## _cr15173 ## Source details **Canonical URL:** [_cr15173](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/scr/2015/_cr15173.pdf) ## Other formats - [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/scr/2015/_cr15173.pdf.md) - [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/scr/2015/_cr15173.pdf.json) --- ### Executive summary and stress-testing scope - Exercise combined top-down solvency tests for bank holding companies (BHCs) and insurance companies, liquidity risk analysis for BHCs and mutual funds, market-price based stress tests, and network/contagion analysis; complemented supervisory (top-down) and company-run (bottom-up) stress tests. - Coverage: banks and nonbanks (including insurance companies and mutual funds), solvency and liquidity risks, and contagion risks. - Data constraints: relied largely on publicly available data; notable gaps in security-level granularity, interconnectedness data, and insurance group-level risk-based capital. ### Macroeconomic stress scenario (solvency design) - Two scenarios: baseline (informed by Blue Chip and IMF WEO as of January 2015) and stress (based on DFAST severely adverse, expanded to a 5-year horizon). - Stress calibration (severity similar to the 2007–09 recession): - Unemployment rate rose by 4 percentage points over a two-year period. - Real GDP was 4.5 percent lower than the baseline by the end of 2015 (GDP growth rates were negative for 5 quarters). - Equity prices fell by 60 percent in one year. - House prices declined by 25 percent over the first two years. - Corporate spreads rose by 330 basis points. - Mortgage rates increased by 80 basis points. ### High-level system outcomes - Base year system-wide CET1: 11.7 percent (base period). - System-wide CET1 ratio would fall by 2½ percentage points in 2015 (or 2¼ percentage points relative to the baseline scenario). - No BHCs fall below IMF hurdle rate in the first year; two BHCs fall below in 2016; an additional eleven BHCs do so during the recovery period. - Recapitalization needs peak in 2019 at 180 percent of 2014 net income (corresponding to 1 percent of 2019 nominal GDP). - SRISK peaked at approximately one trillion dollars in early 2009 and has fallen to 300 billion dollars (approximately 2 percent of GDP). ### Top-line policy recommendations (summary) - Conduct regular, comprehensive stress tests across major financial sub-sectors including macro-financial spillovers, liquidity–solvency interactions, and network effects. - Institute regular liquidity stress testing for banks and mutual funds; collect consistent interbank exposure data; link liquidity, solvency and network analyses. - Develop consolidated, group-level insurance stress tests for systemically important or materially interconnected groups; improve disclosure of market risk sensitivities. - Strengthen mutual fund liquidity guidance and perform regular system-level liquidity analyses. --- ### Bank Holding Companies (BHCs): scope, methodology and key solvency findings - Scope and data: - 31 largest BHCs, same as DFAST, constituting 85 percent of sectoral assets. - Publicly available consolidated data as of September 2014. - Capital framework and hurdle: - Assessed Basel III CET1 ratios versus hurdle = Basel III minimum (per transition schedule) + capital conservation buffer + GSIB surcharge (phased in). - GSIB surcharge ranged from 1.0 to 2.5 percent and was phased-in between January 1, 2016 and end of 2018. - Basel III phase-in minimum CET1: 4.0 4.5 4.5 4.5 4.5 4.5 (2014–2019 as presented). - Capital conservation buffer phase-in values: 0.625 1.250 1.875 2.5 (phase-in amounts). - Total CET1 ratio (I.+II.): 4.0 4.5 5.125 5.75 6.375 7.0 (by year as presented). - Modeling and projections: - RWAs projected via panel regressions relating RWA growth to total asset growth and interest rates; average standardized-approach increase of RWAs taken as 9 percent in 2015. - Aggregate provisions benchmark model projected 2014Q4–2016Q4 provisions at $390bn; panel models projected $280bn over same period. - Dividend rule: dividends payable out of current year’s profit using Basel III capital conservation rule; maximum payout = 2014 dividend payout ratio; sensitivity tested holding dividends constant at 2014 levels. - Key quantitative solvency results: - System-wide CET1 falls by 2½ percentage points in 2015 relative to base year; CET1 regulatory threshold (hurdle rate) in base: 4.5 percent. - Provisions in 2015 were four times higher than at end-2014 (subtracting -2.3 percentage points from CET1 ratio). - Trading losses contributed -0.6 percentage points to CET1 in 2015; negative AOCI and higher deductions each contributed -0.3 percentage points. - Implementation of the standardized approach raised total RWAs by about 9 percent in 2015. - Recovery period (2017–2019) drivers: assumed increases in RWAs, higher hurdle rate due to Basel III phase-in, sluggish interest income growth, and higher dividend distributions led to additional CET1 deterioration. - Recapitalization needs peak in 2019 at 180 percent of annualized 2014Q1–Q3 net income (≈ 1 percent of 2019 nominal GDP). - Sensitivities (selected): - Constant dividends at 2014 levels would subtract additional 30bps from system-wide CET1 each year. - Holding loans constant (no fall in downturn) and assets growing 2 percent raises system-wide CET1 fall by additional 80bps in 2015. - A scenario with constant loans and dividends and no operational RWAs: system-wide CET1 fell by 380 bps. --- ### BHC liquidity analysis (bank-by-bank liquidity metric) - Scope and metric: - Same 31 BHCs; liquidity metric = ratio of liquid assets to total cash outflow using FR Y-9C public data; haircuts from LCR; run-off rates calibrated on 2008/2009 episode. - Two stress horizons: 1 quarter and 3 quarters; sensitivity uses LCR run-off rates. - Key liquid-asset classification and haircuts (selected): - Level 1 assets: Cash and balances due from depository institutions: haircut 0; U.S. Treasury securities: haircut 0. - Level 2A: U.S. government agency obligations; MBSs (GNMA, FNMA, FHLMC): haircut 15. - Level 2B: Other domestic debt securities, other foreign debt securities, investments in mutual funds and other equity securities: haircut 50. - Main findings: - Most BHCs have enough liquid assets to meet a liquidity shock similar to the 2008/2009 event. - Several BHCs would face liquidity pressures from deposit outflows in short run and large unused commitments over longer horizon. - Under LCR-approximation (much larger shock), many BHCs’ liquid assets would be insufficient due to large wholesale funding withdrawals. - Metric caveats: - Included encumbered assets (inflates numerator) and excluded pledged assets (deflates numerator); inflows not considered; derivatives outflows not considered. - Public data lack liability tenor information, inflows from maturing transactions, and contractual derivatives terms. --- ### Network stress testing of largest GSIBs and contagion analysis - Coverage: six largest U.S. GSIBs (52 percent of BHCs’ total assets) due to data limits. - Method: Espinosa and Solé (2011) network stress-test using confidential FRB dataset with capital, credit/funding exposures, and single-name CDS notionals. - Baseline simulation parameters (examples): - 35 percent of funding provided by defaulting institution cannot be replaced. - Asset fire sales at 50 percent loss relative to book value. - PCA distress threshold: capital to assets ratio below 4 percent. - Baseline results: - Six BHCs hold enough capital to sustain a range of credit and funding shocks to individual counterparties; contagion risk appears contained under baseline parameters. - Direct exposures within the six-bank network are not large enough relative to initial capital to trigger second-round spillovers in most simulations. - Off-balance-sheet exposures (e.g., CDS) can dramatically change risk profiles; CDS can hedge exposures to some counterparties but increase exposure to others. - Severe contagion illustrative scenario (FSAP team): - Assumptions: 65 percent rollover of short-term funding (i.e., 35 percent not rolled), asset fire sales at 25 percent of book value, all BHCs start with capital = 8 percent of RWAs, distress defined as >5 percent initial capital loss in one round. - Under these assumptions, distress in one BHC could trigger three successive contagion rounds affecting multiple BHCs. - Recommendations: - Improve monitoring and stress-testing of off-balance-sheet exposures. - Expand data coverage (e.g., exposures to money market funds) and incorporate richer market dynamics (e.g., downward spirals) in contagion simulations. --- ### Insurance sector stress tests: scope, scenarios, valuation, and findings - Sample and scope: - 44 insurance groups in top-down sample (22 life, 15 P&C, 5 health, 2 credit/mortgage); sample represents 40 percent of domestic sector by gross written premiums; includes three FSB/FSOC-designated global systemically important insurers. - Data: SNL Financial, Bloomberg, NAIC aggregate data. - Scenario and valuation approaches: - Market stresses based on DFAST specifications; swaps rates assumed to fall between -21 and -143 basis points across maturities (1y to 30y); market shocks assumed instantaneous. - Two valuation regimes: “fully market-consistent” valuation and a statutory accounting exercise (amortized cost / impairment rules). - Hurdle rate: - In absence of group-level risk-based capital, IMF hurdle = complete extinction of shareholder equity; groups with negative shareholder equity after stress deemed distressed. - Key market-consistent results: - Total losses in shareholder equity = $267 billion (45 percent of pre-stress equity). - Life insurers: $187 billion. - P&C insurers: $72 billion. - Health insurers: $6 billion. - Credit insurers: $2 billion. - At least 17 distressed companies under the equivalent of the “severely adverse” scenario (16 life insurers and 1 credit insurer). - Under instantaneous market-consistent shocks, shareholder equity of distressed groups could turn negative. - Largest contributor: corporate bond portfolio credit spread increase accounts for 62 percent of overall loss. - Statutory-accounting exercise: - Reduction in statutory capital for life and P&C companies: $111 billion, representing 22 percent of pre-stress statutory capital. - Median statutory losses: life 26 percent; P&C 23 percent. - One life insurer becomes distressed (statutory capital negative); one other life insurer has reduction >75 percent. - Attribution: impairments 69 percent; catastrophic/pandemic 18 percent; defaults 13 percent; forced-sale realized losses 1 percent. - Catastrophe/pandemic sensitivities: - Florida hurricane (1-in-250) could cause insured losses around $80 billion (net of reinsurance). - California earthquake (1-in-250) could cause net losses around $34 billion. - Pandemic (1.5 additional deaths per 1,000; 1-in-200) could cost life industry $20 - $25 billion (gross). - Low-for-long interest-rate risk: - Industry spread between net portfolio yield and guaranteed credited rate declined by 57 basis points for 2006–2013. - Net investment yields need to remain above 4 percent to ensure profitability; current annuity guarantees include 1 percent floor; average net portfolio yield in 2013 was 5.11 percent. - Modeling caveats: - Separate accounts excluded; liability duration assumed (life liabilities exceed asset duration by two years; non-life/health by one year); hedging not modeled. --- ### Mutual funds liquidity analysis (top-down) — key methodology and results - Coverage and data: - Top-down liquidity analysis covering around 9,000 mutual funds (~80 percent of industry); CRSP holdings data; data cut-off 2014Q3. - Stress shock: - One-time, tail redemption shock defined as first percentile of net flow distribution by fund style (period 1998–2014Q3). - Asset-selling assumptions: - Approach 1 (“pro rata”): sell assets pro rata to preserve weights. - Approach 2 (“waterfall”): sell in descending liquidity order according to assumed proportions (examples: Cash 100%; Government securities 100%; MBS 100%; Equity 20% then 40% tiers; Corporate bonds 20% then 40%; Munis 100% in ordering). - Key quantitative findings: - Municipal bond sales to meet redemptions could be: - Three times dealers’ inventory under assumed ordering. - Four times dealers’ inventory under pro-rata selling. - Corporate bond sales could be up to seven times dealers’ inventory under severe stress. - Security-level data constraints: - CUSIP coverage limited (56 percent in CUSIP master; 8 percent in Mergent for fixed income); substantial equity reporting bias; security-level liquidity assessment not pursued further. - Caveats and policy implications: - Results subject to data limitations and aggregation bias; mutual funds investing in municipal and corporate bonds may face fire-sale risk. - Authorities encouraged to perform regular liquidity risk analysis for mutual funds and strengthen industry guidance. --- ### Market-price based analysis, Systemic Risk Dashboard, and Contingent Claims Analysis (CCA) - Multi-model dashboard: - Uses market-based measures (including SRISK), early-warning indicators, credit-risk networks, and SyRIN. - SRISK: fallen from ~one trillion dollars (early 2009) to 300 billion dollars (≈ 2 percent of GDP). - SyRIN and asset shares: - Banks <30 percent of system assets; nonbanks >70 percent. - Marginal contribution to systemic risk (MCSR) of High Yield and Insurance sectors is large relative to size. - CCA approach and findings: - CCA uses equity prices and accounting to estimate forward-looking default probabilities (Merton-type structural model mapped to one-year PDs). - Sample: 210 institutions across eight sectors (banks, insurers, asset managers, NBFIs, corporates, GSEs, foreign banks, foreign insurers). - Under the IMF/DFAST stress scenario: - Median one-year default probability for U.S. financial system projected to rise from 0.20 percent to 0.65 percent (≈ two-thirds of 2008–09 peak). - Banks, insurers, NBFIs, and foreign banks exit the “safe zone” (0.5 percent boundary) during peak stress in 2015–16; corporates, asset managers, and foreign insurers do not breach 0.5 percent. - Spillover findings: - Shocks to credit profiles of U.S. asset managers, NBFIs, insurers, and foreign banks increase U.S. bank credit risk most. - Spillovers from U.S. to rest of world can be large (especially foreign banking sector); spillbacks from rest of world to U.S. are relatively modest. - Example metric: excluding macro factors, non-U.S. domiciled banks and insurers account for 3 percent of median U.S. default probabilities; projected to fall to 2 percent at 2015–16 peak. --- ### Selected quantitative and structural indicators (highlights) - CET1 capital ratio doubled since crisis to 12 percent at end-2014Q3. - Leverage ratio (CET1/total assets) more than doubled to 8 percent since end-2008. - Total equity increased by 70 percent since crisis; retained earnings and surpluses doubled since crisis. - Delinquency rates: 3.5 percent (current); NPLs: 2.5 percent. - GSIBs: - Total assets ≈ $10 trillion; represent 75 percent of total assets of BHCs in stress test. - Securities (trading and AFS) account for 30 percent of GSIBs total assets. - Investment/AFS securities portfolio: MBSs 45 percent; Treasury securities ~20 percent. - Intra-financial assets: 22 percent of GSIB assets; intra-financial liabilities: 16 percent of total liabilities. - Mutual fund stress coverage: ~9,000 funds (~80 percent of industry). - Insurance sample aggregated assets: increased by 25 percent to $4.5 trillion (2009–2013). - Life insurers’ assets: $3.1 trillion; non-life: $1.1 trillion; health: $0.3 trillion. - Shareholder equity to assets (2014): life 8.3 percent; non-life 22.9 percent; health 33.1 percent. - Catastrophe scenario estimates: - Florida hurricane (1-in-250) net insured loss ~ $80 billion. - California earthquake (1-in-250) net insured loss ~ $34 billion. - Pandemic (1.5 excess deaths per 1,000) life industry gross cost $20 - $25 billion. --- ### Implementation and methodological priorities for authorities (selected) - Reexamine DFAST assumptions on loan and balance-sheet growth and dividend distributions to ensure historical consistency. - Link liquidity, solvency, and network analyses in an integrated systemic stress-testing framework. - Expand FRB data on interbank and off-balance-sheet exposures; include exposures to money market funds. - Establish a regular liquidity stress-testing framework for banks (CLAR noted as complementary to DFA). - For insurers: develop consolidated group-level stress tests for systemically important and materially interconnected groups; harmonize public disclosure of market risk sensitivities. - For mutual funds: regular top-down liquidity risk analysis; clarify and strengthen industry guidance; authorities to mobilize resources and standardize stress tests for funds with consolidated assets ≥ $10 billion. *Source: EXECUTIVE SUMMARY and IMF staff calculations prepared as part of the 2015 U.S. FSAP (cr15173).* ### EXECUTIVE SUMMARY ___________________________________________________________________________ 6 ### EXECUTIVE SUMMARY ### Overview of the stress testing exercise - A range of stress tests was used to quantify the potential impacts of risks and vulnerabilities in bank and nonbank sectors. - The exercise reflected a broader evaluation of potential risks embodied in the Risk Assessment Matrix (Appendix I). - Approaches included: - Top-down solvency tests for bank holding companies (BHCs) and insurance companies. - Liquidity risk analysis for BHCs and mutual funds. - Market-price based stress tests. - The exercise was informed by supervisory (top-down) stress tests performed by U.S. authorities and company-run (bottom-up) stress tests performed by BHCs. - Coverage: banks and nonbanks (including insurance companies and mutual funds), solvency and liquidity risks, and contagion risks. - IMF staff tests complement Dodd-Frank Act stress test (DFAST) results for BHCs. ### Data availability and constraints - Analysis relied largely on publicly available data due to constraints on sharing confidential supervisory information. - Public data were extensive but had notable gaps: - Lack of security-level granularity limited full-fledged liquidity stress testing for BHCs and mutual funds. - Data on interconnectedness among financial institutions have important gaps, though authorities assisted with a contagion stress test for a sample of large BHCs. - Insurance sector analysis constrained by valuation practices in the United States, complexity of the insurance business and its regulation, and the absence of group-level risk-based capital. ### Macroeconomic stress scenario (solvency test design) - Two scenarios: baseline and stress, with a range of sensitivity checks. - Baseline informed by the Blue Chip Economic Consensus and broadly reflected IMF‘s World Economic Outlook (WEO) projections as of January 2015. - Stress scenario based on the severely adverse scenario from DFAST, expanded to a 5-year horizon. - Scenario calibration (severity similar to the 2007–09 recession): - Unemployment rate rose by 4 percentage points over a two-year period. - Real GDP was 4.5 percent lower than the baseline by the end of 2015 (GDP growth rates were negative for 5 quarters). - Equity prices fell by 60 percent in one year. - House prices declined by 25 percent over the first two years. - Corporate spreads rose by 330 basis points. - Mortgage rates increased by 80 basis points. ### Bank holding companies (BHCs) — scope and capital framework - Stress tests covered the largest 31 BHCs—the same as under DFAST—constituting 85 percent of sectoral assets. - Tests used publicly available, consolidated data as of September 2014. - Solvency stress tests assessed Basel III Common Equity Tier 1 (CET1) ratios against a hurdle rate consisting of: - The regulatory minimum consistent with the Basel III transition schedule, - Augmented by the capital conservation buffer and a capital surcharge for Globally Systemically Important Banks (GSIBs), both phased in over the forecast period. - The IMF hurdle rates were more stringent than in DFAST, where the hurdle rate was consistent with the Basel III transition schedule. ### BHC solvency test key findings - The system-wide CET1 ratio would fall by 2½ percentage points in 2015 (or 2¼ percentage points relative to the baseline scenario). - No BHCs would fall below the hurdle rate in the first year, reflecting