bank: - Secured by Level 1 assets: all buckets: 0.0. - Secured by Level 2A assets — unconditionally callable: 1.3 (O/N), 2.2 (2day～1W), 2.8 (1～2W); others: 15.0 (O/N), 15.0 (2day～1W), 15.0 (1～2W); later buckets: 0.0. - Secured by Level 2B assets — unconditionally callable: 4.3 (O/N), 7.4 (2day～1W), 9.4 (1～2W); others: 50.0 (O/N), 50.0 (2day～1W), 50.0 (1～2W); later buckets: 0.0. - Secured by other assets — unconditionally callable: 8.7 (O/N), 14.9 (2day～1W), 18.8 (1～2W); others: 100.0 (O/N), 100.0 (2day～1W), 100.0 (1～2W); later buckets: 0.0. ### Outflows from derivatives (3-Month Scenario) - Unconditionally callable: 5.2 (O/N), 8.9 (2day～1W), 11.3 (1～2W), 12.5 (2W～1M), 12.6 (1M～2M), 9.5 (2M～3M); others: 60.0 across O/N through 2M～3M. ### Undrawn lines and others - Undrawn volume of committed credit/liquidity lines and Others — unconditionally callable and others: 0.0 across listed buckets. *Source: IMF staff calculations.* *1/ The run-off rates for “undrawn volume of committed credit/liquidity lines” and “others” are calculated as the size-weighted average of the Liquidity Coverage Ratio-run off rates for each subcategory of these items from the more granular data that JFSA provided.* ### Appendix Table 6. Japan: Cash Flow-Based Liquidity Analysis: Roll-Off Rates (3-Month Scenario) ### Appendix Table 6. Japan: Cash Flow-Based Liquidity Analysis: Roll-Off Rates (3-Month Scenario) ### Cash-Inflows — Secured lending (reverse repos) - Secured by Level 1 assets - thereof: unconditionally callable: 100.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 - thereof: others: 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 - Secured by Level 2A assets - thereof: unconditionally callable: 100.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 - thereof: others: 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 - Secured by Level 2B assets - thereof: unconditionally callable: 100.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 - thereof: others: 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 - Secured by other assets - thereof: unconditionally callable: 100.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 - thereof: others: 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 ### Cash-Inflows — Contractual inflows from fully performing loans - Loans to non-consolidated financial entities - thereof: unconditionally callable: 100.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 - thereof: others: 100.0 100.0 90.0 90.0 90.0 90.0 100.0 100.0 100.0 - Unsecured loans to individual group entities and central branch - Consolidated foreign non-financial legal entities - unconditionally callable: 100.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 - others: 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 - Consolidated domestic non-financial legal entities - unconditionally callable: 100.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 - others: 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 - Consolidated foreign financial legal entities - unconditionally callable: 100.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 - others: 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 - Consolidated domestic financial legal entities - unconditionally callable: 100.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 - others: 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 - Head office and branch accounts (only for foreign banks) - unconditionally callable: 100.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 - others: 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 *Source: IMF staff calculations.* ### Appendix Table 7. Japan: Cash Flow-Based Liquidity Analysis: Haircuts and Sales of Assets (3-Month Scenario) ### Counterbalancing Capacity — HQLA and related haircut ratios (U.S. dollars, euros; in percent) - HQLA — Cash and banknotes (Japan, USA, EU, UK, others) - 0.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 0.0 - (Each jurisdiction line shows identical sequence: 0.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 0.0) - Central bank reserves (Japan, USA, EU, UK, others) - 0.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 0.0 - Securities issued or guaranteed by sovereigns, central banks and PSEs (Japan, USA, EU, UK, others) - 0.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 0.0 - CP and bonds not included anywhere above (Japan, USA, EU, UK, others) - 15.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 0.0 - Non-financial common equity shares (Japan, USA, EU, UK, others) - 50.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 0.0 ### Counterbalancing Capacity — CP and bonds not included anywhere above (detailed) - CP and bonds not included anywhere above (Japan, USA, EU, UK, others) - 25.0 50.