already high capital positions. - Capital ratios fall below the hurdle rate in subsequent years: - Two BHCs fall below the hurdle rate in 2016. - An additional eleven BHCs do so during the recovery period. - Recapitalization needs peak in 2019 at 180 percent of 2014 net income (corresponding to 1 percent of 2019 nominal GDP). - Drivers: - In the period of negative economic growth, effects on capital ratios were largely driven by credit and trading losses. - During the recovery period (2017–19), changes in capital ratios were mainly due to assumed increases in risk weighted assets, a higher hurdle rate due to phase-in of Basel III deductions, sluggish interest income growth, and higher dividend distributions. - Caveat: Results are conservative because BHCs would likely not increase balance sheets if doing so would push them below regulatory thresholds; nevertheless, some BHCs could face difficulties supporting credit growth after a large shock. ### BHC liquidity analysis — key findings - Staff’s liquidity risk analysis suggests that most BHCs have enough liquid assets to meet a liquidity shock similar to the 2008/2009 event. - A few BHCs would face liquidity pressures due to deposit outflows in the short run and large unused commitments over a longer stress horizon. - Under a much larger shock approximating the LCR run-off rates, liquid assets for many BHCs would not be sufficient due to large withdrawal of wholesale funding. ### Insurance sector stress tests — scope and limitations - Top-down group-level stress test covered 43 insurance groups. - Limitations: incomplete data, U.S. valuation practices, complexity of insurance business and regulation, absence of group-level risk-based capital. - Tests examined market stresses under a “fully market-consistent” valuation and included an additional exercise based on statutory accounting. - Comparison: Results broadly in line with the top-down stress test performed by the NAIC. ### Insurance sector stress test findings - Market stresses would cause substantial reduction of shareholder equity for life insurers under a “fully market-consistent” valuation; non-life insurers would be less affected. - The equivalent of the “severely adverse” scenario would lead to at least 17 distressed companies (16 life insurers and 1 credit insurer). - Shareholder equity of these groups would turn negative if materialization of market risk shocks occurred instantaneously, effectively reducing the full coverage of insurance liabilities with eligible assets. - Under current valuation regime, asset shocks would not be fully recognized instantly; impairments are required only when losses are significant or prolonged. - Statutory accounting exercise shows the sector in a more robust shape because mark-to-market losses would only partially be reflected in the balance sheet. ### Complementary analyses and contagion - The exercise included market-price based stress tests and network/contagion analysis for a sample of large BHCs (authorities assisted in contagion testing). - The IMF’s suite of tests intends to capture solvency, liquidity, and contagion channels across banks and nonbanks to provide a comprehensive assessment. *Source: EXECUTIVE SUMMARY, IMF staff note prepared as part of the 2015 U.S. FSAP (cr15173). The work was carried out between October 2014 and March 2015.* ### 8. Calculations for managed funds underscore the liquidity risk in the asset management ### 8. Calculations for managed funds underscore the liquidity risk in the asset management industry ### Mutual funds: liquidity risk and stress-test findings - Coverage and method: - Top-down liquidity analysis covering some 9,000 mutual funds representing around 80 percent of the industry. - Stress shock defined as a one time, tail event (1 percent) redemption shock. - Assets sold by mutual funds hit by a stress shock were compared to data on dealers' inventory. - Key findings: - Mutual funds that invest in municipal bonds and corporate bonds might face liquidity problems. - Tests illustrate the danger that these funds might sell their assets at a fire-sale discount to meet redemptions. - Results are subject to important data limitations. - Policy recommendation: - The authorities are encouraged to perform liquidity risk analysis for the mutual fund industry on a regular basis and further strengthen guidance to the industry in this regard. ### Network stress testing of large domestic GSIBs - Methodology: - Network stress-test simulating credit and funding shocks within a network of institutions and tracking contagion effects in terms of capital losses and path of bank failures. - Methodology allows assessment of systemic impact arising from existing off-balance sheet financial linkages (e.g., credit default swaps). - Coverage and limits: - Exercise focused only on six largest U.S. GSIBs, accounting for 52 percent of BHCs’ total assets, due to data limitations. - Findings: - The six largest BHCs hold enough capital to sustain a range of credit and funding shocks to individual counterparties within the network; contagion risk appears contained. - Direct exposures within the six-bank holding company network are not large enough relative to initial capital to lead to second-round spillovers. - Risk transfer instruments, such as credit default swaps, alter dramatically the risk profile of financial institutions. - Recommendations: - Importance of monitoring and stress-testing off-balance sheet exposures. - Need to expand the data on exposures included in the network (e.g., exposure of the BHCs to money market funds). - Consider richer market dynamics in simulations (e.g., downward spirals in the value of certain financial assets). ### Nonbank financial institutions and cross-sector systemic risk - Quantitative assessment: - Nonbank financial institutions and markets account for a majority of systemic risk. - Banks are sizable, accounting for some 30 percent of assets, and their contribution to systemic risk remains large. - Marginal contributions to systemic risk outside of the banking system have increased. - Interconnectedness: - Systemic risks posed by bond funds, money market funds, separate funds and equity managers may be small in absolute terms, but large in relation to their size, reflecting their interconnectedness with banks and other financial sector components. ### Market-price based stress tests: cross-sector spillovers and international effects - Role and limitations: - Market-price based analysis provides useful complementary information given the development, liquidity, transparency, and sophistication of U.S. financial markets. - Important limitations reflect modeling assumptions and the quality of underlying market data. - Results highlighted: - The system as a whole appears less vulnerable to large adverse macroeconomic shocks than during the recent crisis, but default probabilities would still rise to levels that would suggest significant systemic stress. - Cross-sector spillovers amplify the effects of shocks: - U.S. banks, insurers, and other non-bank financial institutions tend to be adversely affected by credit risk shocks originating in other domestic sectors. - Spillovers from insurance companies to banks are small when macroeconomic conditions are held constant, but large otherwise, suggesting insurer risk stems from their own vulnerability to macroeconomic shocks. - Spillovers from the United States to the rest of the world can be large: - Particularly true for the foreign banking sector. - The foreign insurance sector is most negatively affected by adverse changes in the credit profile of U.S. asset managers and insurers. - Spillbacks from the rest of the world appear to be relatively modest. ### Stress testing enhancements and recommendations (summarized) - General recommendations: - Conduct regular, comprehensive stress tests for all major financial sub-sectors that capture the impact of macro-financial factors, spillover and feedback effects between institutions and interactions between solvency and liquidity. - Conduct more intensive monitoring of systemic financial sector risks, including the use of market-based solvency and shortfall measures. - Banking recommendations: - Conduct liquidity stress testing on a regular basis. - Collect consistent interbank exposure data, and run network analysis on a regular basis. - Link liquidity, solvency and network analyses in a systemic risk stress testing framework. - Reexamine DFAST solvency stress test assumptions (e.g., regarding loan and balance sheet growth and dividend distribution) to confirm consistency with historical evidence. - Insurance recommendations: - Develop and perform insurance stress tests on a consolidated, group-level basis for systemically important or materially interconnected groups. - Improve public disclosure by requiring insurance companies to disclose market risk sensitivities in a more harmonized manner. - Mutual funds recommendation: - Develop and perform liquidity risk analysis for the mutual fund industry on a regular basis; further strengthen guidance to the industry on liquidity risk analysis. *Source: IMF staff summary of stress-testing analysis and recommendations contained in the 2015 FSAP chapter.* ### 25. Total equity has increased by 70 percent driven by retained earnings and surpluses, ### 25. Total equity has increased by 70 percent driven by retained earnings and surpluses, ### Equity and Capital Positions - Total equity has increased by 70 percent driven by retained earnings and surpluses, which have doubled since the crisis, largely in response to the higher regulatory requirements. - CET 1 capital ratio has doubled since the crisis to 12 percent at the end of 2014Q3. - The leverage ratio (defined as CET1 over total assets) has more than doubled to 8 percent since the end of 2008. ### Off-balance-sheet Activity - Off balance sheet activity has fallen since 2009 mostly due to lower holdings of derivatives, notwithstanding an increase in unused commitments. - Derivatives (credit equivalent) have fallen by 30 percent since the crisis. - Unused commitments have increased by 5 percent since the crisis and still represent the largest off balance sheet item. - Composition of unused commitments: - Consumer credit card lines: 40 percent. - Commercial and industrial loans: 25 percent. - Structural shifts in unused commitments since the crisis: - Unused credit card lines have fallen. - Unused commitments on commercial and industrial loans and loans to financial institutions have increased by 30 percent. - Securities lending is the second largest off balance sheet item with a share of around 20 percent of total off balance sheet activity. - Derivatives portfolio composition: - Interest rate contracts: 82 percent. - Foreign exchange contracts: 14 percent. - Credit derivatives have been cut in half since 2009 and represent a small proportion of derivatives activities; most contracts are related to purchased or sold investment grade credit default swaps. - Swaps and forward contracts dominate the derivative contracts. - Almost the whole derivative portfolio is held for trading. - Collateral and counterparties in OTC derivatives: - In most OTC derivatives transactions cash is the main collateral. - Major counterparties are banks and securities firms and non-financial corporate firms. ### Asset Quality and Underwriting Standards - Delinquency and NPL trends: - Delinquency rates have been cut in half since end 2009 and now stand at 3.5 percent. - NPLs have been cut in half since end 2009 and now stand at 2.5 percent. - Most bad loans consist of residential mortgage loans. - Non-real estate mortgage delinquent loans are at the levels before the crisis. - Net charge-off rates are considerably lower than in 2009 but still higher than before the crisis due to higher charge-offs for consumer loans. - Regulatory surveys from the OCC suggest looser underwriting in commercial real estate, commercial and industrial loans, and auto loans with some banks having significant exposure to subprime auto loans. - LTVs for CRE loans are approaching their pre-crisis levels, indicating continuing monitoring is needed. - The largest BHCs’ direct loan exposure to energy-related companies is in the range of 1.2 to 5 percent of their total loans. ### Profitability and Expenses - BHCs have posted all time high profits in 2014Q3, but there is a large dispersion of profitability indicators across BHCs. - Drivers of improved net income since the last FSAP: - Lower provisions, which have come down to pre-crisis levels. - Higher non-interest income. - Net interest margins continue to compress due to protracted low interest rates, increased holding of liquid assets because of regulatory requirements, and heightened loan competition. - Non-interest expenses are 35 percent higher in 2014Q3 than in 2008, partly due to litigation-related charges and cyber security protection. - Return metrics: - Return on equity (ROE) is about 50 percent smaller than before the crisis, mainly due to higher capitalization of BHCs. - Return on assets are 20 bps lower than before the crisis. - Many BHCs are seeking to enhance their ROE by looking for new business, principally through new loan growth or reconfiguration of business models, which increases the risk of relaxation of underwriting standards. - Litigation: The large BHCs continue to be plagued by very elevated litigation-related charges. ### Differences in BHC Business Models - Three broad types based on largest asset category: 1. Lending-focused BHCs: - High proportion of loans. - Lower leverage. - Lower off balance sheet activities. - Less involved in wholesale activities (reverse repo and trading). - Use deposits as a major funding source. - More profitable than other types. 2. Capital markets / trading-focused BHCs: - Large trading activities. - Less involved in lending. - More involved in reverse repo transactions. - Funded more in the wholesale market. - Higher leverage. - Higher NPLs. - May target riskier loans searching for yield. 3. AFS-securities / custodian-type BHCs (largest three BHCs): - Largest asset item pertains to AFS securities. - Less involved in lending, trading and repo transactions. - Higher leverage and off balance sheet activities. - Large deposit base. ### Global Systemically Important Banks (GSIBs) - Total assets of GSIBs have reached about $10 trillion and represent 75 percent of total assets of all BHCs included in the stress test. - When derivatives positions and securities financing transactions are added, their total exposure is about 35 percent higher than their total assets. - Around 25 percent of their assets are related to foreign exposures. - On average, GSIBs derive about ¼ of their total net revenue from foreign business. - The largest component of foreign loans pertains to commercial and industrial loans. - Securitization, mostly of mortgage loans, represents 11 percent of GSIBs assets but with large differences across BHCs. - Securities (trading and AFS) account for 30 percent of GSIBs total assets. - Investment and AFS securities portfolio structure: - MBSs: 45 percent. - Treasury securities: around 20 percent. - Together these represent almost 2/3 of portfolios. - Interconnectedness: - Intra-financial system assets represent 22 percent of GSIBs’ assets. - Intra-financial system liabilities represent 16 percent of their total liabilities. - Largest component of intra-financial system assets: fair value and potential future exposure of OTC derivatives (52 percent), followed by deposits (20 percent). - Most intra-financial liabilities pertain to deposits (47 percent; most of deposits were due to non-bank financial institutions) and OTC derivatives (35 percent). - Almost half of OTC derivatives are cleared through a central counterparty. ### Macroeconomic Scenarios Used in Solvency Stress Tests - Two macroeconomic scenarios examined over a five year horizon: a baseline and a stress scenario. - Scenarios developed by the FRB in consultations with the OCC and the FDIC (over July and August 2014). - Scenarios consisted of future paths of 28 economic and financial variables (six measures of economic activity and prices, four measures of developments in equity and property prices, six measures of interest rates and variables for the euro area, the United Kingdom, developing Asia, and Japan). - Baseline and stress for initial three years reflected DFAST supervisory baseline and severely adverse scenarios respectively; baseline was very similar to the IMF’s latest WEO projections for the first three years. - For 2018Q1 to 2019Q4, IMF staff extended paths for selected key indicators based on the latest WEO projections for the baseline; stress paths converge to the baseline by end of horizon. - Variables extended via simple OLS regressions and projected key indicators as exogenous variables; business and consumer interest rates and federal funds rate were added and projected from 2014Q4 using regression models. - Baseline features: - Sustained, moderate expansion converging to a growth rate of about 2 percent and unemployment rate reaching 4.4 percent by the end of 2019. - Gradual normalization in federal funds rate and Treasury yields starts in second quarter of 2015. - All asset prices (equity, house and commercial property) rise steadily. - Stress scenario features: - Similar in severity to the 2007–09 recession; based on the severely adverse scenario of the DFAST extended to 5 years. - Severely adverse scenario in the 2015 DFAST (shock applied from 2014Q4) characterized by: - 4 percentage point rise in the unemployment rate over a two-year period. - Real GDP on average 6.6 percentage points lower than the baseline in 2015. - Equity prices fall by 60 percent in the first year. - House prices decline by 25 percent over the first two years. - Corporate spreads rise significantly in 2015. - Mortgage rates increase by 80 basis points. - Market volatility rises to levels reached in the 2007–09 recession. - Short-term interest rates remain at zero by end of 2017, then normalize; long-term Treasury yields drop to 1 percent in 2014Q4 then edge up slowly. - Oil price assumed to rise to about $110 per barrel. - After 2017 most variables converge to baseline levels. - Scenario complemented with sensitivity analyses (including interest rate spike) and supervisory DFAST calculations incorporated projected losses from operational risk events. ### Capital Definitions and Hurdle Rates in the Stress Test - Capital definition applied corresponded to Basel III capital standards; applied to all BHCs recognizing only advanced approaches BHCs were subject to Basel III capital rules in 2014, and non-advanced approached BHCs became subject from January 1, 2015. - Hurdle rates included: - CET1 minimum requirement. - Capital conservation buffer. - GSIB surcharge. - Solvency stress test assessed BHCs CET1 ratios against regulatory thresholds consistent with the Basel III transition schedule while accounting for the capital conservation buffer and a G-SIBs capital surcharge (calculated using the BCBS framework). - GSIB surcharge: - Ranged from 1.0 to 2.5 percent, following FSB buckets, and was phased-in between January 1, 2016 and end of 2018. - Phase-in and transition considerations: - Phase in of deductions from CET1: 20 40 60 80 100 100 (as presented). - Phase out of existing AOCI capital adjustments: 38 06 04 02 00 0 (as presented). - Table 4 highlights phased Basel III minimums (percent) across 2014–2019 including: - I. Basel III phase-in minimum CET1: 4.0 4.5 4.5 4.5 4.5 4.5 (by year). - II. Capital conservation buffer: 0.625 1.250 1.875 2.5 (phase-in amounts). - III. Total CET1 ratio (I.+II.): 4.0 4.5 5.125 5.75 6.375 7.0 (by year). - IV. GSIB surcharge \10.25 0.50 0.75 1.00 (phase-in levels). - V. Total CET for GSIBs (III. + IV.) shown across years with ranges (e.g., 7.125-8.25, 8.0-9.5). - Capital ratio application distinctions: - Advanced approaches BHCs: revised (Basel III) capital framework applied to capital numerator and denominator, with projected RWAs and standardized RWAs switch including operational risk RWAs in projections. - Other BHCs: Basel I-based but Basel III used for capital numerator in Q4 2014; revised (Basel III) applied in subsequent years; denominator treatments varied (general approach RWAs, projected RWAs). - Tier 1 leverage ratio considered but detailed Basel III leverage ratio analysis was out of scope due to lack of granular off-balance-sheet data. *Source: IMF Staff (excerpts from the IMF Financial Sector Assessment Program chapter).