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 0.0 ### Equity shares and Others (haircut progression) - Equity shares not included anywhere above (Japan, USA, EU, UK, others) - 50.0 50.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 - Others (Japan, USA, EU, UK, others) - 50.0 50.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 *Source: IMF staff calculations.* ### Appendix Table 8. Japan: Cash Flow-Based Liquidity Analysis: Haircut Ratios (3-Month Scenario) (For Japanese yen; in percent) ### BoJ haircut ratios through discount window and for USD-funds supplying operations — HQLA and reserves - Cash and banknotes; Central bank reserves (Japan, USA, EU, UK, others) - BoJ haircut ratios through discount window: 0.0 1.4 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 - BoJ haircut ratios for USD-funds supplying operations: 0.0 1.4 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 ### Securities issued or guaranteed by sovereigns (Japan, USA, EU, UK, others) - BoJ haircut ratios through discount window / for USD-funds supplying operations: - 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 0.0 0.0 0.0 ### CP and bonds not included above (Japan, USA, EU, UK, others) - BoJ haircut ratios through discount window / for USD-funds supplying operations: - 4.0 4.0 15.0 15.0 15.0 15.0 15.0 15.0 0.0 0.0 0.0 ### Non-financial common equity shares 1/ (Japan, USA, EU, UK, others) - BoJ haircut ratios through discount window / for USD-funds supplying operations: - 41.9 51.6 51.6 51.6 51.6 51.6 51.6 0.0 0.0 0.0 ### CP and bonds not included anywhere above — alternate listing - CP and bonds not included anywhere above (Japan, USA, EU, UK, others) - 25.0 25.0 25.0 25.0 25.0 25.0 25.0 0.0 0.0 0.0 ### Equity shares not included anywhere above 1/ (Japan, USA, EU, UK, others) - 41.9 51.6 51.6 51.6 51.6 51.6 51.6 0.0 0.0 0.0 ### Others (Japan, USA, EU, UK, others) - 50.0 50.0 50.0 50.0 50.0 50.0 50.0 0.0 0.0 0.0 Note: 1/ The market haircut ratios on domestic equities are calculated based on the estimated price impact of sales of equities on equity prices (see Appendix IV for details). *Source: IMF staff calculations.* ### Appendix Table 9. Japan: Cash Flow-Based Liquidity Analysis: Run-Off Rates (1-Year Scenario) ### Cash-Outflows — Retail funding due (including small business customers but excluding SMEs) - Stable deposits - Sight deposits: 0.3 0.5 0.6 0.6 0.6 0.6 0.6 0.6 0.5 - Others: 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 - Less stable deposits - Sight deposits: 0.6 1.0 1.1 1.2 1.3 1.3 1.3 1.3 1.0 - Others: 10.0 10.0 10.0 10.0 10.0 10.0 10.0 10.0 10.0 - Term deposits: 10.0 10.0 10.0 10.0 10.0 10.0 10.0 10.0 10.0 ### Cash-Outflows — Unsecured wholesale funding due - Total operational deposits - Sight deposits: 1.6 2.4 2.8 3.0 3.1 3.2 3.2 3.2 2.4 - Other deposits: 25.0 25.0 25.0 25.0 25.0 25.0 25.0 25.0 25.0 - Certificate of deposits - Unconditionally callable: 1.6 2.4 2.8 3.0 3.1 3.2 3.2 3.2 2.4 - Other certificate of deposits: 25.0 25.0 25.0 25.0 25.0 25.0 25.0 25.0 25.0 - Debt securities (excluding certificate of deposits) - Unconditionally callable: 4.5 6.8 7.9 8.5 8.8 8.9 8.9 8.9 6.7 - Other debt securities: 70.0 70.0 70.0 70.0 70.0 70.0 70.0 70.0 70.0 ### Cash-Outflows — Deposits from non-financial corporates, sovereigns, central banks, MDBs and PSEs - Sight deposits: 2.6 3.9 4.5 4.9 5.0 5.1 5.1 5.1 3.8 - Other deposits: 40.0 40.0 40.0 40.0 40.0 40.0 40.0 40.0 40.0 ### Cash-Outflows — Funding from individual group entities and central branch - Consolidated foreign non-financial legal entities - Unconditionally callable: 2.6 3.9 4.5 4.9 5.0 5.1 5.1 5.1 3.8 - Others: 40.0 40.0 40.0 40.0 40.0 40.0 40.0 40.0 40.0 - Consolidated domestic non-financial legal entities - Unconditionally callable: 2.6 3.9 4.5 4.9 5.0 5.1 5.1 5.1 3.8 - Others: 40.0 40.0 40.0 40.0 40.0 40.0 40.0 40.0 40.0 - Consolidated foreign financial legal entities - Unconditionally callable: 6.5 9.7 11.3 12.1 12.5 12.7 12.8 12.8 9.6 - Others: 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 - Consolidated domestic financial legal entities - Unconditionally callable: 6.5 9.7 11.3 12.1 12.5 12.7 12.8 12.8 9.6 - Others: 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 - Head office and branch accounts (only for foreign banks) - Unconditionally callable: 6.5 9.7 11.3 12.1 12.5 12.7 12.8 12.8 9.6 - Others: 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 ### Cash-Outflows — Non-consolidated financial institutions / others (detailed) - Thereof: non-consolidated financial institutions - unconditionally callable: 6.5 9.7 11.3 12.1 12.5 12.7 12.8 12.8 9.6 - others: 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 - Thereof: others (same rates repeated) - unconditionally callable: 6.5 9.7 11.3 12.1 12.5 12.7 12.8 12.8 9.6 - others: 