* ### 39. CET1 capital for the base period was estimated for non-advanced approaches BHCs. ### _cr15173 - 39. CET1 capital for the base period was estimated for non-advanced approaches BHCs. ### CET1 estimation for non-advanced approaches BHCs - Non-advanced approaches BHCs became subject to Basel III capital rules from 2015Q1 and did not report CET1 capital (on Schedule HC-R, Part I.B. of FR Y-9C) as of September 2014. - SNL’s estimate of CET1 (after deductions and adjustments) was used for non-advanced approaches BHCs. - SNL calculates CET1 as: Tier 1 Capital - Non-qualifying Perpetual Preferred Stock - Preferred Stock & Surplus - Qualifying: Non Controlling Interests—Qualifying: Restricted Core Capital Elements—Qualifying: Mandatory Convertible Securities. ### Deductions from CET1 and phase-in strategy - Most deductions pertained to goodwill, intangible assets, and deferred tax assets (DTAs). - Key implementation points: - Deductions were reported by advanced approaches BHCs only (in Schedule HC-R). For non-advanced approaches BHCs, intangible assets deducted from CET1 were approximated by intangible assets other than goodwill and mortgage servicing assets (MSA) (from Schedule HC-M) adjusted for deferred tax liabilities (DTLs) associated with intangible assets. - A deduction related to goodwill, net of deferred tax liabilities, was reported by all BHCs. - A deduction related to DTAs for non-advanced approaches BHCs was approximated by DTAs deducted from Tier 1 capital. - The nominal value of all deductions was assumed to stay constant over the stress testing horizon (as in the supervisory DFAST). Each deduction was considered separately since there was no transition provision for goodwill while intangible assets and DTAs followed the Basel III transition provisions. - Deductions were calculated by multiplying intangible assets and DTAs by the transition provision factor. - Only the difference between the deduction in period t+1 and t was subtracted from CET1 capital in period t since CET1 in period t was already defined as CET1 after adjustments and deductions. - No assumptions were made about banks' behavioral responses to phase-ins. - Numeric notes from the source: - Other adjustments and deductions account, on average, for 3.2 of total deductions or 0.3 percent of CET1 after adjustments and deductions. - The amount for intangible assets and DTAs reported in Schedule HC-R (BHCA P842, P843) in 2014Q3 were already a result of applying transition provisions in 2014 (20 percent). ### Treatment of Accumulated Other Comprehensive Income (AOCI) - Treatment reflected Basel III transition arrangements: - For advanced approaches BHCs, only 20 percent of AOCI was incorporated into CET1 capital in 2014 and an additional 20 percent in every year after 2014. - It was assumed that non-advanced approaches BHCs would opt out of including AOCI (consistent with supervisory DFAST). - In the IMF top-down test aggregate AOCI was modeled (structure of AOCI not publicly available). Only the difference between AOCI in period t+1 and t was added to CET1 capital in period t since CET1 in period t was already defined as CET1 after AOCI. - No assumptions were made on possible behavioral responses such as reallocating securities from AFS to HTM as higher proportion of AOCI flows through to CET1 capital. ### Risk-weighted assets (RWAs) projection methodology - Total RWAs were projected for each participating BHC. Modeling separately credit RWAs and market RWAs was challenging with publicly available data. - Approach: - Year-on-year growth rate of total RWAs was modeled in a panel regression model with fixed effects as a function of year-on-year growth rate of total assets. - Interest rates were added as an exogenous variable to make RWA projection risk sensitive. - It was assumed that the relationship between RWAs and total assets found before 2016 would hold also after 2016, despite BHCs being able to qualify for advanced approach credit risk RWAs from January 1, 2016. - Operational risk RWAs: - Included in total RWAs for advanced approaches BHCs that exited the parallel run (given Collins amendment requirements). - Operational risk capital charge for non-advanced approaches banking organizations and advanced-approaches BHCs that have not exited the parallel run was not applied. - It was assumed that the share of operational RWAs in total assets stays the same over the stress testing horizon; RWAs for advanced approaches BHCs that exited the parallel run were projected without operational RWAs, and projection of operational RWAs was then added to projected RWAs to calculate projected total RWAs. - Increase in RWAs due to standardized approach: - The increase in total RWAs due to implementation of standardized approach was applied to projected total RWAs starting in 2015 onwards. - Credit RWAs under the standardized approach were not possible to calculate exactly with available data; the increase was applied based on the calculated average increase in RWAs reported for the 2014 DFA stress testing exercise. - The average increase of RWAs due to implementation of standardized approach was 9 percent. - BCBS Basel III monitoring exercise estimates cited: increase in total RWAs to be 8.3 percent for Group 1 banks and 6.5 percent for Group 2 banks. - Market risk threshold and conversion: - A BHC is subject to market risk capital rules if consolidated trading activity equals: (1) 10 percent or more of the bank holding company’s total assets as reported in its FR Y-9C report for the previous quarter, or (2) $1 billion or more. - A BHC’s market risk equivalent assets equal its measure for market risk multiplied by 12.5 (the reciprocal of the minimum 8.0 percent capital ratio). ### Models, data, and behavioral assumptions - Data and modeling: - Quarterly data from 1991 to 2014Q3 from FR Y-9C report and a set of panel regression models were used to forecast each BHC’s main balance sheet and income statement components. - Projections of balance sheets were used for projecting total RWAs and income statement items. Projections of RWAs and net income, with assumptions on dividend distribution, Basel III deductions and AOCI determined capital requirements over the stress testing horizon. - Asset disposals and acquisitions over time were not considered (unlike DFAST). - Modeling choices and robustness: - When panel modeling was not appropriate, aggregate-level modeling was used and results distributed to BHCs by market share or 2014 DFAST results. - Several model specifications were considered: lags of dependent variable, fixed effects versus pooled estimates, different dependent variable definitions (year-on-year, q-o-q and shares), and handling of outliers. - The “best” model was chosen based on R^2 and statistical significance and sign of coefficients. Some variables were excluded due to multicollinearity or marginal contribution to R^2. - Many models perform best with variables specified as year-on-year growth rates. Only one equation (the loan equation) included a lagged dependent variable. ### Balance sheet growth and loan projections - Total assets: - Growth rate of total assets was assumed equal to growth rate of the largest asset category (often total loans), reflecting smaller volatility of total assets’ growth. - Year-on-year growth rates of total assets were “filtered” and projected using pooled panel regression model with year-on-year growth rate of loans as the only independent variable. - Loans: - Modeling individual BHC’s total loans used a panel, fixed-effects model of year-on-year growth rate of net total loans at bank-by-bank level. - Independent variables included: year-on-year growth rate of real GDP, year-on-year changes in interest rates, and the VIX. The model also included lagged dependent variable to account for persistence. - It was assumed that lower economic activity, increases in interest rates and higher market turbulence (proxy for risk aversion) would lower loan demand. - Growth rate of deposits was assumed equal to growth rate of loans (historical correlation equal to 0.85). - Sensitivity analysis considered different growth rates of loans and total assets in the stress scenario. - Trading assets for broker-dealers: - Total assets of Goldman Sachs and Morgan Stanley were primarily determined by trading assets, projected separately in a panel regression with fixed effects where growth of trading assets was driven by VIX, real GDP growth rate, and the federal funds rate. - Growth rates of total assets for these two companies over the stress horizon were assumed equal to growth rate of projected trading assets. - Modeling limitations: - Other balance sheet items (cash, HTM securities, ATM securities, repos, trading assets and other assets) could not be consistently modeled with available data; this implied projecting specific categories of credit risk RWAs was not possible and differences across BHCs’ stress test results could not be explained by differences in loan portfolio composition. - Classification issues: owner-occupied nonfarm nonresidential properties were treated as CRE loans due to reporting aggregation by many BHCs. _Italic: Source: IMF Staff._ ### 50. The projections of revenues, expenses, and loan losses were based on the IMF’s ### _cr15173 - 50. The projections of revenues, expenses, and loan losses were based on the IMF’s ### Projection methodology and data sources - Projections of revenues, expenses, and loan losses were based on the IMF’s projections of the balance sheet for each BHC over the planning horizon. - Most components of pre-provision net revenue (including components of net interest income, noninterest income, and noninterest expenses, Table 8) were modeled using data on historical revenues and operating and other non-credit-related expenses reported on the FR Y-9C in a simple panel regression model framework (Tables 23 and 24). - Projections of all independent variables were taken from the scenarios. - The choice of regressors started from the New York Fed’s CLASS model (Hirtle and others, 2014) and models of net charge-off rates in Francisco B. Covas, Ben Rump, and Egon Zakrajsek (2014). ### Provisions for loan losses - Benchmark (first) approach: - Aggregate provisions modeled as the ratio of total provisions over aggregate net loans as a function of: - real GDP growth, - differenced unemployment rate, - credit spreads, - growth rates of house prices, - growth rate of VIX, - with interaction term between growth rate of VIX and a dummy equal to 1 when VIX growth is positive to capture amplified effects in stress periods. - Projected aggregate provisions distributed among BHCs using each bank’s share of provisions in total provision in the 2014 DFAST exercise. - This aggregate provisions model captured the spike in provisions during the 2008/2009 crisis. - Sensitivity (second) approach: - Assumed provisions equal net charge offs (as in the last FSAP stress testing exercise). - Used models of net charge-off rates by loan types (Table 9) that relate historical net charge-offs to macroeconomic and financial market variables; predicted net charge-off rates multiplied by loan balances. - Growth rate of each loan type assumed equal to projected growth rate of total loans. - Limitations noted: - Provisions increase more quickly than realized net charge-offs during stress times. - Net charge-offs lag provisions by 2 to 3 quarters. - Sensitivity also explored projecting total net charge-offs and total provisions (instead of by loan types). ### Net interest income, non-interest income, trading income, AOCI, and expenses - Net interest income: - Projected using fixed effects panel regression with dependent variable = annual difference of net interest income. - Explanatory variables: annual difference of (total net loans × loan interest rates) and annual difference of (total interest bearing deposits × deposit rate). - Projections of loans taken from BHC balance sheet projections; interest-bearing deposits assumed to grow at year-on-year growth rate of loans. - Deposit interest rates assumed equal to the Federal funds rate. - Loan interest rate defined as a weighted average of mortgage, business, consumption lending rate and federal funds rate, with weights reflecting each BHC’s loan portfolio structure (real estate loans, CRE loans, business loans, consumer loans and loans to financial institutions, Table 7). - Non-interest income excluding trading income: - Projected via panel regression with dependent variable = non-interest income (excluding trading income) over total assets. - Independent variables: growth rate of VIX, unemployment rate, lending rates. - Rationale: higher VIX and higher unemployment expected to reduce fee- and commission-related income; higher lending rates may shift banks toward interest income. - Trading income: - Modeled as aggregate trading income (including gains on AFS and HTM securities) over total aggregate assets. - Independent variables: year-on-year growth rate of VIX, interaction of VIX growth and positive-growth dummy, change in credit spread, change in term spread. - Projected trading income distributed among BHCs based on 2014 shares of trading income and gains on AFS and HTM securities. - Non-interest expenses: - Modeled as year-on-year growth rate in a panel regression with year-on-year growth rate of total assets as sole independent variable and fixed effects. - Assumption: non-interest expenses depend on size of business and thus size of balance sheet. - Taxes: - Set at 28 percent—the pooled average level of the tax rate over the last 25 years. - Extraordinary items and minority interest: - Assumed equal to zero. - AOCI: - Aggregate total AOCI modeled as ratio of AOCI to aggregate assets in regression with yearly change in BBB yields, 10 year Treasury bond, real GDP growth, and interaction of real GDP growth with a dummy equal to 1 when GDP growth is negative. - Projected total AOCI distributed among BHCs using proportions from 2014 DFAST AOCI losses. - Items not considered: - (i) losses related to operational risk events, mortgage repurchases, or OREO, - (ii) HFS/FVO loan losses (data not publicly available), - (iii) deferred tax assets (DTAs). ### Models of net charge-offs (loan-type mappings in Table 9) - Real estate loans: - Residential Real Estate: Real GDP, unemployment rate, house prices, mortgage rates, VIX. - Commercial Real Estate: Real GDP, unemployment, CRE prices, corporate interest rates, VIX. - Business loans: Unemployment, VIX, corporate spreads. - Consumer loans: Unemployment, house prices, consumer interest rates. - Loans to financial institutions: Real GDP, federal funds rate, VIX. - TOTAL: Real GDP, unemployment rate, house prices, lending rates, corporate spreads. - Note: Net charge offs for loans to government were not considered as they were very small in the last 15 years. ### Capital action assumptions and dividend policy - Assumed BHCs do not issue new shares or make repurchases during the stress test horizon (similar to supervisory DFAST), except employee stock ownership plan issuances considered in supervisory DFAST. - Dividend distribution rule: - Dividends payable out of current year’s profit using the Basel III capital conservation rule taking into account transition provisions and GSIB surcharge (Table 10). - Dividends paid out of current period net income after taxes by BHCs in compliance with the capital requirement equal to the hurdle rate. - Maximum allowed dividend payout equal to dividend payout ratio in 2014. - If a bank fell below the hurdle rate before dividend distribution, it was capital constrained and followed the schedule in Table 10; if a bank fell below the hurdle rate because of dividend distribution, payout limited to ensure hurdle rate not breached. - Rule applied only if bank earned positive net income; negative net income implies no dividend payout. - Sensitivity test: assumption that dividends remained fixed (in nominal terms) at their 2014 value was explored. ### Sensitivity analyses performed - Variations included: - (i) loans and balance sheet grow at similar rate as supervisory DFAST; - (ii) constant dividend distribution like supervisory DFAST; - (iii) hurdle rates from DFAST and assuming total RWAs = credit risk RWAs + market risk RWAs (excluding operational RWAs); - (iv) assuming all DFAST elements in same scenario (constant dividends, fixed loan supply, total RWAs = credit + market RWAs only). - Additional sensitivity tests: - (i) including an oil price shock in the stress scenario; - (ii) calculating impact of a large interest rate shock in first year of stress scenario; - (iii) using different measures to calculate provisions in stress scenario; - (iv) extending scope to include all banking organizations with consolidated assets of $50 billion or more subject to Basel III. ### Results — solvency stress test and key statistics - Base period capital and early impact: - Base year system-wide CET1: 11.7 percent. - CET1 regulatory threshold (hurdle rate): 4.5 percent. - No BHCs fell below the hurdle rate during the first year of severe economic distress (2015). - System-wide CET1 ratio fell by 2½ percentage points in 2015 relative to the base year, or 2.2 percentage points relative to the baseline scenario in 2015. - First quarter 2014Q4 dynamics: - Net income fell from annualized cumulative high in 2014Q3 of about $120 billion to a $30 billion net loss in 2014Q4 due to initial negative shock; credit losses doubled compared to 2014Q3 levels. - Lower RWAs (due to slowdown in lending) and lower total assets cushioned negative effect on CET1 in 2014Q4. - Capital conservation rule limited dividend impact from 2014Q4 onwards. - Drivers in 2015 downturn: - Provisions and trading losses were main drivers of CET1 dynamics. - Provisions in 2015 were four times higher than at end-2014 (subtracting -2.3 percentage points from CET1 ratio which is 8 times higher than in the base period). - Trading losses contributed -0.6 percentage points to CET1. - Negative AOCI and higher deductions from CET1 each contributed -0.3 percentage points. - Net interest income fell by 5 percent compared to end-2014 due to lower loan demand despite higher spreads. - Noninterest income fell by more than 10 percent. - Implementation of the standardized approach to calculate RWAs raised total RWAs by about 9 percent in 2015, but RWAs stayed stable relative to 2014 due to lower loans and total assets. - Recovery and subsequent years: - 2016: Two BHCs fall below the hurdle rate in first year of recovery; system-wide CET1 falls an additional 0.3 percentage points in 2016. One BHC fell due mainly to relatively low profitability in the baseline. - 2017–2018: Eleven additional BHCs fall below the hurdle rate by end of stress-testing horizon due to rapid balance-sheet expansion. System-wide CET1 falls by additional 0.6 percentage point in 2017–2018 after recovering in last year by 0.1 percentage points. - By 2019, despite recovery in system-wide capital position, a number of BHCs still fall below hurdle rates. - Drivers in recovery: higher RWAs due to balance-sheet expansion; provisions and non-interest expense contributed to net losses in 2016; in final three years RWAs subtract 80bps from CET1 each year and non-interest expenses grow due to balance-sheet expansion by 10 percent annually. - Baseline projections: - CET1 ratios projected to fall in the baseline driven by increases in RWAs despite positive net incomes. - System-wide CET1 ratio fell by 90bps in 2019 compared to base period. - Dividend distribution subtracted on average 40bps from CET1. - Even under baseline, 3 BHCs fell below the hurdle rate in 2018 and seven more in 2019; these are the same BHCs that failed under the stress scenario mainly due to rapid balance-sheet expansion. - Recapitalization needs: - Recapitalization needed to bring all BHCs to hurdle thresholds peaks in 2019 at 180 percent of their annualized 2014Q1–Q3 net income. - This peak recapitalization corresponds to 1 percent of 2019 nominal GDP. ### Sensitivity results (selected quantitative impacts) - Loan dynamics (loans and total assets do not fall during downturn; loans remain at 2014Q3 level, total assets grow at yearly rate of 2 percent): - CET1 ratios fall more in 2015 and less in subsequent periods. - System-wide CET1 ratio fell by additional 80bps due to higher RWAs compared to the stress scenario but increased on average by 40bps a year in the last three years due to lower increases in RWAs. - Dividend distribution: - Holding dividends constant at 2014Q3 level (as in DFAST) would subtract additional 30bps from system-wide CET1 each year. - DFAST hurdle rate and no operational RWAs: - Despite lower hurdle rates and higher initial capital ratios, same two BHCs would fall below thresholds; additional 3 BHCs would fall below threshold in recovery period. - Constant loans and dividends, no operational RWAs: - System-wide CET1 fell by 380 bps in the scenario variation examined. *Source: IMF Staff.* ### 2015. Under this scenario two additional BHCs would fall below the regulatory minimum of ### _cr15173 - 2015. Under this scenario two additional BHCs would fall below the regulatory minimum of ### Scenario shocks and sensitivity analysis - Regulatory minimum considered: 4.5 percent, mainly affected in 2015 due to dividend distribution which, in the benchmark, are cut to zero due to negative net income in 2016. - Oil price shock: - Oil prices fell by 25 percent 2014Q4 and additional 60 percent by the end of 2015. - This oil price shock would increase total provisions by around 3 percent. - Interest rate shock: - Applied to the 3-month Treasury yields of 450bps in 2015. - Higher short-term rates would reduce loans growth rate and subsequently total assets and RWAs. - Higher short-term rates would result in higher credit losses (by 11 percent in 2015) and losses on AOCI (almost twice as higher in 2014Q4–2015Q4 than in the stress scenario). - Overall, these interest-rate-induced losses would not be large enough to make a material impact on the results relative to other variables. - The effects of higher interest rates on GDP growth were not analyzed. - Provisions model sensitivity: - Aggregate provisions model projected provision in the period 2014Q4–2016Q4 at the level of $390bn. - A panel model of total provisions and a panel model of net charge offs both projected total provision of $280bn over the same period. - Compared to the benchmark stress scenario, the panel-model projection would correspond to system-wide CET1 ratio increase of about 80bps in 2015 and 15bps in 2016. - Panel models project lower provision in the first two years but more persistent projections resulting in higher total provision over the whole stress testing horizon than the aggregate provisions model. ### Structure of loan losses - Modeling net charge-offs by loan type shows distribution of losses by exposure: - Around 35 percent of losses in 2015–2016 come from consumer loans. - 25 percent comes from residential real estate exposures. - The share of commercial real estate and business loans is similar to 2008–2009. - Loans to financial institutions (and other loans) would account for a much higher proportion of losses given their size in 2014, which is much higher than before the crisis. ### Expanded coverage of the stress test - Expanded coverage scope: include all deposit-taking financial institutions with asset size of $50 billion and more that reported Basel III capital ratios as of 2014Q3. - Inclusion required adding: - Additional two BHCs and two large savings banks. - Results for added institutions: - One financial institution would fall below the threshold in 2016 mainly due to higher provisions and relatively low capitalization in the base period. - Total recapitalization needs would increase by 5 percent due to the recapitalization of this financial institution. ### CET1 ratio sensitivity (notes from Figure 4) - The distribution of projected CET1 ratios for 31 BHCs is summarized using quartiles, medians, and extremes; system-wide CET1 ratios defined as projected total CET1 capital over total RWAs (sum over all BHCs). - Sensitivity experiments referenced include: - Benchmark (stress scenario) - Constant loans - Provisions = Net charge offs - Constant dividend - Interest rate shock - No operational RWAs - No operational RWAs, constant div., constant loans *Source: IMF Staff calculations.* ### Network analysis for large BHCs — data and methodology - Purpose: assess potential spillovers among six largest U.S. G-SIBs using the Espinosa and Solé (2011) network stress-test methodology. - Confidential dataset maintained by FRB contained: - Capital levels, credit and funding exposures, and credit default swaps (CDS) contracts for six systemic BHCs. - The six institutions hold capital of around 9 to 12 percent of their total assets. - Credit exposure data: estimates of credit loss borne if one of the other five institutions defaulted, including direct loan losses and indirect losses (derivative replacement, losses on owned securities issued by the defaulting BHC). - Funding exposure data: amount of funding received from the other five counterparties (secured and unsecured funding, repo transactions of all maturities). - Risk transfer data: single-name notional CDS exposures between the six BHCs (does not include CDS index trades or CDS options). ### Simulation design and main results - Simulations conducted quarterly for each quarter between 2013Q1 and 2014Q3. - Four sets of simulations: 1. Domino effects triggered if each of the six BHCs defaulted one at a time on their credit commitments. 2. Credit-plus-funding event: default leads to liquidity squeeze for institutions funded by the defaulting institution, compounding credit shock with a funding shock and associated fire sale losses. 3–4. Builds on the first two sets by incorporating credit default exposures (CDS) of each BHC. - Key assumptions in baseline network stress-test (example parameters used in initial simulations): - 35 percent of the funding provided by a defaulting institution cannot be replaced. - Asset fire sales conducted at 50 percent loss relative to book value. - A bank is considered under distress if its capital to assets ratio falls below the PCA threshold of four percent. - Main findings: - The six BHCs hold enough capital to sustain a range of credit and funding shocks to individual counterparties within the network; most simulations did not trigger contagion chains among the six institutions. - Direct exposures within the six-BHC network are not large enough relative to initial capital to lead to second-round spillovers. - Losses could still be substantial for some BHCs (example: BHC 4 could suffer losses of up to 2.5 percent of its initial capital), potentially leading to funding difficulties if market concerns arise. - Off-balance sheet exposures (e.g., CDS) can dramatically alter risk profiles; for example, CDS affect BHC 5’s losses differently across counterparties—hedging exposures to BHCs 1, 2, and 3, but increasing exposure to BHCs 4 and 6 via CDS. - Robustness tests by the FRB: - Assigned extreme values to parameters measuring funding squeeze severity, loss of asset value due to fire sales, and capital level defining distress. - Robustness test scenarios included: - Assuming none of the funding provided by the defaulting institution can be replaced. - Assuming fire sales of assets take place at 1/4 of the book value of the assets. - Raising to 6 percent the capital threshold below which distress in one bank is transmitted to others. - Results appear robust to these stressed parameter values. ### Severe contagion scenario (FSAP team calculations) - A combination of severe factors can trigger contagion chains: - Assumed only 65 percent of short-term funding provided by a defaulting institution is rolled over by other market participants (i.e., 35 percent not rolled over). - Assumed asset fire sales take place at 25 percent of book value. - Assumed all BHCs have initial capital levels equivalent to 8 percent of risk-weighted assets. - Defined distress: if a bank loses more than 5 percent of its initial capital in one round, it suffers “distress” which triggers the next round of contagion. - Example contagion path under these assumptions: - Distress in BHC 1 would lead to three successive rounds of contagion affecting BHC 4 first, then BHCs 5 and 6 in a second round, and finally BHC 3. BHC 2 would not go into distress in this scenario. - Caveats: these calculations are illustrative and require more detailed data (initial capital levels, exposures to specific financial instruments, reliance on short-term funding) to construct more accurate scenarios. *Source: IMF Staff calculations and FRB data reported to IMF staff.* ### 81. There are important differences between the IMF approach and the authorities’ ### _cr15173 - 81. There are important differences between the IMF approach and the authorities’ ### Differences between IMF top-down stress test and authorities’ (DFAST/FRB) approach - Hurdle rates in the IMF stress test were more stringent. - Differences in calculation of risk-weighted assets (RWAs) and capital ratios, including the inclusion of operational RWAs in the IMF stress tests, on average resulted in lower ratios for advanced approaches BHCs. - Level of data granularity differed: the authorities used much more granular data. - Dividend distribution rule: IMF used a rule that was a function of the level of capital and was a less conservative assumption than the one used by the authorities. - Projection approach: - IMF stress test projected loans and total assets. - DFAST forecasted loans and total assets conditional on the assumption that credit supply is maintained at long-run historical levels. - Some losses were not covered by the IMF stress test due to lack of historical data. - Stress horizon: - Authorities used a 9-quarter stress horizon and 3-year forecast scenarios. - IMF test focused on a stress testing horizon that spanned over 5 years. - IMF forecasts over the first three years were consistent with the forecasts of the FRB under the baseline and stress scenarios. ### Public disclosure constraints on DFAST results - Detailed information on capital, RWAs and income statement items were not publicly available. - Published items included: - Only the minimum and the ending capital ratios over the stress testing horizon of nine quarter. - RWAs as of the end of the stress testing horizon. - Income statement items as a cumulative value of: pre-provision revenue, other revenue, provisions, realized losses/gains on securities (AFS/HTM), trading and counterparty losses, other losses/gains, net income before taxes and other comprehensive income. - AOCI included in capital was published as of the end of the stress testing horizon. - Loan losses (cumulative over the nine quarter period) and loan losses structure by type of loan. - All results were presented both in aggregate for the 31 BHCs and for individual BHCs for severely adverse and adverse scenarios. ### DFAST supervisory results and IMF comparison (severely adverse scenario) - Aggregate resilience: supervisory DFAST suggests that, in the aggregate, BHCs are resilient to shocks for the severely adverse scenario; all BHCs stay above the regulatory minima. - IMF hurdle rates: Even if the IMF hurdle rates were used, no BHC failed the stress test. - CET1 ratio projections over the nine-quarter stress testing horizon (supervisory DFAST): - System-wide CET1 would fall to 7.6 percent (its minimum) and to 7.8 by the end of 2016. - IMF estimate of the system-wide CET1 in 2014Q3 was 12.4 percent. - The CET1 ratio at its minimum would fall by 480 basis points compared to 2014Q3. - Drivers of CET1 decline: increase in RWAs (mainly reflecting implementation of standardized approach in 2015 and the assumption on loan supply) and net income losses projected to be -$222 billion. ### Loss drivers and income performance (severely adverse) - Projected losses (major components): - Loan losses: $340 billion. - Trading and counterparty credit losses from a global market shock: $103 billion. - Mortgage and consumer loans: - Projected losses on mortgage and consumer loans represent 56 percent of projected loan losses, driven by higher unemployment rate and lower house prices. - Definitions in footnotes: - Mortgage loans: first-lien mortgages, domestic and junior liens and HELOCs, domestic mortgage loans. - Consumer loans: credit cards and other consumer loans. - Largest losses by loan type: credit card losses of $83 billion. - Nine-quarter cumulative loss rate: 6.1 percent (with significant differences across BHCs), described as more severe than any recession since the 1930ties. - Trading and counterparty losses by BHCs: trading losses at six BHCs and counterparty losses at eight BHCs ranged between $1bn and $24 billion across the eight BHCs. - Pre-provision revenue: low level mainly reflected low projected net interest income and non-interest income, consistent with low interest rates and flattening of the yield curve early in the stress horizon and falling asset prices, rising equity market volatility and falling economic activity. ### Adverse scenario outcomes - Adverse scenario effect: BHCs would report moderate declines in capital ratios; scenario simulates a mild recession but with a sharp increase in short term rates that affect BHCs’ funding costs. - Projected capital ratios under the adverse scenario are smaller than those under the severely adverse scenario. - Main difference versus severely adverse: higher pre-provision revenue driven by higher net interest income due to higher interest rates. - Public results did not indicate any impact of higher interest rates on loan delinquency. - AOCI is three times larger than in the severely adverse scenario due to higher interest rates. ### Aggregate comparisons: CET1 and RWAs (DFAST vs IMF) - Aggregate CET1 projections in the DFAST and IMF top-down approach are similar (reference to Figure 7). - Minimum CET1 ratio in the benchmark IMF stress test is higher than in the DFAST, mainly reflecting less conservative loan dynamics and lower dividend distributions for firms constrained by the hurdle rate. - When DFAST assumptions on loan supply, dividend distribution and operational RWAs were introduced in the IMF stress test, the CET1 ratio in the benchmark IMF case came down close to the DFAST CET1 ratio. - RWAs: - Total RWAs in DFAST were 6 percent higher than in the IMF benchmark case (reflecting difference in credit dynamics that outweigh the addition of operational RWAs to the IMF model). - Total RWAs in DFAST were also 4 percent higher than in the IMF stress test with DFAST assumptions. - If the DFAST estimates of the 2015 standardized RWAs increase for individual BHCs were applied in the IMF stress test the total RWAs would be very similar. - Aggregated capital level in DFAST was higher than in IMF’s stress test. - Net income losses were very similar across the two stress tests: projected provisions were almost the same and the sum of pre-provision revenue and trading, counterparty and other losses was very similar to pre-provision revenue in the IMF stress test. - Possible contributors to higher aggregate capital in DFAST (not fully decomposable from published results): (i) taxes; (ii) deductions from CET1; (iii) dividends; (iv) extra-ordinary items; (v) change in valuation of allowances. - Deductions and dividends were presumably the same in both stress tests in the case where IMF stress test took DFAST assumptions, but details were not part of the published results. ### Bank-specific divergence across exercises - About one third of BHCs in the DFAST have either higher or lower CET1 ratio by 2.5 percentage points compared to IMF estimates. - For a few large BHCs IMF estimates of CET1 ratios were higher due to high trading and counterparty default losses in the DFAST for these BHCs. - For other BHCs, IMF estimates of net income losses were mainly higher, including for the two BHCs that would fall below the capital hurdle rates. - Differences may be due to the IMF’s modeling approach and lower granularity of data in the IMF stress testing framework. ### Company-run DFAST for 31 BHCs — scope and timing - Three bottom-up stress tests were reviewed by agencies: FRB (DFA company-run stress test), OCC, and FDIC. - Cut-off date of these tests: September 2014. - Coverage: - DFA company-run stress test covered BHCs with consolidated assets of $10 billion or more, accounting for about 90 percent of total BHC assets. - OCC’s tests covered national banks and federal savings associations with total consolidated assets over $10 billion. - FDIC’s tests covered FDIC-insured state banks and savings associations with total consolidated assets of more than $10 billion (not members of the Federal Reserve System). - Instructions and scenarios issued on October 23, 2014. - Company-run stress testing results were reported to primary supervisors on January 6, 2015 (for companies with assets > $50 billion) and on March 31, 2015 (for companies with assets between $10 billion and $50 billion). - For comparability with IMF top-down and supervisory DFAST, the note focuses on the DFA bottom-up stress test results for largest BHCs (total assets of $50 billion and more). ### Capital definitions and regulatory transition in company-run stress tests - Capital definitions corresponded to local regulation: - Basel III (subject to phase-in) for advanced approaches BHCs and non-advanced approaches BHCs (from January 1, 2015). - Basel I for non-advanced approach BHCs for the first quarter of the stress test horizon (last quarter of 2014). - Companies were required to assume consistency with the Basel III transition schedule. - Stress tests incorporated transition arrangements and minimum capital requirements from the revised regulatory capital framework implementing Basel III capital reforms from January 2014 and Basel I capital standards for non-advanced banking organizations applied for 2014 only. - Starting in 2015, the revised capital framework introduced a new standardized approach for risk weighting assets, replacing calculation of risk weights using the general risk-based capital approach. ### Disclosure and comparability limitations for company-run results - Company-run results were disclosed by BHCs in supervisory DFAST format. - Detailed information on capital, RWAs and income statement items were not publicly available. - Six BHCs did not publish their RWAs, precluding aggregate analysis for the 31 BHCs. ### Company-run DFAST results and comparison with supervisory DFAST and IMF stress test - On average, company-run results were more optimistic than supervisory DFAST or IMF stress test. - Aggregate resilience: tests suggest BHCs are resilient to shocks for the severely adverse scenario; all BHCs stay above regulatory minima even if IMF hurdle rates were used. - Un-weighted system-wide average CET1 over nine-quarter stress horizon in company-run stress test: - Fell from 12.9 percent to 9.4 percent (its minimum). - Average CET1 ratio comparisons: - Company-run average CET1 is 90bps higher than in supervisory DFAST. - Company-run average CET1 is 175bps higher than in the IMF stress test. - Drivers of higher CET1 in company-run: - Lower increase in RWAs (for BHCs that reported RWAs). - Lower net income losses projected to be -$190 billion (compared to -$222 billion in supervisory DFAST and -$224 billion in IMF stress test). ### Differences in projected losses and loan-loss structure (company-run) - Company-run projected losses: - Loan losses: $250 billion. - Trading and counterparty losses: $108 billion (almost the same as in supervisory DFAST). - Projected loan losses were much lower in company-run than in FRB supervisory or IMF estimates. - Structure of loan losses differences: - Company-run: consumer loans represent 45 percent of projected loan losses. - Supervisory DFAST and IMF: consumer loans represent 35 percent of projected loan losses. - Correlations and BHC-level divergences: - Correlation between CET1 ratios in the company-run stress test and in the supervisory stress test is 0.80. - Correlation between CET1 ratios in the company-run stress test and in the IMF stress test is 0.47. - Correlation between CET1 ratios in the supervisory stress test and in the IMF stress test is 0.49. - In company-run vs supervisory DFAST, five BHCs have higher CET1 ratio by 2.5 percentage points due to lower losses and RWAs projected by BHCs. ### IMF staff liquidity risk analysis for BHCs (as of 2014Q3) - Analysis covered the same BHCs as the solvency stress test and was done on a bank-by-bank basis. - Due to data constraints, the LCR or the NSFR could not be calculated. - The IMF liquidity metric was defined and calculated based on publicly available FR Y-9C report data (Schedules HC, HC-B, HC-D, HC-E, HC-L and HC-M). - The liquidity metric measured whether BHCs have adequate levels of liquid assets convertible