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 ### Secured wholesale funding due - Conducted with domestic central bank - unconditionally callable: 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 - others: 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 - Not conducted with domestic central bank - Secured by Level 1 assets - unconditionally callable: 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 - others: 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 - Secured by Level 2A assets - unconditionally callable: 1.9 2.9 3.4 3.6 3.8 3.8 3.8 3.8 2.9 - others: 30.0 30.0 30.0 30.0 30.0 30.0 30.0 30.0 30.0 - Secured by Level 2B assets - unconditionally callable: 5.2 7.8 9.1 9.7 10.0 10.1 10.2 10.2 7.7 - others: 80.0 80.0 80.0 80.0 80.0 80.0 80.0 80.0 80.0 - Secured by other assets - unconditionally callable: 6.5 9.7 11.3 12.1 12.5 12.7 12.8 12.8 9.6 - others: 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 ### Outflows from derivatives - Thereof: unconditionally callable: 5.2 7.8 9.1 9.7 10.0 10.1 10.2 10.2 7.7 - Thereof: others: 80.0 80.0 80.0 80.0 80.0 80.0 80.0 80.0 80.0 ### Undrawn volume of committed credit/liquidity lines and Others - Run-off rates for “undrawn volume of committed credit/liquidity lines” and “others” are calculated as the size-weighted average of the Liquidity Coverage Ratio run-off rates for each subcategory from the more granular data provided by JFSA (no numeric values listed in table). *Source: IMF staff calculations.* ### Appendix Table 10. Japan: Cash Flow-Based Liquidity Analysis: Roll-Off Rates (1-Year Scenario) ### Cash-Inflows — Secured lending (reverse repos) - Secured by Level 1 assets - unconditionally callable: 100.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 - others: 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 - Secured by Level 2A assets - unconditionally callable: 100.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 - others: 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 - Secured by Level 2B assets - unconditionally callable: 100.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 - others: 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 - Secured by other assets - unconditionally callable: 100.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 - others: 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 ### Cash-Inflows — Contractual inflows from fully performing loans (1-Year Scenario) - Loans to non-consolidated financial entities - unconditionally callable: 100.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 - others: 100.0 100.0 90.0 90.0 90.0 90.0 90.0 90.0 90.0 - Unsecured loans to individual group entities and central branch (multiple consolidated categories) - For each of consolidated foreign non-financial legal entities; consolidated domestic non-financial legal entities; consolidated foreign financial legal entities; consolidated domestic financial legal entities; head office and branch accounts: - unconditionally callable: 100.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 - others: 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 *Source: IMF staff calculations.* ### Appendix Table 10. Japan: Cash Flow-Based Liquidity Analysis: Roll-Off Rates (1-Year Scenario) (concluded) ### Appendix Table 10–12 and Appendices VI–VIII (Selected Extracts) ### Cash-Inflows: Roll-Off Rates (1-Year Scenario) — Selected rows (Appendix Table 10, concluded) - Header: (In percent) — Cash-Inflows O/N 2day～1W 1～2W 2W～1M 1M～2M 2M～3M 3M～6M 6M～9M 9M～1Y - loans to central banks - thereof: unconditionally callable: 100.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 - thereof: others: 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 - loans to other entities - thereof: unconditionally callable: 100.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 - thereof: others: 100.0 100.0 95.0 95.0 95.0 95.0 95.0 95.0 95.0 - Contractual inflows from securities - thereof: unconditionally callable: 100.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 - thereof: other contractual inflows from securities: 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 - Inflows from derivatives - thereof: unconditionally callable: 100.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 - thereof: other inflows from derivatives: 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 - Others - thereof: unconditionally callable: 100.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 - thereof: others: 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 *Source: IMF staff calculations.* ### Counterbalancing Capacity — Haircuts and Sales of Assets (1-Year Scenario) (For U.S. dollars, euros) — Selected rows (Appendix Table 11) - Header: Counterbalancing Capacity Haircut Ratios O/N 2day～1W 1～2W 2W～1M 1M～2M 2M～3M 3M～6M 6M～9M 9M～1Y - HQLA — Cash and banknotes - thereof: Japan 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.0 - thereof: USA 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.0 - thereof: EU 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.0 - thereof: UK 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.0 - thereof: others 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.0 - Central bank reserves - thereof: Japan 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.0 - thereof: USA 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.0 - thereof: EU 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.0 - thereof: UK 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.0 - thereof: others 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.0 - securities issued or guaranteed by sovereigns, central banks, and PSEs - thereof: Japan 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.0 - thereof: USA 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.0 - thereof: EU 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.0 - thereof: UK 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.0 - thereof: others 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.0 - CP and bonds not included anywhere above - thereof: Japan 15.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 0 - thereof: USA 15.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 0 - thereof: EU 15.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 0 - thereof: UK 15.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 0 - thereof: others 15.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 0 - non-financial common equity shares - thereof: Japan 50.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 0 - thereof: USA 50.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 0 - thereof: EU 50.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 0 - thereof: UK 50.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 0 - thereof: others 50.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 0 - CP and bonds not included anywhere above (alternative rows) - thereof: Japan 25.0 50.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 0 - thereof: USA 25.0 50.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 0 - thereof: EU 25.0 50.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 0 - thereof: UK 25.0 50.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 0 - thereof: others 25.0 50.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 0 *Source: IMF staff calculations.* ### Counterbalancing Capacity — Haircut Ratios (1-Year Scenario) (For Japanese yen) — BoJ haircut ratios (Appendix Table 12) - Header: BoJ haircut ratios through discount window / BoJ haircut ratios for USD-funds supplying operations O/N 2day～1W 1～2W 2W～1M 1M～2M 2M～3M 3M～6M 6M～9M 9M～1Y - HQLA — Cash and banknotes - thereof: Japan 0.0 1.4 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 - thereof: USA 0.0 1.4 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 - thereof: EU 0.0 1.4 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 - thereof: UK 0.0 1.4 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 - thereof: others 0.0 1.4 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 - Central bank reserves - thereof: Japan 0.0 1.4 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 - thereof: USA 0.0 1.4 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 - thereof: EU 0.0 1.4 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 - thereof: UK 0.0 1.4 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 - thereof: others 0.0 1.4 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 - securities issued or guaranteed by sovereigns, central banks, and PSEs - thereof: Japan 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 - thereof: USA 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 - thereof: EU 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 - thereof: UK 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 - thereof: others 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 - CP and bonds not included anywhere above - thereof: Japan 4.0 4.0 15.0 15.0 15.0 15.0 15.0 15.0 15.0 15.0 - thereof: USA 4.0 4.0 15.0 15.0 15.0 15.0 15.0 15.0 15.0 15.0 - thereof: EU 4.0 4.0 15.0 15.0 15.0 15.0 15.0 15.0 15.0 15.0 - thereof: UK 4.0 4.0 15.0 15.0 15.0 15.0 15.0 15.0 15.0 15.0 - thereof: others 4.0 4.0 15.0 15.0 15.0 15.0 15.0 15.0 15.0 15.0 - non-financial common equity shares 1/ - thereof: Japan 59.2 69.0 69.0 69.0 69.0 69.0 69.0 69.0 69.0 - thereof: USA 59.2 69.0 69.0 69.0 69.0 69.0 69.0 69.0 69.0 - thereof: EU 59.2 69.0 69.0 69.0 69.0 69.0 69.0 69.0 69.0 - thereof: UK 59.2 69.0 69.0 69.0 69.0 69.0 69.0 69.0 69.0 - thereof: others 59.2 69.0 69.0 69.0 69.0 69.0 69.0 69.0 69.0 - CP and bonds not included anywhere above (alternative rows) - thereof: Japan 25.0 25.0 25.0 25.0 25.0 25.0 25.0 25.0 25.0 - thereof: USA 25.0 25.0 25.0 25.0 25.0 25.0 25.0 25.0 25.0 - thereof: EU 25.0 25.0 25.0 25.0 25.0 25.0 25.0 25.0 25.0 - thereof: UK 25.0 25.0 25.0 25.0 25.0 25.0 25.0 25.0 25.0 - thereof: others 25.0 25.0 25.0 25.0 25.0 25.0 25.0 25.0 25.0 - Equity shares and Others (concluded) - Equitiy shares not included anywhere above — thereof: Japan 50.0 50.