into cash to meet liquidity needs, defined as the ratio between stocks of liquid assets and total cash outflow. - Items included in numerator and denominator were informed by LCR definitions used in BCBS (2013), but there was no attempt to replicate the LCR calculation based on publicly available data. - Haircuts were taken from the LCR and run-off rates were calibrated based on the 2008/2009 episode. - Two stress horizons for withdrawal of funding were assumed: 1 quarter and 3 quarters. - In sensitivity analysis run-off rates were calibrated based on the LCR. ### Liquid assets and haircut structure (Table excerpts) - Haircut and asset categorizations included: - Level 1 assets: - Cash and balances due from depository institutions: haircut 0 - U.S. Treasury securities: haircut 0 - Level 2A assets: - U.S. government agency obligations: haircut 15 - MBSs (guaranteed by GNMA, issued by FNMA, FHLMC, issued or guaranteed by U.S. government agencies or sponsored agencies): haircut 15 - Level 2B assets: - Other domestic debt securities, other foreign debt securities, investments in mutual funds and other equity securities: haircut 50 ### Outflow items and run-off rates (selected entries from Table 14) - Historical run-off rates (1q), historical run-off rates (3q), LCR run-off rates (examples): - Time deposits of less than $100,000: HIST. RUN-OFF RATE, 1q = 0; HIST. RUN-OFF RATE, 3q = 153 - Interest bearing demand deposits, NOW, ATS and other transaction accounts: HIST. RUN-OFF RATE, 1q = 30; LCR RUN-OFF RATE = 3 - Time deposits of $100,000 or more: HIST. RUN-OFF RATE, 1q = 82; LCR RUN-OFF RATE = 15 - Money market deposit accounts and other savings accounts: HIST. RUN-OFF RATE, 1q = 0; HIST. RUN-OFF RATE, 3q = 0; LCR RUN-OFF RATE = 10 - Noninterest bearing balances: HIST. RUN-OFF RATE, 1q = 128; LCR RUN-OFF RATE = 10 - Deposits in foreign offices: HIST. RUN-OFF RATE, 1q = 41; LCR RUN-OFF RATE = 55 - Unsecured borrowing: HIST. RUN-OFF RATE, 1q = 41; HIST. RUN-OFF RATE, 3q = 110 - Subordinated notes and debentures: HIST. RUN-OFF RATE, 1q = 44; HIST. RUN-OFF RATE, 3q = 100 - Securities sold under agreement to repurchase: LCR RUN-OFF RATE = 15 - Secured borrowing: HIST. RUN-OFF RATE, 1q = 13; HIST. RUN-OFF RATE, 3q = 22; LCR RUN-OFF RATE = 50 - Liabilities for short position: HIST. RUN-OFF RATE, 1q = 13; LCR RUN-OFF RATE = 5 - Currently undrawn committed credit and liquidity facilities (examples): - Revolving, open-end loans secured by 1-4 family residential properties: HIST. RUN-OFF RATE, 1q = 5; LCR RUN-OFF RATE = 35 - Unused consumer credit card lines: HIST. RUN-OFF RATE, 1q = 11; LCR RUN-OFF RATE = 25 - Commitments to fund CRE, construction and land development loans secured by RE: HIST. RUN-OFF RATE, 1q = 6; LCR RUN-OFF RATE = 14 - Securities underwriting: HIST. RUN-OFF RATE, 1q = 44; LCR RUN-OFF RATE = 70 - Commercial and industrial loans: HIST. RUN-OFF RATE, 1q = 25; LCR RUN-OFF RATE = 10 - Loans to financial institutions: HIST. RUN-OFF RATE, 1q = 25; LCR RUN-OFF RATE = 40 - Other contingent funding liabilities (examples): - Financial standby letters of credit: HIST. RUN-OFF RATE, 1q = 24; LCR RUN-OFF RATE = 10 - Performance standby letter of credit: HIST. RUN-OFF RATE, 1q = 21; LCR RUN-OFF RATE = 8 - Commercial and similar letters of credit: HIST. RUN-OFF RATE, 1q = 52; LCR RUN-OFF RATE = 61 *Source: IMF Staff* ### 102. The results of the analysis give some comfort that the system is able to meet liquidity ### _cr15173 - 102. The results of the analysis give some comfort that the system is able to meet liquidity ### Liquidity analysis: key findings - The analysis indicates that the system is able to meet liquidity requirements overall, but there are pockets of vulnerability. - Most, but not all, BHCs have enough liquid assets to meet a liquidity shock similar to the 2008/2009 event (Figure 8). - Several BHCs would face liquidity pressures due to deposit outflows in the short run and large unused commitments over a longer stress horizon. - Under a much larger shock characterized by the LCR run-off rates, liquid assets for many BHCs would not be sufficient to meet liquidity needs due to large withdrawal of wholesale funding. - Wholesale funding is important because the run-off rate on wholesale funding in the LCR approximation is much larger than what happened in 2008/2009. - Footnote: The Fed’s estimates indicate that the majority of the largest U.S. banks would meet the LCR. ### Liquidity metric, scope and data limitations (caveats) - The liquidity metric used: - Includes both unencumbered and encumbered liquid assets, which inflates liquid assets relative to an ideal measure that would include only unencumbered liquid assets. - Does not include pledged assets to BHCs, which deflates the liquid assets measure. - Did not consider inflows as part of the liquidity analysis, which would make the liquidity metric smaller than it would otherwise be. - Did not consider outflows on derivatives contracts. - Data and sensitivity limitations: - Due to unavailability of granular data, the actual run-off rates used in the sensitivity analysis had far smaller variation than those in the LCR. - Current regulatory reports lack important elements to run an accurate liquidity stress test, such as liability tenor information, inflows from maturing transactions, and relevant contractual terms embedded in derivatives contracts. - Footnote: Publicly available data do not allow differentiating between encumbered and unencumbered assets. ### Figure 8 (conceptual summary) - Figure 8 shows the distribution (whisker boxes) of the liquidity metric (Liquid assets/Outflow) for 31 BHCs, including median, first and third quartiles, un-weighted average (diamonds), minimum and maximum. - The figure compares: - Historical run-off rates (1 quarter and 3 quarters), - An approximation of the LCR run-off rates, - Composition of liquid assets and outflows by instrument (e.g., Cash, U.S. Treasury securities, GSE MBSs, Unsec. wholesale funding, Sec. wholesale funding, Undrawn commitments, Other Deposits). - The sample size associated with Figure 8: 31 BHCs. *Source: IMF Staff calculations.* --- ### IMF staff’s solvency tests for insurance ### A. Scope of the test - Sample: - 44 insurance groups included in the stress-testing sample. - By business line: 22 predominantly life insurance, 15 property & casualty (P&C), five health insurance, and two in credit and mortgage insurance. - All groups in the sample are publicly listed. - The sample includes the three groups designated both as global systemically important insurers by the Financial Stability Board (FSB) and as systemically important by the Financial Stability Oversight Council (FSOC). - The sample represents 40 percent of the domestic insurance sector in terms of gross written premiums. - NAIC stress test: - A separate top-down stress test performed by the NAIC covered all U.S. life and P&C insurers filing with the NAIC and informed IMF sensitivity analyses. - Data sources: - Publicly available consolidated regulatory returns from SNL Financial and Bloomberg. - Further sector-wide data provided by the NAIC for greater granularity. - NAIC legal entity data (including NAIC risk-based capital data) were available to IMF in aggregate format only. ### B. Scenario (market risk stress specifications) - The insurance top-down stress test was built on the DFA stress test (DFAST) specifications focusing on investment assets and market risk. - Market risk stresses included shocks to: - Bond holdings (sovereigns, municipals, and corporates), - Securitizations, - Equity, - Property and other investments (such as hedge funds and private equity). - Swap rates assumed to fall in a range of minus 21 to minus 143 basi–s points for maturities of one year and 30 years, respectively. - Broad characterization: market distress similar to the height of the 2008–09 financial crisis; all stresses were assumed to occur instantaneously. - Adaptations due to data limits: - Equity shock was applied as an overall weighted shock (not per-country) based on the geographical breakdown of equity exposures. - Non-U.S. sovereign bond shock derived and split into NAIC rating categories. - Property (commercial real estate) shock assumed to compress 12 quarters of projected DFAST CRE price developments (2014-Q4 to 2017-Q4) into a single period. - Selected market shock figures and market value changes (as presented in the source): - Market shocks (examples as presented): RMBS-51,3%; CMBS-50,3%; Other Securitizations-28,7%; Mortgage loans-12,2%; Equity interest-28,9%; Property (CRE)-28,3%; Cash 0,0%; Other investments-28,9%. - Bond spread and swap/discount curve changes shown in original table (IMF Staff assumptions based on DFAST). ### C. Valuation approach and capital standard - Accounting context: - U.S. insurers use statutory accounting based on U.S. GAAP, which differs from a “fully market-consistent” valuation. - P&C insurer liabilities are generally not discounted under statutory accounting (adds conservatism). - Life insurance liabilities are discounted either at the rate set when the policy was sold or with a discount rate based on the expected return of assets associated with the insurance liabilities. - NAIC calculated that the discount rate on all life insurance policies averaged approximately 4 percent in 2013. - The average net portfolio yield for life insurance companies was 5.11 percent in 2013. - Asset valuation under statutory accounting: - Amortized cost is the predominant regime for fixed income assets. - For life insurers, fixed income assets in NAIC credit quality buckets 1 to 5 are valued at amortized cost; for P&C insurers, buckets 1 and 2 are valued at amortized cost. - For remaining credit quality buckets (6 in life and 3 to 6 in P&C), the lower value of amortized cost and market value is used. - Under statutory accounting, unrealized gains or losses on amortized-cost assets are not recognized; impairment is applied only when fair value loss is deemed other than temporary, and once impaired a bond cannot be written back up to original fair value after recovery. - IMF top-down stress test hurdle: - In absence of a group capital requirement for insurers, the IMF’s top-down stress test used the complete extinction of shareholder equity as the hurdle rate—groups with negative shareholder equity after stress were deemed to fail the stress test. - Contextual mitigants: surrender penalties, loss of insurability, federal income tax treatment of life insurance, and state insurance regulator powers can limit policyholder runs and ensure relatively ordered run-offs. ### D. Overview of insurance companies’ soundness (aggregated sample-level facts) - Aggregate balance sheet (2009–2013): - Aggregated balance sheet of the 44 insurance groups increased by 25 percent to $4.5 trillion. - Aggregated assets by line: life insurers $3.1 trillion; non-life insurers $1.1 trillion; health insurers $0.3 trillion. - Capital ratios (aggregate): - Ratio of shareholder equity to balance sheet assets in 2014: life insurers 8.3 percent; non-life insurers 22.9 percent; health insurers 33.1 (as stated in source). - Asset composition and duration: - Non-life insurers hold higher amounts of shares and shorter bond portfolio durations than life insurers. - Health insurers have the most conservative asset allocation with relatively high allocations in sovereign bonds. - Median durations (2014 sample medians): life insurers 7.6 years (up from 7.2 years in 2009); non-life companies 5.4 years; health insurers 5.2 years. - Median share of non-investment grade bonds: life insurers 5.4 percent; non-life insurers 4.1 percent; health insurers 4.4 percent. - Trend: share of non-investment grade bonds among life insurers decreased since 2009; increased for non-life insurers (from a lower base). - Revenues and profitability: - Revenue shares in the sample: health insurers 42 percent (mainly recurring premiums); life insurers 32 percent; non-life insurers 26 percent. - Revenue growth 2009–2014: life insurers increased revenues by 36 percent; non-life insurers by 11 percent; health insurers by 46 percent. - Net income 2014: nearly all insurers reported positive net income. - Median returns on equity: health sector 14.0 percent; non-life insurers 11.7 percent; life insurers 8.2 percent. - For P&C insurers, the combined ratio median in 2014 was 94 percent (post-super storm “Sandy” 2011 experience and stabilization thereafter). Over the last five years, changes of the combined ratio were predominantly driven by the loss ratio; operating expenses were stable on average around 32 percent. - Health sector specifics: - Affordable Care Act (ACA) implementation resulted in rising premium income for health insurers and outperformance of stock prices relative to the S&P 500 since March 2010. - Ongoing risks: administrative risks regarding regulated prices, legal challenges to subsidies, unknown medium-term policyholder behavior (notably lapse rates), and uncertain future claims due to shifts in the risk profile of policyholders. ### E. Modeling assumptions and limitations - Fully market-consistent valuation assumption: - Assuming a fully market-consistent valuation impact of the shock, the value of investment assets and shareholder equity declines substantially. - Credit spread increases are multiplied by the duration of the respective asset class (derived from maturity buckets provided in statutory reporting), further lowering investment asset values. - Separate accounts: - Separate accounts were excluded from the stress test because investment losses are generally passed on to policyholders. - Separate accounts are offered by 17 companies in the sample, with a median share of total reserves of 53 percent among those companies offering separate accounts. - No breakdown of investments or detail on guarantees in separate accounts is available in consolidated public filings; it was assumed asset allocation of separate accounts matches group-wide allocation and no economic loss remains with the insurer. - This simplifying assumption may underestimate stress effects where guarantees on separate accounts exist. - Liability duration: - No detailed information on liability duration was available. - Assumed duration of liabilities exceeds asset duration by two years for life insurers and by one year for non-life and health insurers (based on market participant suggestions). - Hedging: - The stress test did not account for any mitigating hedging effects (interest rate hedges via swaps/swaptions; equity hedges via options/futures) because hedging strategies vary substantially across companies; the stress test likely gives a maximum impact. - Sensitivity analyses: - IMF team built sensitivity tests on NAIC approaches for (a) major natural catastrophes, (b) a pandemic, and (c) a prolonged period of low interest rates. - Results of these sensitivity analyses were not added to the main stress scenario outcomes, although concurrence of the scenarios (e.g., stress plus catastrophe) could be assumed. *Italic: Source: IMF Staff calculations and IMF FSAP team analysis as presented in the supplied content.* ### 123. The IMF team has specified three types of natural catastrophes, for which the NAIC has ### _cr15173 - 123. The IMF team has specified three types of natural catastrophes, for which the NAIC has ### Catastrophe and pandemic scenarios specified by the IMF/NAIC - Three P&C catastrophe events specified and modeled as independent: - A Florida hurricane, similar to but worse than hurricane Andrew in 1992, causing an industry-wide loss of at least $40 billion. - A California earthquake, similar to but worse than the Northridge earthquake in 1994, with magnitude of at least 7.2 and insured loss greater than $35 bn. - A series of three major tornados in the U.S. Midwest, each causing an insured loss of at least $4 billion. - Life-sector pandemic shock: - Calibrated as a pandemic with 1.5 additional deaths per 1,000, considered a 1-in-200 year event (Swiss Re 2007). - Note: this increased mortality rate is below 1918–19 influenza levels (more than 5 per 1,000 in the United States), and results do not include potential macroeconomic implications. ### Top-down stress test results (market-consistent accounting) - Overall impact on shareholder equity: - Total losses in shareholder equity amount to $267 billion, or 45 percent of pre-stress equity. - By sector: - Life insurers contribute $187 billion. - P&C insurers contribute $72 billion. - Health insurers contribute $6 billion. - Credit insurers contribute $2 billion. - Out of the total loss, $104 billion is attributed to the distressed companies. - Life sector outcomes: - Wide dispersion; 11 out of 22 groups would report negative shareholder equity after stress under a fully market-consistent accounting regime. - Distressed firms: - Aggregated balance sheet assets of the 11 distressed companies amount to around $490 billion (11 percent of the full sample). - Their pre-stress shareholder equity declines from $55 billion to a negative $48 billion under stress. ### Main drivers and contributions to losses - Largest single contribution: - Corporate bond portfolio credit spread increase (also applied to hybrids) accounts for 62 percent of the overall loss for the full sample. - Other notable contributions (each with 7 percent): - Shock to sovereign bonds and GSEs. - Shock to the securitization portfolio. - Shock to share prices. - Sectoral sensitivities: - P&C insurers more affected by the equity market shock due to larger holdings in shares. - Health insurers would record larger losses stemming from their sovereign bond portfolio. - For credit insurers, the full impact of higher credit spreads and higher default rates may be understated because liabilities would also be affected. ### Natural catastrophe and pandemic impact estimates (as presented) - Summary from text (Table 16 context): - Florida hurricane modeled as a 1-in-250 year event could cause insured losses of around $80 billion (net of reinsurance). - California earthquake with same expected occurrence frequency could result in a loss of $34 billion (net of reinsurance), given residential exposure through the California Earthquake Authority. - Series of three severe tornados in the Midwest shows substantially contained effects, below the other two scenarios. - Pandemic (1.5 additional deaths per 1,000; 1-in-200 year) could cost the U.S. life insurance industry between $20 and $25 billion (estimated on a gross basis, not accounting for reinsurance). - Table excerpt (preserved verbatim from source): - net claims in USD billion 1-in-100 years 1-in-200 years 1-in-250 years - Florida hurricane45 - 5075 - 83 - California earthquake2234 - Pandemic20 - 25 ### Low interest rate ("low for long") risk to life insurers - Historical and projected spreads: - For 2006–2013, industry-wide spread between net portfolio yield and guaranteed credited rate declined by 57 basis points. - NAIC analysis horizon: 2014 until 2018; investment yield assumed to decline linearly based on trend observed since 2006. - Profitability and solvency thresholds: - Net investment yields would need to remain above 4 percent to ensure profitability. - Currently issued annuities contain an interest rate guarantee of 1 percent. - Potential impacts: - Continued low rates could erode profitability and, if negative spreads arise, could lead to widespread losses and a weaker capital position. - Modeling assumptions are restrictive (e.g., asset allocation) and results require intense monitoring. ### Asset allocation, duration, and credit quality changes (Table 17) - Duration and credit quality changes (preserved numeric values): - Life: - Duration 2009: 7.2 - Duration 2014: 7.6 - Share of non-IG in bond portfolio 2009 (in percent): 7.5 - Share of non-IG in bond portfolio 2013 (in percent): 5.4 - Share of BBB in IG bond portfolio 2009 (in percent): 33.9 - Share of BBB in IG bond portfolio 2009 (in percent): 37.6 - Non-life: - Duration 2009: 5.1 - Duration 2014: 5.4 - Share of non-IG in bond portfolio 2009 (in percent): 1.9 - Share of non-IG in bond portfolio 2013 (in percent): 4.1 - Share of BBB in IG bond portfolio 2009 (in percent): 8.7 - Share of BBB in IG bond portfolio 2009 (in percent): 15.0 - Health: - Duration 2009: 4.2 - Duration 2014: 5.2 - Share of non-IG in bond portfolio 2009 (in percent): 1.0 - Share of non-IG in bond portfolio 2013 (in percent): 4.4 - Share of BBB in IG bond portfolio 2009 (in percent): 10.1 - Share of BBB in IG bond