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 - Others — thereof: Japan 50.0 50.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 - Others — thereof: USA 50.0 50.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 - Others — thereof: EU 50.0 50.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 - Others — thereof: UK 50.0 50.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 - Others — thereof: others 50.0 50.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 1/ The market haircut ratios on domestic equities are calculated based on the estimated price impact of sales of equities on equity prices (see Appendix IV for details). *Source: IMF staff calculations.* ### Spillovers Analysis Using Equity Returns (Appendix VI) - Methodology: - Uses Diebold and Yilmaz’s (2014) approach: a financial spillover from firm A to firm B is the share of variation in firm B’s equity returns shocks attributable to shocks to firm A’s equity returns. - Focuses on idiosyncratic shocks; excludes co-movement driven by common factors. - Data and estimation: - Sample: Total Firms (300); Japanese Firms (115); Foreign Firms (185) broken down into Banks (114), Other finance (67), Non financial (4). Japanese sample includes Citi Banks (4), International Active Banks (9), Other Regional Banks (59), Asset Management (3), Insurance (7), Securities & Investment Banking (16), Other Finance (14), Non Financial (3). - VAR of equity returns estimated using a lasso-estimator (Zou and Hastie 2005) due to high-dimensional sample. - Robustness: VAR augmented with VIX index as global control variable; alternative implementation using volatility of equity returns (Diebold and Yilmaz 2009) — results unchanged. - Model specification (text form): - Y is a vector of equity returns for all firms, X is either empty or the VIX. A(L) and B(L) are lag polynomials. ε is an error term. DH is the H-step ahead generalized forecast error variance decomposition matrix. - Focus on the 3-week ahead forecast error. - Identification: - Uses generalized forecast-error variance decomposition (GVD) via Pesaran and Shin (1998) to obtain order-invariant identification of shocks. *Source: IMF staff.* ### Gravity Model Analysis: Determinants of Spillovers (Appendix Table 1, estimates) - Specification: s_ij = exp(X_ij β + ε_ij) estimated by Poisson Pseudo-Maximum Likelihood (PPML) with firm fixed effects. - Selected coefficient estimates (columns labeled All firms / Japan only / Japanese banks): - Size: -0.147*** (0.000) ; -0.109*** (0.000) ; -0.124*** (0.000) - Revenue growth: 0.418*** (0.000) ; 0.168** (0.016) ; 0.185*** (0.009) - ROA: -0.104*** (0.000) ; -0.132*** (0.000) ; -0.095*** (0.000) - z-Score: -0.051*** (0.000) ; -0.049*** (0.000) ; -0.051*** (0.000) - Credit claims: -0.113*** (0.000) ; -0.147*** (0.000) ; -0.173*** (0.000) - Institutional ownership: 0.154*** (0.000) ; -0.135*** (0.000) ; -0.123*** (0.000) - Wholesale funding: -0.001 (0.649) ; -0.007** (0.012) ; 0.023*** (0.000) - Distance: -0.539*** (0.000) ; -0.001*** (0.000) ; -0.001*** (0.000) - Observations: 81,454 ; 10,864 ; 8,064 - Robust pval in parentheses. Significance: ***p<0.01, ** p<0.05, *<0.1 *Source: IMF staff.* ### Network Analysis of Balance Sheet Exposures — Model and Calibration (Appendix VII) - Model (Espinosa-Vega and Solé 2011 framework): - Simulates failure of a financial firm and tracks spillover effects through direct exposures and “domino effect” via third parties. - Considers two separate shocks: - a) Credit shock: impact of a financial firm defaulting on its liabilities to other firms. - b) Funding shock: impact of a firm deleveraging by withdrawing funding from other financial firms, forcing deleveraging and fire sales at a discount. - The approach tracks spillovers on both asset and liability sides. - Funding shock calibration (selected assumptions): - Short-term money placements and undrawn commitments: simulation assumes 0 percent roll-over or 100 percent funding withdrawal. - Fixed income: 86 percent rollover. Given the average duration of fixed income is seven years, it is assumed that 1/7 of fixed income cannot be rolled over. - Uncollateralized lending: simulation assumes 50 percent rollover. - Asset price loss: loss factor calibrated as 1/(1+δ) using share of illiquid investment assets. - δ assumptions: - Banks: δ is assumed to be 0.3 for city banks and 0.4 for regional banks (calibrated from share of illiquid securities available for sale including state and municipal securities, and other investment (including equity stakes) as percent of total investment). - Insurance: δ = 0.7 (derived from share of corporate bonds, other securities (investment securities), shares and other variable interest instruments, and other investments (policy loans + mortgage and loans + real estate + other)). - Securities firms: δ = 0.2 (derived from share of equity Investments in associates and other securities as a percent of total investment). - Credit shock calibration: - Assumes 100 percent loss given default. - Exposures computed net of collateral and other risk mitigation measures. - Alternative loss given default values used for robustness checks. - Combined scenario: - Credit and funding shock scenario uses the combined shocks with the same assumptions as above. *Source: IMF staff.* ### Stress Test Matrix (STeM) for the Banking Sector: Solvency, Liquidity, and Contagion Risks (Appendix VIII) — Selected entries - Domain: BANKING SECTOR: SOLVENCY RISK — Institutional Perimeter - Bottom-Up by Banks: - Institutions included: 4 banks - Market share: 65 percent - Data and baseline date: Institutions’ own data as of end of March 2017. Scope of consolidation: consolidated financial group. Coverage of sovereign exposures: Domestic and foreign sovereign bonds (disaggregated, granular data) in banking and trading books. - Top-Down by Authorities: - Institutions included: 18 banks; Largest internationally active and domestic banks plus several largest regional banks, excluding Japan Post Bank and Norinchukin bank. - Market share: 85 percent - Data and baseline date: Supervisory data as of end of March 2016. Scope of consolidation: banking group consolidated basis. Coverage of sovereign exposures: Domestic and foreign sovereign bonds (disaggregated by maturity) in banking and trading books. - Top-Down by FSAP Team: - Institutions included: 20 banks; All major banks including Japan Post Bank and Norinchukin bank. - Market share: 90 percent - Data and baseline date: Supervisory data as of end of September 2016. Scope of consolidation: banking group consolidated basis. Coverage of sovereign exposures: Domestic and foreign sovereign bonds (aggregated positions only) in banking and trading books. *Source: IMF staff calculations.* ### 2. Channels of ### 2. Channels of Risk Propagation ### Methodology and Models - Bottom-Up by Banks - Banks’ internal models, which are verified in advance by JFSA. - Satellite models for macro-financial linkages: banks’ internal models. - Deleveraging is not allowed. - Models for credit losses (credit costs), pre-impairment income, credit growth. - Model integrates solvency and credit growth channel via second-round effects. - Results include estimation with and without deleveraging. - Stress test horizon: 3 years. - Top-Down by Authorities - BoJ’s Top Down stress testing model based on supervisory data (balance sheet model). - Models for credit losses, pre-impairment income, credit growth; expert judgment. - Models based on Bayesian Model Averaging (BMA) methodology to limit selection bias. - Models will integrate solvency-funding liquidity feedback mechanism. - Sovereign risk parameters and feedback from financial stress to real economy estimated using Vitek’s (2015) DSGE model. - Results include estimation with and without deleveraging. - Stress test horizon: 3 years. - Top-Down by FSAP Team - IMF Top Down stress testing model (“workbox”) modified according to the JFSA supervisory reporting requirements. - Model is based on supervisory data (Balance sheet model). - Models for credit losses, pre-impairment income, credit growth; expert judgment. - Integrates solvency-funding liquidity feedback mechanism. - Sovereign risk parameters calculated using Vitek’s (2015) DSGE model. - Feedback from financial stress to real economy, second-round effects estimated using Vitek’s (2015) DSGE model. - Results include estimation with and without deleveraging. - Stress test horizon: 5 years. ### Scenario Analysis (Macro Shocks) - Common macro scenario elements (applies across approaches): - Shocks to GDP, inflation, interest rates, exchange rate, unemployment, property prices, equity prices, haircuts on government securities. - Losses in moderate adverse scenario: - The loss of real GDP in the moderate adverse scenario is about 5.0 percent over the 5-year horizon compared to the baseline scenario, equivalent to about 2 standard deviations of historical real GDP growth rate. - Losses in severe adverse scenario: - The total loss of real GDP in severe adverse scenario is about 7.6 percent over the 5-year horizon compared to the baseline scenario, equivalent to about 3 standard deviations of historical