portfolio 2009 (in percent): 14.9 - Observed trends: - Insurers increased portfolio durations since 2009. - Life insurers decreased share of bonds below investment grade (from 7.5 to 5.4 percent). - Non-life and health insurers increased non-IG share (from low bases). - Within IG, relative share of BBB-rated assets increased. - Some movement into alternative investments (hedge funds, private equity), concentrated among larger insurers. ### Statutory accounting stress testing and differences (Box 2 / IMF additional exercise) - IMF statutory accounting exercise (end-2014 data) modeling steps included: - Mark-to-market impairment for equity, corporate bonds below investment grade, and other investment assets (based on DFAST severely adverse scenario). - Default losses in corporate bond and securitization portfolios (default rates based on mid-2008 to mid-2009 observations, assumed 50 percent loss given default). - Catastrophic event (P&C) and pandemic (life) resulting in net cash outflows of $25 billion in each sector, distributed by market share. - Forced liquidation of assets to match outflows, with sales assumed in order: U.S. treasury bonds, municipals, and GSE issues first. - Aggregated statutory capital impact: - Reduction in statutory capital for life and P&C companies under this approach: $111 billion, representing 22 percent of pre-stress statutory capital. - Median loss: life sample 26 percent; P&C insurers 23 percent. - One life insurer becomes distressed (statutory capital negative); one other life insurer experiences reduction >75 percent. - Attribution of statutory capital changes: - Impairment of investments: 69 percent of change. - Catastrophic/pandemic event: 18 percent. - Defaults in investment grade corporate bond and securitizations: 13 percent. - Realized losses from forced sell-off: 1 percent. - Comparison with NAIC stress test: - Broad alignment with NAIC statutory-accounting results, but differences exist due to scenario design (e.g., NAIC stress test does not include pandemic impact in life insurers’ overall results). ### Insurer mitigation options and behavioral responses - Possible mitigating actions insurers can take: - Reduce discretionary benefits in profit-sharing products to pass on investment losses to policyholders. - De-risk investments under a risk-based solvency regime to lower capital requirements and raise solvency ratios. - Cede risks to reinsurers or alternative risk transfers. - Actively manage dividend policies (subsidiary-to-top and top-to-shareholder flows). - Limits of modeling: - Wide range of management actions and limited publicly available data prevent a single general modeling result. ### Liquidity risks in open-ended mutual funds (IMF staff analysis) - Redemption risk characteristics: - Liabilities are liquid (regulatory obligation to meet investor redemption demand in-cash within 7 days) while investments may be less liquid. - Funds may lack balance-sheet liquidity capacity or robust-to-severe stress back-up lines, potentially forcing asset sales at steep discounts. - Fixed-income expansion and liquidity concerns: - Rapid expansion into corporate bonds, emerging market debt, bank loans, and municipal bonds raises concerns that investor runs could exacerbate asset market stress. - These markets are less liquid than equities, U.S. Treasury, and GSE securities. - Notable decrease in trading liquidity in corporate bonds since the crisis. *Source: _cr15173 - 123. The IMF team has specified three types of natural catastrophes, for which the NAIC has* ### 137. IMF staff has performed an analysis of liquidity in U.S. mutual funds, and important ### IMF staff analysis of liquidity in U.S. mutual funds ### Scope and objective of the analysis - Purpose: Measure whether markets could absorb severe redemption pressures forcing mutual funds to liquidate positions; exploratory analysis due to nascent analytical basis and incomplete public data. - Coverage: Close to 9,000 mutual funds representing around 80 percent of the industry. - Data cut-off date: 2014Q3. - Data source: Granular mutual fund data from the Center for Research in Security Prices (CRSP). ### Top-down liquidity risk methodology - Stress type: One time, tail event redemption shock. - Definition of shock: First percentile of the distribution of net flow rates by fund style. Net flow rates defined monthly as a simple average of net flow rates of all mutual funds of the same style over the period 1998-2014Q3. - Market-level focus: Mortgage, corporate, municipal, government bonds; both index funds and hybrid funds included. - Proxy for trading liquidity: Dealer inventory in specific asset markets used as an indicator of market makers’ willingness/ability to intermediate and of general market demand for an asset class. ### Assumptions on asset sales to meet redemptions - Two alternative approaches for how funds sell assets when hit by the shock: - Approach 1 (“pro rata”): Pro-rata selling to keep portfolio weights intact (natural for index funds). - Approach 2 (“waterfall”): Rank-order assets by liquidity (as captured by LCR haircut hierarchy) and sell in descending order of liquidity. Assumed ordering and proportion sold (from most to least liquid): - 1. Cash — 100% - 2. Government securities — 100% - 3. MBS — 100% (assuming most are GSE-backed MBSs) - 4. Equity — 20% - 5. Corporate bonds — 20% - 6. Munis — 100% - 7. Equity — 40% - 8. Corporate bonds — 40% - Aggregation: Realized asset sales due to the tail event were summed across all funds and for each asset market, then compared to dealers’ inventory. ### Key quantitative findings - Municipal bonds and corporate bonds may face significant stress under the tail event redemption shock. - Municipal bonds potentially sold to meet redemptions could be: - Three times larger than dealers’ inventory (in the assumed ordering of assets sold). - Four times larger than dealers’ inventory (in the pro-rata asset selling case). - Corporate bonds potentially sold by mutual funds under severe stress could be up to seven times larger than dealers’ inventory. - Figure referenced: Sold assets under stress and dealers' inventories, million USD (shows Sold assets- assumed ordering; Dealers' inventory; Sold assets- pro-rata). ### Security-level analysis and data constraints - A separate security-level exercise combined CRSP portfolio security information with CUSIP Global Service’s CUSIP Master File and the Mergent Fixed Income Securities Database to aggregate holdings into homogenous asset classes. - Data limitations observed: - CUSIP master file contained information for 56 percent of the securities with CUSIPs in the CRSP holdings database; of these records, about 90 percent were for equities. - Mergent database contained information for only 8 percent of the securities with CUSIPs in the CRSP holdings database. - Evidence of significant domestic equity reporting bias and limited coverage in issuer information made aggregated asset classes unrepresentative of the broader mutual fund holdings universe. - Result: Liquidity assessments using these aggregated security-level asset classes were not pursued further. ### Caveats and limitations highlighted by IMF staff - First, not all mutual funds were included; a larger sample would increase assets sold under stress and underscore vulnerabilities. - Second, data granularity was limited for both funds’ asset structures and dealers’ inventory structures, preventing discrimination (e.g., domestic vs. foreign corporate bonds were bundled together). - Third, the top-down, style-based approach aggregated across market segments (e.g., investment grade and high yield corporate bonds were not separated), which may affect results. - Fourth, the analytical basis for measuring mutual fund liquidity is still nascent; absent an asset pricing model, dealer inventory was used as a proxy for market liquidity, which has pros and cons. ### “Ideal” liquidity analysis (described by IMF staff) - Required inputs and steps for an ideal analysis at individual fund level: - Determine tailored liquidity waterfalls based upon investment mandates. - Apply extreme redemption shocks to assess immediate cash demand. - Use the liquidity waterfall to estimate which, and how much of, each security should be sold to meet redemptions. - Use cumulative information on security bid-ask spreads and trading volumes to estimate the sales price of each security sold. - Aggregate values of securities sold across funds into granular asset classes and compare to obtainable values under ordinary market conditions to identify asset classes prone to fire-sale dynamics. - Data needed: Detailed supervisory data on fund holdings, investment mandates, trading volumes, and market-level liquidity information. ### Policy recommendations and next steps - Authorities should step up work to assess susceptibility of markets to extreme mutual fund redemptions. - Authorities should further clarify guidance on liquidity risk analysis performed by the industry. - Authorities should mobilize resources necessary to regularly conduct liquidity risk analyses as part of their overall approach to mutual fund industry oversight. ### Transition to market-price based stress tests - IMF staff complement the top-down analysis with market-price based analysis and stress tests to capture information embodied in market prices and higher-frequency correlations. - Caveats: Market-price based findings are sensitive to methodological choices, imperfect proxies, sample period selection, heterogeneity across definitions, and potential endogeneity; findings are informative approximations rather than precise estimates. *Source: IMF staff analysis in IMF technical note (CR15173), chapter on mutual fund liquidity and market-price based stress tests.* ### 150. To limit model risk, this technical note adopts a multi-model approach to identifying ### _cr15173 - 150. To limit model risk, this technical note adopts a multi-model approach to identifying ### Multi-model approach and Systemic Risk Dashboard - Purpose: adopt a multi-faceted “systemic risk dashboard” to identify diverse dimensions of threats to financial stability and to limit “model risk” (excessive reliance on a single modeling framework). - The dashboard employs a range of measures to inform risk-based assessment and guard against model risk. ### Market-based measures and SRISK - SRISK is an estimate of the capital that a financial firm would need to raise if a severe financial crisis were to occur. - Findings: - SRISK capital shortfalls peaked at approximately one trillion dollars in early 2009. - SRISK has now fallen to 300 billion dollars. - This level constitutes approximately 2 percent of GDP. - Caveat: SRISK exclusively measures systemic risk in the banking sector; nonbank risks are covered in subsequent analysis. ### Early warning indicators, equity, and housing - Early warning indicators: - Financial cycles and credit-to-GDP gap measures signal that the United States banking system is relatively healthy from a cyclical perspective. - Equity price indicators: - Percentage deviations of observed equity prices from APT-implied theoretical prices and the Equity Composite Z-score indicate equity prices may be approaching unsustainably high levels (indicators are either positive or trending positive). - Housing and credit measures: - Credit-based measures are still well below their pre-crisis levels. - Growth in housing prices has fully regained its pre-crisis momentum and needs careful monitoring. ### Credit risk networks and contagion - Credit risk migration: - Susceptibility of institutions to credit risk migration has declined materially since 2007–09, with fewer dense, red connections in current-period networks versus 2007–09. - The role played by GSIBs in transfer of credit risk remains important. - Methodology note (network construction): - Networks created by applying bivariate Granger-Causality tests to time-series of 198 institutions’ first-differenced default probabilities. - Tests applied 21 times with combinations of three significance levels (1%, 5%, 10%) and seven lag orders (1 through 7). - Connection-color robustness: red = found in 95% or more of trials; dark gray = 70% or more; light gray = at least 50%. Thick black = membership in minimum spanning tree. ### Financial account net exposures (sector–instrument linkages) - Key inter-sector linkages and net exposures: - Asset managers, insurers, pensions, and households are sizable net claimants on corporate equities and corporate bonds issued by corporates and NBFIs. - Asset managers, insurers, pensions, households, and banks are notable net claimants on the government via treasury, agency, and municipal debt securities. - Households and pensions are the largest net claimants on fund shares; banks, insurers, and non-financial corporate also hold net claims on this sector. - Money market instruments act as an important conduit between households, asset managers, and banks. - Banks, corporates, households, and GSEs have net exposures to all other sectors and to abroad via deposits, loans, and mortgages. - Government and pensions are significantly exposed to households and corporates via pension entitlements. - Rest of the world exposures: - On a net basis, the rest of the world is exposed to the U.S. financial system primarily through net claims on money market instruments, corporate equities, and other direct investment. - Foreign net claims on the United States take the form of shares in investment funds, government securities, corporate bonds, loans, mortgages, and miscellaneous deposits. - OTC CDS exposures: - Publicly available locational data on OTC credit-default-swaps (CDS) show sizable outstanding exposures for both the rest of the world and the aggregate U.S. financial system. - Additional detail on the nature of these credit risk exposures cannot be obtained without access to confidential trade information. ### SyRIN (Systemic Risk Indicators) framework findings - Framework description: - SyRIN employs distress-based metrics; “distress” is defined as an event where the sector return index falls to a level that history would suggest should only be observed 1 percent of the time. - SyRIN models a multivariate joint probability density that accounts for financial system interconnectedness and uses a robust, non-parametric approach incorporating endogenous changes in sector co-dependence. - Asset shares and contributions: - In terms of asset size, banks account for less than 30 percent of the U.S. financial system; more than 70 percent is accounted for by nonbanks (shares broadly unchanged between 2010Q4 and 2014Q4). - The marginal contribution to systemic risk (MCSR) of the U.S. high yield and insurance sectors is disproportionately large relative to their share of financial system assets. - The pension sector’s ratio of MCSR to size is approximately 0.6, indicating the pension sector contributes less to systemic risk relative to its size and appears to diversify risk from the financial system. - Longer-term trends: ratios of MCSR relative to size for High Yield Bond Funds, IG Bond Funds and Hedge Funds have shown a significant increasing trend. ### Overall assessment and policy implications - Summary: - Systemic risk indicators suggest threats to financial stability have diminished since the peak of the crisis but remain worthy of continued close monitoring, especially with regard to nonbanks. - Financial stability indicators in the United States have broadly improved, but areas of concern remain, notably: - Equity price levels appear stretched. - Housing price growth has regained pre-crisis momentum and requires monitoring. - Banks remain vulnerable to distress originating in the high yield and insurance sectors. - Next analytical step noted in the technical note: - The subsequent section uses a hybrid market/balance-sheet framework known as Contingent Claims Analysis (CCA) to explore how financial risks may evolve under the IMF/DFAST stress scenario. *UNITED STATES — INTERNATIONAL MONETARY FUND.* ### 152. Contingent Claims Analysis (CCA) ### Contingent Claims Analysis (CCA) ### Overview and purpose - Uses equity prices and accounting information to measure the credit risk of institutions with publicly traded equity. - Provides forward-looking default probabilities that take into account leverage levels and market participants’ views on credit-quality. - Provides a standardized benchmark of credit risk (default probabilities) for cross-sector and cross-country comparisons. - Limitations: - Can only be applied to entities with either publicly-traded equity or very liquid CDS spreads. - Cannot capture liquidity or (financing) roll-over risk. - CCA risk measures are calculated at the firm-level and complement aggregate/systemic risk measures with micro-based financial risk information. ### Theoretical foundation - Default risk driven by level and uncertainty in asset values relative to promised debt payments ("default barrier"). - Equity modeled as an implicit call option on assets with exercise price equal to the default barrier (Merton, 1973). - Risky debt modeled as default-free debt minus an implicit put option (expected loss due to default). - Accounting information on debts and market information on equity prices calibrate CCA balance sheet risk indicators and estimate forward-looking probabilities of default. ### Distinction from SyRIN - CCA uses “default probabilities” derived from a structural model (Merton, 1973): chance that market-adjusted assets fall below a liability threshold. - SyRIN uses “distress probabilities”: chance that a sector’s market equity falls below the 1st percentile of its historical equity price distribution. - Sector coverage differs: SyRIN includes pensions and high yield bond funds; CCA includes foreign banking and insurance sectors. ### CCA Stress Test Data and sample - Historical default probability estimates acquired from CreditEdge. - CreditEdge production steps: i. For each institution, daily equity values, equity volatilities, and default barriers are calculated. ii. Inputs used to solve two structural CCA equations to estimate market-implied asset values and market-implied asset volatility. iii. Compute “Distance-to-default” indicators and map to empirically observed one-year default probabilities using Moody’s historical default database. - Investment-grade “safe zone” empirical boundary: expected one-year default probability of 0.5 percent or less. - Sample covered 210 institutions across eight sectors: - Asset Managers: 41 (10 billion USD plus market cap) - NBFIs: 13 (10 billion USD plus market cap) - Insurers: 44 (20 billion USD plus market cap) - Corporates: 32 (Includes the largest non-financial DJIA public companies, automakers that received government support, “new economy” technology companies with large and rapidly growing market capitalization) - Banks: 46 (20 billion USD plus market cap) - GSEs: 2 (Must have entered government conservatorship) - Foreign Banks: 20 (All foreign banks designated by the FSB as GSIBs) - Foreign Insurers: 12 (All foreign insurers designated by the FSB as GSIIs plus the largest non-U.S. domiciled global insurers) - Total: 210 - Note: Default probabilities for GSEs corresponding to the period following their conservatorship were omitted due to data quality constraints. ### Variables and data frequency - Daily default probability data from CreditEdge. - Daily, monthly, and quarterly macroeconomic data from Haver Analytics and Bloomberg. - Inter-connectivity data calculated by IMF staff on a rolling monthly basis (2004–2014). - All quarterly and monthly variables, including connectivity, temporally disaggregated to daily using the Chow-Lin Max-Log methodology. - Connectivity projection assumptions: i. The connectivity trend observed in 2008–2010 would repeat itself. ii. Then remain at its post-crisis historical average once this value has been reached. - All default probabilities used are median one-year expected default probabilities; sector and U.S. financial system medians computed from pooled observations (excluding GSEs for U.S. system). ### Modeling methodology (GAMLSS) - Employed General Adaptive Models of Location, Shape, and Scale (GAMLSS) to model median 1-year default probabilities. - GAMLSS advantages: allows wide variety of distributions and models first four moments as functions of exogenous conditions (addresses tail-risks, nonlinearities, deviations from normality). - Modeling iterative eight-step process: i. Fit approximately 100 different distributions to response variable. ii. Compare fits using Generalized Akaike Information Criteria (GAIC) and select best distribution. iii. Use distribution to run information-criteria based selection to identify independent variables with most linear explanatory power. iv. Transform variables not selected using additive terms and repeat selection. v. Experiment with additive terms to enhance fit. vi. Compare experimental models via GAIC and select best-scoring model. vii. Repeat steps (i)–(vi) for each distributional parameter (mean, variance, skewness, kurtosis). viii. Use diagnostics on residuals; if inadequate, repeat process with next-best GAIC distribution. - Over-fitting controls: - Stiff penalties on model complexity when computing GAIC (penalty value k=2.8). - Back-testing regime with quantile-based sampling to partition dataset into validation, testing, and training datasets. - Validation dataset used for initial distributional fits; training dataset for estimating regression coefficients and variable selection; testing dataset for out-of-sample performance assessment (compare global deviance and mean prediction error). ### Key model findings on drivers of U.S. credit risk - Median 1-year default probabilities for the U.S. financial system: - Expected to fall linearly with rises in long-term Treasury yields, housing and commercial real-estate prices, and with appreciation of the U.S. dollar against the Euro. - Expected to increase linearly with BBB-rated corporate bond yields and with sector median default probabilities (banks, insurers, asset managers, other NBFIs, foreign banks, foreign insurers). - Variables with greatest effects on financial system credit risk include changes in BBB corporate bond yields, long-term Treasury yields, and asset manager default probabilities. - Non-linear relationships identified: - U.S. credit risk slightly rises with euro-area consumer inflation. - U.S. credit risk falls with declines in real GDP growth and the VIX and Dow Jones Industrial indices. - Overall U.S. credit risk rises with increases in corporate credit risk, but only in a non-linear statistically significant manner. - Diagnostics and specification details summarized in referenced appendices (Appendix Table 5; Appendix Figure 8). ### Connectivity measure construction and behavior - Single connectivity variable used in each GAMLSS model: global clustering coefficient derived from institutional default probability time-series of five domestic sectors (banks, insurers, asset managers, other NBFIs, nonfinancial corporates). - Three-step calculation: - Perform Spearman Rank Correlation Tests to identify institutional default probabilities correlated at the .0001 percent significance level. - Construct adjacency matrix from test results and derive a correlation network. - Calculate the network’s global clustering coefficient score. - Calculated on one-month rolling windows spanning 2004Q3 to 2014Q3; monthly values temporally disaggregated to daily via Chow-Lin Max-Log. - Statistical tests indicate a structural break occurred in 2008: - Post-break connectivity has a higher mean and lower variance, suggesting the financial system is more “consistently connected” post-2008. - Clustering coefficient scale: 0 to 1 (probability that points in network share direct relationship). ### CCA stress test scenarios and controls - Stress test scenarios identical to those used in top-down balance-sheet based stress tests for banks (IMF/DFAST scenarios). - Models estimated separately for each sector and for the overall U.S. financial system. - Controls implemented: - Quantile-based default probabilities as control for idiosyncratic firm-level risk. - Macroeconomic variables to control for macro risk. - Statistically significant connectivity measure to control for relational/intersectoral changes. - Intersectoral spillovers captured via relationships estimated between sector default probabilities, given the stated controls. ### Stress test projections and results - Under the stress scenario: - Expected one-year median default probabilities for the U.S. financial system projected to increase from 0.20 percent to 0.65 percent. - This projection is about two-thirds the level seen at the height of the 2008–09 financial crisis. - Sectoral projections: - On average, default probabilities for all sectors expected to increase to about two thirds of their 2008–09 levels, except corporates and asset managers which are expected to experience smaller increases. - Using 0.5 percent as the low credit risk boundary: - Under the stress scenario, banks, insurers, NBFIs, and foreign banks exit the “safe zone” during peak stress occurring in 2015–16. - Corporates, asset managers, and foreign insurers do not breach the 0.5 percent threshold even under severely adverse macroeconomic conditions. - Projection notes: - To better show projection details, the CCA presentation truncates the y-axis at 1 percent (the historical 75th percentile reached a maximum value of 2.5 percent in 2008-09). - Only overall U.S. financial system projections explicitly incorporate changes in estimated default probabilities of other sectors; individual sector projections use macroeconomic and connectivity factors only. *Source: IMF staff calculations and methodology as described in the CCA stress testing section.* ### 166. The analysis shows the importance of spillovers across sectors. Spillovers from the United ### _cr15173 - 166. The analysis shows the importance of spillovers across sectors. Spillovers from the United ### Spillovers across sectors: key findings - Spillovers from the United States to the rest of the world were found to be large, but spillbacks from rest of the world to the United States appear relatively limited. - The estimated effect of credit risk shocks (Figure 22) varies from sector to sector. - Shocks to the credit profiles of U.S. asset managers, NBFIs, insurers, and foreign banks increased U.S. bank credit risk the most. - Increases in overall U.S. credit risk and credit risk in foreign insurers most negatively impacted the foreign banking sector. - The foreign insurance sector is most negatively affected by adverse changes in the credit profile of U.S. asset managers and insurers. - The effect of a rise in default risk in foreign banks and foreign insurers on U.S. default risk appears small. - Excluding the impact of macro factors, non-U.S. domiciled banks and insurers were found to collectively account for only 3 percent of median U.S. default probabilities. - Under the 2015–16 peak period of the stress scenario, this contribution is projected to fall by an additional one-third, to 2 percent of the total attributable to non-macro factor components. ### Figure 22: Spillover exercise (description and method) - Exercise: Effect on Sector X of a 0.5 percent Default Probability Increase in Sector Y. - The figure shows the marginal contribution (in percent) of a one-time shock on a recipient sector’s default probabilities given a severe 50 basis point (0.5 percent) shock to an originating sector’s default probabilities. - Results calculated using the GAMLSS inter-sector relationships estimated as of end 2014Q3. - Sector end of period default probabilities were separately increased by 50 basis points and shocks were measured as the deviation in a recipient sector’s probability of default from its unshocked historical baseline. - Dashed red lines denote cross-border effects of credit risk shocks. - Spillover effects were captured by incorporating historical estimates of median sector default probabilities as independent variables into the final specification of each of the sector projection models used to generate the results shown in Appendix Figure 8. ### Methodological note on insurer-to-bank effects - The effect of insurers on banks may appear quantitatively smaller than suggested by the SyRIN approach used in the previous section. - This reflects methodological differences between the two approaches: - SyRIN focuses on the joint probability of distress. - The present approach explicitly takes into account the effect of macroeconomic variables on probabilities of distress. - Interpretation: There are risks of spillovers from insurers to banks, but they partly reflect common (macroeconomic) factors. ### Recommendations for improvement (policy and methodological) - Link liquidity, solvency and network analysis in a systemic risk stress testing framework. - Example: Bank of Canada’s Macro Financial Risk Assessment Framework captures solvency, liquidity and spillover effects within a single stress testing framework (Bank of Canada, 2014). - Reexamine solvency stress test assumptions to make them consistent with historical evidence, including assumptions on loan and asset growth and dividend distribution. - Increase coverage of the tests. - The FRB should include the largest SLHCs in the supervisory stress tests once they start performing company-run stress tests (from 2017). - Establish a regular liquidity stress testing framework for banks. - The announced Comprehensive Liquidity Analysis and Review (CLAR), expected to be launched by the end of 2015, is a step in the right direction. This will complement solvency testing under the DFA. - Expand FRB data on interbank exposures to improve modeling of network contagion. - Network contagion exercise indicates need for richer set of dynamics and broader range of counterparties. - Insurance sector improvements: - Develop and perform insurance stress tests on a consolidated, group-level basis, especially for groups that are: - designated as systemically important; - engaged in material group-internal risk transfer, e.g., via captives; or - exposed to non-linear market risks through sale of products with guarantees or optionalities, e.g., variable annuities. - Improve public disclosure by requiring insurance companies to disclose market risk sensitivities in a more harmonized manner. - Mutual fund industry: - Conduct regular system-focused liquidity risk analysis for the mutual fund industry. - Authorities indicate forthcoming rule-making to standardize stress tests by mutual funds with consolidated assets of $10 billion or more. - Authorities are encouraged to clarify guidance to the industry on liquidity risk analysis and to begin regular top-down analysis to provide a more holistic picture of the industry’s contribution to systemic risk. - Monitoring and market-based measures: - Conduct more intensive monitoring of systemic financial sector risks, including use of market-based solvency and shortfall measures. - Market-price based stress tests provide forward-looking, higher-frequency, market consensus information that can add value as a “cross-check” to traditional stress tests and can be extended to sectors not traditionally subject to bank-like supervisory oversight. ### Risk Assessment Matrix: selected highlights and quantified scenarios - Risk 1. Abrupt surge in market volatility - Likelihood of Severe Realization of Threat in the Next 1–3 Years: High - Expected Impact on Financial Stability if Threat is Realized: High - Noted facts: - Over the past five years, the share of credit instruments held by retail funds has increased substantially, to 37 percent of total credit holdings. - A 50 bps permanent increase in 10-year interest rates could subtract about ½ percent of GDP after two years. - Risk 2. Imbalances from protracted period of low interest rates - Likelihood: Medium - Expected Impact: High - If unaddressed, distortions could lead to financial instability with significant economic costs and large spillovers to the rest of the world. - Risk 3. Operational risk - Likelihood: High - Expected Impact: Medium - Example: If a large solar storm similar in size to the 1859 event hit the world now (an event with an estimated 12 percent likelihood in the next 10 years), cost estimates are 2 trillion dollars with power and satellite outages lasting for months. - Risk 4. Protracted period of slower growth and lower inflation in advanced and emerging economies - Likelihood: High - Expected Impact: Medium - Slower growth could subtract about ½ percent of GDP after two years. - Risk 5. Geopolitical risks - Likelihood: Medium - Expected Impact: Low - A sustained 15 percent increase in oil prices above baseline would subtract about 0.2 percent of GDP after two years. - Risk 6. Bond market stress from a reassessment in sovereign risk - Likelihood: Low - Expected Impact: High - A 200bps increase in the benchmark Treasury yields would subtract 2.5 and 1.5 percentage points from U.S. growth in 2015 and 2016, respectively. ### Stress Test Matrix: Banking sector solvency risk (selected assumptions) - Institutional perimeter - Institutions included: 31 largest bank holding companies (with total consolidated assets of $50 billion or more) - Market share: 85 percent of total banking holding companies’ assets - Data and baseline date: - BHCs’ own data; Consolidated banking group; Baseline date: 2014 Q3 - Supervisory data (FR Y-14); Consolidated banking group; Baseline date: 2014 Q3 - Publicly available data (FR Y-9C); Consolidated banking group; Baseline date: 2014 Q3 - Channels of risk propagation / Methodology - Bottom-up: BHCs’ internal models consistent with supervisory expectations. - Top-down by Authorities: Fed’s top-down approach similar to balance sheet-based approach based on granular data. - Top-down by FSAP Team: Balance sheet-based approach with satellite models for macro-financial linkages. - Satellite/Projection approaches include: - Pre-provision net revenue forecasted using autoregressive models relating components of a BHC’s revenues and non-credit-related expenses to BHC characteristics and macroeconomic variables. - Income statement and balance sheet items (loans and trading assets) modeled using (i) panel regression with fixed effects, (ii) univariate regression models using aggregate data (for trading income, AOCI and provisions effects); both types include macro variables as explanatory variables. - Loan losses projected separately for different categories of loans based on obligor type, collateral, loan structure and accounting treatment; loan data came from FR Y-14 report. - Stress test horizon - Bottom-Up by Financial Institutions: 2014Q4-2016Q4 - Top-Down by Authorities: 2014Q4-2016Q4 - Top-down by FSAP Team: 2014Q4-2019Q4 *Italic: Excerpted from IMF staff calculations and accompanying narrative in the provided content.* ### 3. Tail shocks ### _cr15173 - 3. Tail shocks ### Scenario analysis - Baseline: reflected the average projections from the most recently available consensus view of the macroeconomic outlook expressed by professional forecasters, government agencies and other public sector organizations. - Adverse: represented conditions that are more adverse than those in the baseline scenario but less severe than those in the severely adverse scenario. It was formulated as one-half or two-thirds of the deviation of the paths of the variables relative to the baseline or severely adverse scenario. It also includes specific risk assumptions not found in the other scenarios, such as the rise in short-term interest rates. - Severely adverse (stress): reflected conditions of a specific recreation of a post-war U.S. recession. In the scenario: - unemployment rate rose by a 4 percentage point rise over a two-year period. - Real GDP was 4.5 percent lower than the baseline by the end of 2015 (GDP growth rates were negative for 5 quarters). - equity prices fell by 60 percent in one year. - house prices declined by 25 percent over the first two years. - corporate spreads rose by 330 basis points. - mortgage rates increased by 80 basis points. - DFA baseline scenario: broadly similar to IMF projections that follow the January 2015 WEO. - DFA stress: DFA severely adverse scenario that characterizes a V-shaped recession over 2014Q4-2016 followed by a three year recovery. - BHCs’ developed scenarios: baseline and severely adverse which reflected a company’s idiosyncratic risks and unique vulnerabilities to different factors including firm-specific events. ### Sensitivity analysis / one-time add-on shocks - Global market shock: - add-on shock applied to six companies with significant trading activity to their trading book and private equity positions (including their CVA) as of October 16, 2014. - entailed estimating trading and counterparty mark-to-market losses and incremental default risk on their trading exposures. - assumed to be an instantaneous event which immediately affects the market value of the trading assets and liabilities. - Counterparty default shock: - add-on shock involved the instantaneous and unexpected default of the companies’ counterparty with the largest net stressed losses as of October 16, 2014. - applied to eight large and highly interconnected companies. - applied to companies’ securities lending and repurchase/reverse repos and derivative exposures. - Other one-time or episodic shocks listed: - Interest rate spike in 2015. - Oil price shock (large drop in 2014-2015). - Modeling and operational assumptions: - Provisions modeled as net charge offs in a panel regression framework, also by loan types. - Constant dividend distribution. - Constant loan supply and total assets growing at 2.5 percent per year. - No operational RWAs in calculation of total RWAs. - DFAST hurdle rates. ### Risks and Buffers (general) - Credit risk: - granular sectoral exposures assessed (households, corporates, sovereign, financial institutions exposures). - Taxes: - Tax rate: After-tax net income (or loss) was calculated by applying a consistent tax rate to pre-tax net income (or loss) for all BHCs. - Taxes: set at the pooled average tax rate over the last 20 years. - Behavioral adjustments and balance-sheet projection rules: - companies projected balances for each major segment of the balance sheet consistent with scenario conditions and competitive and strategic factors. - For supervisory scenarios two sets of capital actions were used: BHC’s planned capital action and DFA stress test capital action assumptions. - Under the DFA stress test capital action assumption: - for the first quarter of the stress horizon, companies took into account its actual capital actions as of the end of the calendar quarter. - For each of the second through ninth quarter, companies were required to use a standardized set of capital action assumptions specified in the Dodd-Frank Act. - For each of the second through ninth quarter common stock dividend payments were assumed to continue at the same level as the previous year. - FRB projection of BHC balance sheets began with models relating total assets and subcomponents to nominal GDP and other macroeconomic factors including a measure of loan supply; relationships estimated using aggregate data used to project growth under supervisory scenarios. - In the adverse and severely adverse scenarios, the measure of loan supply was assumed to remain at its long-run historical average over the planning horizon. - Industry assets, loans, and trading assets were allocated to each BHC based on its shares of these positions at the beginning of the planning horizon. - All types of loans at all BHCs were assumed to grow at the same rate, meaning the mix of loans at each BHC was constant over the planning horizon. - Balance sheet growth and deposits growth were assumed to be equal to credit growth modeled and forecasted using a panel regression with fixed effects and macro variables as exogenous variables. - Dividend payout schedule followed capital conservation rule also taking into account GSIB surcharge; BHCs could distribute maximum dividend amount equal to dividend payout ratio (dividends over net income) in the base-year if they are not capital constrained; dividends were paid out only if bank records profits. - Asset disposals and acquisitions over time were not considered; the portfolio composition remained unchanged over time, with maturing exposures replaced with similar ones. - Any capital actions designed to offset the impact of the stress scenario on the bank are not allowed. - Trading assets were assumed to grow at the growth rate of total assets, adjusted for changes in mark-to-market values reflecting industry-wide trading asset composition. - Securities growth at each BHC was set so that total asset growth equals the projected rate given the growth of other asset types at that firm. - Balance sheet projections incorporated expected changes to a BHC’s business