real GDP growth rate. ### Sensitivity Analysis - Counterparty credit risk shock (Bottom-Up by Banks): - Simulation of default of the two weakest counterparties (with the lowest credit rating) within the ten largest ones. - Market shocks (Bottom-Up and Top-Down by FSAP Team): - Shocks to real estate prices (20 percent decline). - 50 percent drop in domestic equity prices (both, Topix and Nikkei indexes). - Top-Down by Authorities: not applicable for counterparty shock. ### Risks, Behavioral Adjustments, and Buffers - Risks/factors assessed (common elements across methodologies): - Credit losses, profitability, funding costs, market risk, fixed income holdings of banks/sovereigns, counterparty risk, exchange rate, taxes. - Behavioral adjustments — Bottom-Up by Banks: - Static balance sheet: no change in composition nor deleveraging (negative credit growth) allowed on solo, sub-consolidated and consolidated basis for baseline and adverse scenarios. - Interest expenses cannot decline under the adverse scenarios. - Interest income on defaulted assets is not allowed under the adverse scenarios. - Non-interest expenses may decline in adverse scenarios, capped to the average observed in 2008−16. - Dividend payout: if bank meet all regulatory requirements, dividend payout as planned by bank. - Behavioral adjustments — Top-Down by Authorities: - Dynamic. No deleveraging (negative credit growth) in domestic exposures allowed on bank-consolidated basis for baseline and adverse scenarios. - Interest expenses cannot decline under the adverse scenarios. - Interest income on defaulted assets is not allowed under the adverse scenarios. - Dividend payout: if bank meet all regulatory requirements, dividend payout as planned by bank. - Behavioral adjustments — Top-Down by FSAP Team: - Quasi dynamic: no change in composition of balance sheet but deleveraging (negative credit growth) allowed on solo, sub-consolidated and consolidated basis for baseline and adverse scenarios. - Dynamic balance sheet adjustment allows for deleveraging as well as asset disposal in combined liquidity and solvency scenario only. - Interest expenses change under the adverse scenarios per the interest rate and funding costs projections. - Funding costs linked to macro scenarios as well as banks capital buffers. - Solvency and liquidity tests linked via funding costs and fire-sales effects. - Interest income on defaulted assets is not allowed under the adverse scenarios. - Non-interest expenses may decline in adverse scenarios, capped to the average observed in 2008−16. - Dividend payout: if bank meet all regulatory requirements, dividend payout as planned by bank. - Calibration of risk parameters: - Bottom-Up by Banks: PDs and LGDs from banks’ internal models verified by JFSA through the cycle, point in time for both credit losses and stressed RWA calculations; separate PDs/LGDs for foreign exposures; credit costs based on loan portfolio migration within the five supervisory categories; separate estimation for foreign exposures. - Top-Down by Authorities: PDs and LGDs through the cycle, point in time (without defaulted exposures) by major exposure class for both credit losses and stressed RWA calculations; EDFs for foreign exposures combined with historic credit losses. - Regulatory/accounting standards and metrics: - Hurdle rate: Basel III. - Capital metrics: Basel III; CET1, T1, CAR. - RWAs change because of change in PDs/LGDs. ### Reporting and Outputs - Common output presentation across methodologies: - Capital shortfall, system wide. - Number of banks that pass or fail; percentage of assets that fail. - Contribution to changes in RWAs. - Contribution to changes in income. --- ### Banking Sector: Liquidity Risk ### Institutional Perimeter and Data - Institutions included: - 16 banks on the bank-solo basis (or 14 banks on the financial group-consolidated basis). - Market share: - 76 percent of total banking sector assets. - Data and baseline date: - Supervisory data; Bank-solo basis or financial group-consolidated basis. - Baseline date: December 31, 2016. ### Channels of Risk Propagation and Methodology - LCR by currency (yen, U.S. dollar, and euro). - Cash flow-based analysis using maturity buckets by currency. - Link the cash flow-based liquidity analysis in U.S. dollar with solvency risk by increasing run-off rates for cash outflows including those from FX swaps, as well as funding costs and capital ratios. ### Risks, Buffers, and Market Liquidity - Risks: - Funding liquidity risk, rollover risk (roll-off rates). - Market liquidity risk: Markov regime-switching models estimate impact of market liquidity shocks on equity and JGB prices during stress; estimated price impact used to calculate corresponding haircut ratios on these assets. - Haircut ratios on equities are dynamic and depend on the total amount of sales by all banks. - Buffers: - Liquid assets/Counterbalancing capacity, assuming HQLA in different jurisdictions can be transferred without restrictions. ### Tail Shocks (Size of the shock) - 2-week mild stress scenario: - Run-off rates for yen-denominated retail deposits calibrated based on historical cases of capital injection, nationalization, and bankruptcy. - 3-month intermediate stress scenario: - Higher run-off rates on retail deposits, unsecured wholesale funding, and undrawn committed credit/liquidity lines on top of the mild stress scenario (comparable with LCR parameters). - 1-year severe stress scenario: - Higher run-off rates on secured wholesale funding (particularly FX swaps) on top of the intermediate stress scenario. ### Regulatory Standards and Reporting - Regulatory standards: - Threshold for cash flow-based analysis: net cumulative funding gap falls below 0. - Threshold for LCRs set to 100 percent. - Fail criteria for cash flow-based liquidity analysis in foreign currencies: need to use yen liquid assets (either through market or through the BoJ’s U.S. dollar funds-supplying operations). - Fail criteria for cash flow-based analysis in yen: need for the BoJ’s emergency liquidity assistance and ratios below 100 percent of LCR (for LCR type of tests). - Output presentation: - Number of banks with negative net cumulative funding gaps by currency and by consolidation basis. - Distribution of LCRs by currency and by type of bank. --- ### Banking Sector: Contagion Risk ### Institutional Perimeter and Data - Institutions included: - 10 banks and trust banks, 10 regional banks, 13 insurances, and 5 securities firms, and 10 major foreign firms. - Market share: - About 80 percent of the financial sector. - Data and baseline date: - Authorities’ data collected for FSAP, September 2016. ### Channels and Methodology - Balance-sheet model: Espinosa-Vega and Sole (2010). - Market-based model: Diebold and Yilmaz’s (2014) generalized forecast error variance decomposition approach. ### Tail Shocks (Size of the shock) - Pure contagion: - Default of institutions, 80 percent loss given default, 50 percent funding roll-over ratio. ### Reporting and Outputs - Output presentation: - Capital shortfall, by bank. - Capital shortfall, system wide. - Number of failed institutions given defaults. - Market-based analysis: Variance Decomposition (spillover contribution to equity prices). --- ### Appendix IX. Stress Test Matrix (STeM) for the Insurance Sector — Solvency Risk ### Institutional Perimeter and Data - Institutions included: - 7 life, 6 non-life. - Market share: - 73 percent in life, 92 percent in non-life (based on annualized new business premiums). - Data and reference date: - Statutory reporting. - Reference date: March 31, 2016. ### Channels of Risk Propagation and Methodology - Investment assets: - Market value changes after price shocks, affecting the solvency margin. - Sensitivity analysis: - Effect on available capital and solvency margin. - Time horizon: - Bottom-Up: Instantaneous shock. - Top-Down by IMF and Authorities: 3-year projection (only in the baseline and the severe adverse scenario) and Instantaneous shock. ### Scenarios and Sensitivities - Scenario analysis: - Baseline. - Moderate adverse scenario. - Severe adverse scenario. - Sensitivity analysis: - Longevity shock: permanent 20 percent decline in mortality rates. - Pandemic event: temporary 35 percent increase in disability/morbidity rates, temporary 10 percent increase in mortality rates. - Catastrophic events: (1) Great Kantō earthquake, (2) Typhoon Mireille, (3) Hurricane Andrew. - Sensitivity to market risk variables and interest rates (Top-Down). ### Risks, Buffers, Behavioral Adjustments - Risks/factors assessed: - Market risks: interest rates, share prices, property prices, FX rates, credit spreads. - Credit risks: default of largest financial and non-financial counterparty. - Underwriting risks: catastrophe events, lapses. - Summation of risks, no diversification effects. - Buffers: - Bottom-Up: Buffers inherent to product design and regulatory framework. - Top-Down: None. - Behavioral adjustments: - Bottom-Up: Management actions limited to non-discretionary rules in place at the reference date. - Top-Down: None. ### Regulatory Standards and Reporting - Regulatory/accounting standards: - J-GAAP. - Output presentation: - Impact on solvency margins. - Impact on net income (Bottom-Up). - Contribution of individual shocks. - Dispersion measures of solvency ratios and net income (Bottom-Up). - Dispersion measures of solvency ratios (Top-Down). *Source: cr17285 - 2. Channels of Risk Propagation (IMF FSAP material).* --- _Source: https://www.imf.org/-/media/files/publications/cr/2017/cr17285.pdf_