plan, such as mergers, acquisition, and divestitures, that are likely to have a material impact on its capital adequacy and funding profile. ### Regulatory and Market-Based Standards and Parameters - Calibration of risk parameters: - Company own calibration of parameters. - PDs were generally modeled as part of a transition process in which loans move from one payment status to another (e.g., from current to delinquent) in response to economic conditions. - LGD were typically defined as a percentage of EAD and was based on historical data. - Total RWAs were modeled (in a panel regression) as a function of total assets and interest rates to make RWAs risk sensitive. - 2015 switch to standardized approach to calculate RWAs was calibrated using Fed’s estimate of RWAs using standardized approached and using generalized approach. - Operational RWAs (for advanced approaches BHCs that exited the parallel run) were assumed to grow with total assets. - Regulatory/Accounting and Market-Based Standards: - Capital standards: The capital definition corresponded to that required by local regulation, i.e., Basel III for advanced approaches BHCs and non-advanced approaches BHCs (from January 1, 2015) and Basel I for non-advanced approached BHC for the first quarter of the stress test. - Capital standards: Basel III capital standards for all BHCs (CET1 estimated for non-advanced approaches BHCs); standardized approached to calculating RWAs from January 2015. - Hurdle rate: Basel III schedule (regulatory minimum for CET1), capital conservation buffer, GSIB. - Capital metrics: Tier 1 common capital ratio, common equity tier 1 ratio, Tier 1 capital ratio, total capital ratio and tier 1 leverage ratio. - Hurdle rates: Basel III minimum capital requirements; capital conservation and GSIB surcharge were not considered as part of hurdle rates. - Capital surcharge (using FSB buckets); phase ins of deductions from CET1 (for all BHCs) and phase outs of AOCI were also taken into account (from advanced approaches BHCs). - Risk-weighted assets: - generalized approach (2014Q3-2014Q4); standardized approached to calculating RWAs from January 2015. - RWAs were projected based on corresponding projections of on- and off-balance sheet exposures and their risk attributes consistent with the scenario conditions and standardized RWAs under the revised capital framework. - Two components of RWAs were projected: market RWAs and two types of credit RWAs, generalized RWAs under the capital framework. - Generalized risk weights were imputed from FR Y-9C data and held fixed throughout the forecast horizon. - Credit RWAs were calculated under the revised capital framework implementing the Basel III regulatory capital regime in the United States (standardized approach RWAs) from 2015q1 onwards; the weights used to calculate credit RWAs were held fixed throughout the planning horizon. - Market RWAs were broken down into cyclical and non-cyclical components. The cyclical group, including value at risk and the incremental risk charge, was projected based on the volatility of the trading portfolio of the BHCs. The non-cyclical group, including value at risk, stressed value at risk, the specific risk charge, and the comprehensive risk charge, was assumed to evolve according to projections of the BHCs’ trading assets. - Operational RWAs were not considered in some parts of the analysis. ### Banking sector: Liquidity risk - Institutional perimeter: - Top-down by FSAP Team included 31 largest BHCs. - Market share: 75 percent. - Data and baseline date: Publicly available data (FR Y-9C); consolidated banking group as of 2014Q3. - Channels of risk propagation / Methodology: - Liquidity metric was defined as the ratio of liquid assets over outflows using balances outstanding. - Tail shocks: - Shocks reflected in adjustment factors (haircuts and run-off rates) applied to high-quality liquid assets and outflows; factors were informed by historical dynamics of liquid assets and outflow categories. - Risks and buffers: - Risks: Funding liquidity risk; Market liquidity shock. - Buffers: Liquid assets. - Calibration of risk parameters and regulatory standards: - Benchmark: haircuts calibrated based on the LCR; run-off rates calibrated based on historical experience. - Sensitivity analysis: run-off rates calibrated based on the LCR. - Threshold set to 1. ### Banking sector: Spillover risks - Institutional perimeter: - Top-down by FSAP Team included 6 largest BHCs. - Market share: 52 percent of BHCs’ assets. - Data and baseline date: 2014Q3. - Methodology: - Espinosa and Sole (2013) network analysis. ### Insurance sector: Solvency risks - Institutional perimeter: - Top-down by Authorities: Full coverage of solo entities: 751 life insurers; 2,569 P&C insurers. - Top-down by FSAP Team: 43 Insurance groups, of which: 20 life insurers; 16 P&C insurers; 5 health insurers; 2 credit insurers. - Market share: Authorities: 100 percent; FSAP Team: 40 percent (gross written premiums). - Data and baseline date: Authorities and FSAP Team: statutory accounting and U.S. GAAP; baseline date end-2013. - Channels of risk propagation / Methodology: - Investment assets: market value changes after price shocks, affecting total adjusted capital (TAC) and shareholder equity. - In sensitivity analysis: profitability negatively affected through prolonged period of interest rates: spread between net portfolio yields and guaranteed credited rates declining, ultimately resulting in weaker capital position. - Stress test horizon: - Instantaneous shock (except for “low-for-long” analysis: 5 years, i.e. 2014-2018). - Tail shocks / Sensitivity analysis: - Natural catastrophes: - Florida hurricane (similar to but worse than “Andrew”, 1992). - California earthquake (similar to but worse than “Northridge” event, 1994). - Series of tornados (three tornados, classified as EF5, in the Midwest of the United States). - Prolonged period of low interest rates (“low-for-long”): assuming unchanged level of interest rates until 2018. - Pandemic. - Risks, shock magnitudes and aggregation: - Market value shocks (Authorities): - Sovereign bonds: +2%. - Municipal bonds: -5%. - CMBS and RMBS: between -5% and -28%. - Mortgage loans: -5%. - Equity: -40% for unaffiliated, -30% for affiliated, -10% for preferred. - Property: -30%. - Other investment assets: -30%. - Potential loss from guarantees in variable annuities: 10% of the maximum guaranteed amount. - Catastrophic risk: 1-in-250 year event for the Florida hurricane and the California earthquake. - Market value shocks (FSAP Team): - CMBS, RMBS and other ABS: between -28.7% and -51.3%. - Mortgage loans: -12.2%. - Equity: -28.9%. - Property: -28.3%. - Other investment assets: -28.9%. - Credit spread shocks: - U.S. sov. bonds: +58.4bp. - Municipal bonds: between +66.0bp (AAA-A) and +429.0bp (BB). - Other sov. bonds: between +56.7bp (AAA-A) and +1,661bp (CC and lower). - Corporate bonds: between +248.1bp (AAA-A) and +1,878bp (CC or lower). - Shock to the USD risk-free interest rate: between -21bp for 1 year and -143bp for 30 years. - Catastrophic risk: 1-in-250 year event for the Florida hurricane and the California earthquake. - Pandemic: higher mortality rate: +1.5 additional deaths per 1,000. - Risk aggregation: sum of individual shocks. - Buffers: - Coverage of risk-based capital (RBC). - Shareholder equity. - Behavioral adjustments: None. ### Asset managers: Liquidity risks - Institutional perimeter: - Top-down by FSAP Team: Largest mutual funds (9.000 mutual funds) divided into different styles of mutual funds. - Data and baseline date: 2014Q3. *Source: _cr15173 - 3. Tail shocks (IMF).* ### 2. Channels of Risk Propagation ### _cr15173 - 2. Channels of Risk Propagation ### Methodology (Mutual Funds / Liquidity Stress) - Comparing redemptions with capacity of market for a particular asset class to be sold in an orderly manner (securities of mutual funds mapped into portfolio categories of NY Federal data on dealers’ inventory). - Assuming a ranking of assets to be sold to meet redemptions. - Redemptions applied to individual mutual funds of the same style; redemptions calculated by style of mutual fund as an average redemption rate across all mutual funds of the same style; shock: 1st percentile of redemption rates’ distribution. - For MMFs: an approximation of the LCR calculated. - Stress test horizon: One quarter (shock characterized by a run on a fund represented by an assumed redemption rate). ### Tail shocks / Sensitivity Analysis (Mutual Funds) - Tail shocks: N.A. - Sensitivity analysis: N.A. ### Risks and Buffers (Mutual Funds) - Risks: - Liquidity risk (a run on a mutual fund). - Buffers: - Liquid assets. - Capacity of a particular market to absorb sell-off of corresponding asset to meet redemptions (by comparing mutual fund’s portfolio of a particular security with dealers’ inventory of the same security). --- ### MARKET-PRICE BASED: CREDIT/DEFAULT RISKS AND CROSS-BORDER SPILLOVER RISKS — Institutional Perimeter - Institutions included: - 210 institutions with publicly traded equity (divided into 7 different sectors, five domestic financial and non-financial sectors and two foreign financial sectors). - 178 institutions are domiciled in the U.S. and 32 are domiciled in non-U.S. jurisdictions. - Domestic sectors include U.S. banks, insurers, non-financial corporates, asset managers, and other non-bank financial institutions. - Foreign sectors include foreign banks (non-U.S. GIBs) and foreign insurers (non-U.S. G-SIIs and other large non-U.S. insurers). - Market share inclusion thresholds: - U.S. Banks and insurers (75 percent or greater of total sector assets). - U.S. corporates, asset managers, and other-non bank financial institutions (40% or greater of total sector assets). - Foreign banks and insurers (100% of GSIB and G-SII total assets). ### Data and Baseline Date - All data from end 2004Q3 to 2014Q3. - Credit risk data from CreditEdge. - Macro data from Haver and Datastream. - Connectivity data computed by IMF using CreditEdge data. - Non-daily data temporally disaggregated to daily frequency using Chow-Lin Max-Log methodology. ### Methodology (Market-Price Based) - Contingent Claims Analysis. - General Adaptive Models of Location, Scale, and Shape (GAMLSS). ### Tail shocks (Market-Price Based) - Size of the shock: - DFA/IMF stress and baseline macroeconomic scenarios. - One time 50 basis point shock to counter-party sector default probabilities. ### Risks (Market-Price Based) - Macro risks. - Interconnectivity risk. - Cross-border and domestic spillover risk. - Credit risk. - Default risk. --- ### Appendix II — Additional Data: Banking Sector (Selected structures and shares) - Data source for figures: FRB, SNL, IMF Staff calculations. - Time coverage shown in figures: 2000–2014 (many series to 2014Q3 or 2014). - Structure of AfS securities portfolio (percent shares shown): - Treasury securities 16% - U.S. government agency obligations 3% - Municipal securities 7% - MBSs 49% - ABSs 4% - Structured financial products 4% - Domestic debt securities 2% - Foreign debt securities 14% - Mutual funds and equity 1% - Structure of MBS portfolio (percent shares shown): - Pass-through RMBS guaranteed by GNMA 10% - Pass-through RMBS issued by FNMA and FHLMC 47% - Other pass-through RMBS 0% - RMBS issued/guaranteed by GSEs 22% - RMBS collateralized by MBS issued/guaranteed by GSEs 0% - Other RMBS 11% - Pass-through CMBS issued/guaranteed by FNMA, FHLMC, GNMA 2% - Other pass-through CMBS 0% - CMBS issued/guaranteed by GSEs 2% - Other CMBS 6% - Structure of trading portfolio (percent shares shown): - Treasury securities 9% - U.S. government agency obligations 2% - Municipal securities 2% - MBSs 10% - ABSs 1% - Debt securities 24% - Real estate loans 2% - Commercial and industrial loans 2% - Consumer loans 0% - Other loans 1% - Other trading assets 29% - Derivatives 18% - Structure of trading liabilities (percent shares shown): - Other trading liabilities 1% - Derivatives 40% - Short sale-Equity 18% - Short sale-Debt securities 41% - Short sale-Other assets 0% - Structure of other borrowed money (percent shares shown): - CP 7% - Other borrowed with maturity <1 24% - Other borrowed with maturity >1 69% - Secured vs. unsecured composition: - Secured 29% - Unsecured 71% - Structure of wholesale funding (percent shares shown): - Other borrowed money 47% - Fed funds/repos 24% - Trading liabilities 18% - Subordinated notes 5% - Brokered deposits 6% - Deposits structure: - Deposits in domestic offices 82% - Deposits in foreign offices 18% - Noninterest bearing 25% - Interest bearing demand deposits 6% - Money market deposits 59% - Time deposits of less than $100.000 5% - Time deposits of more than $100.000 5% - Stable deposits 89% - Structure of loans (selected categories shown across time series): - Residential real estate loans, Commercial real estate loans, Business loans, Consumer loans, Loans to financial institutions and other loans, Loans to foreign government and official institutions (figures shown in series 2000–2014Q3). - Securities and trading series (selected): - HTM securities, AfS securities, Trading asset securities, Trading liabilities series displayed across 2000–2014 and 2009–2014 intervals in charts. - Derivatives position, gross amounts (percent shares shown): - Futures 3% - Forwards 17% - Exchange traded-Written options 2% - Exchange traded-Purchased options 3% - OTC-written options 7% - OTC-purchased options 7% - Swaps 61% - Derivatives by contract type (percent shares shown): - Interest rate contract 82% - FX contracts 14% - Equity derivatives 3% - Commodity and other contracts 1% - Gross notional held for trading 99%; gross notional held for other purposes 1%. - Off-balance sheet items composition (examples): - Revolving, open-end loans secured by 1-4 family residential properties 7% - Unused consumer credit card lines 49% - Other unused credit card lines 3% - Commitments to fund CRE, construction and land development loans secured by RE 2% - Commitments to fund CRE, construction and land development loans not secured by RE 1% - Securities underwriting 0% - Commercial and industrial loans 31% - Loans to financial institutions 7% - OTC derivatives collateral and counterparties (percent shares shown): - Fair value of collateral: Cash US dollar 17%, Cash other securities 2%, US Tresury bonds 1%, US gouvernment agency securities 0%, Corporate bonds 2%, Equity securities 6%, All other collateral 50%, Fair value of collateral 22% (labelled separately). - Counterparties: Banks and securities firms 54%, Monoline Financial guarantors 0%, Hedge funds 2%, Sovereign governments 6%, Corporations and all other counterparties 38%. - Credit derivatives (notional and composition): - CDS 96% - TRS 1% - Credit options 3% - Other credit derivatives 0% - Credit derivatives sold protection: Investment grade 72%, Subinvestment grade 28%. - Credit derivatives purchased protection: Investment grade 73%, Subinvestment grade 27%. - Credit performance indicators (series 2000–2014Q3): - Delinquencies (as % of total net loans) and NPLs (as % of total net loans) shown across business, consumer, CRE, FIs and other loans, RRE and government categories. - Net charge offs series across loan types. - ROE and ROA series (percent scales shown). - GSIBs (selected exposures and shares): - Total exposure and total assets plotted for Bank of America, Goldman Sachs, J.P. Morgan, Morgan Stanley, Wells Fargo, Citigroup, State Street, New York Mellon. - Foreign claims and liabilities (% of total assets) and foreign net revenue (in % of total revenue) displayed across these GSIBs. - Intra-financial system assets and liabilities (% of total assets) for the same GSIB set shown. - Investment securities (percent shares shown for GSIBs): - U.S. Treasury securities 18% - U.S. agency securities (excl MBS) 2% - Municipal securities 8% - Mortgage-backed securities 45% (investment securities) / 42% (AfS securities) - Asset-backed securities 9% (investment securities) / 10% (AfS securities) - Other debt securities 17% (investment securities) / 19% (AfS securities) - Mutual funds and equity securities 1% - Securitization activity (percent shares shown): - 1–4 family residential loans 84% - Home equity lines 1% - Credit card receivables 0% - Auto loans 0% - Commercial and industrial loans 0% - All other loans and leases 15% - Intra-financial system composition examples: - Funds deposited or lent to unaffiliated financial institutions 20% - Undrawn commitments 10% - Secured debt securities 2% - Senior unsecured debt securities 3% - Subordinated debt securities 0% - Commercial paper 0% - Stocks 5% - Net positive current exposure of SFT 8% - OTC derivatives - net positive value 8% - OTC derivatives - Potential exposure 44% - Net negative current exposure of SFTs 18% - OTC derivatives - Net negative fair value 8% - OTC derivatives - Potential future exposure 27% - Deposits due to depository institutions 11% - Deposits due to non-depository financial institutions 36% *Source: _cr15173 - 2. Channels of Risk Propagation (IMF staff chapter content).* ### Appendix Figure 3. GSIBs (continued) ### _cr15173 - Appendix Figure 3. GSIBs (continued) ### GSIBs: OTC Derivatives and Derivatives Structure - OTC derivatives (in % of total assets): chart headings list banks: Bank of America, Goldman Sachs, J.P. Morgan, Morgan Stanley, Wells Fargo, Citigroup, State Street, New York Mellon. - Structure of OTC derivatives: - OTC derivatives cleared through a central counterparty: 53% - OTC derivatives settled bilaterally: 47% - Derivatives: Held for trading: 99%; Non-traded: 1% - Derivatives by type: - Interest rate contracts: 82% - Foreign exchange contracts: 14% - Equity, comm, & other contracts: 4% - Derivatives by instrument: - Futures and forwards: 52% - Written options: 24% - Purchased options: 24% - Derivatives by maturity: - One year or less: 61% - Over 1 year to 5 years: 24% - Over 5 years: 15% - Source cited: FRB, SNL, IMF Staff calculations ### Macroeconomic Scenarios: Main Variables (Appendix Figure 4) - Variables depicted across Q1 2007 through Q1 2019: - GDP growth, q-o-q, annualized, in % - Unemployment rate, in % - 3-month Treasury yield, in % - 10-year Treasury bond yield, in % - Dow Jones Industrial Stock Market Index - House price index - Commercial Real Estate Price Index - Market volatility index - Euro area GDP growth - Developing Asia real GDP, q-o-q rate, % - UK GDP growth - Source cited: FRB, IMF Staff calculations ### Estimation Results: Income Statement Items (Appendix Table 3) - Dependent variables and model specifications summarized (panels/aggregates, shares, lags). Selected coefficient highlights (preserve numeric formatting and significance markers exactly as shown): - (Loans x Lending rate): 0.001775*** - (Deposits x Depoit rate): -0.001586*** - Total assets, y/y: 0.856675*** - Lending rate: -0.000118*** and 0.000133*** - Real GDP, y/y: 0.000260***; -0.000323***; -0.000194*** - Real GDP, y/y x Dum (=1 if y/y<0): 0.000682*** - Unemployment rate: -0.000408*** - d(Unemployment rate): 0.001762*** - d4(Unemployment rate): 0.001174*** - d4(10-year Treasury yield): -0.000512*** - d(10-year Treasury yield - 3 month): 0.000773*** - d4(BBB corporate yield): -0.000112 - BBB corporate yield - 3-month rate: -0.000819*** 0.000130 0.000336*** - Mortgage rate: (coefficients shown in table) - House Price Index (y/y): -6.85E-05***; -8.70E-05*** - Market Volatility Index, y/y: -0.00000313***; 0.00000884***; -6.58E-06 - Market Volatility Index, y/y x Dum(=1 if y/y>0): -1.76E-05***; 1.90E-05** - Market Volatility Index, level: 1.77E-05*** - Constant entries: 26819.86***; 0.009149***; 0.001206***; 2.369656***; -0.001447***; 0.000369 - R^2 entries: 0.53; 0.81; 0.41; 0.39; 0.54; 0.84; 0.56 - Number of observations listed: 2231; 22879; 5223; 29795; 2237 (as shown in table form) ### Estimation Results: Balance Sheet Items, Net Charge offs, RWAs (Appendix Table 4) - Dependent variables and coefficients summarized (net loans, total assets, RWAs, NCOs by loan type). Selected coefficient highlights: - Net loans, y/y: 0.346118***; 0.793545*** - Total assets, y/y: 0.899295*** - Lending rate: 0.907326***; 0.00002 - d4(Lending rate): -2.299608*** - Real GDP, y/y: 1.45978***; -0.00007**; -0.000123**; -0.000467***; -0.000242*** - Unemployment rate: 0.000688***; 0.001498*** - d4(Unemployment rate): 0.001008***; 0.000851*** - BBB corporate yield - 3-month rate: 0.000178***; 0.000271*** - d4(Mortgage rate): 0.000247*** - House Price Index (y/y): -0.000114***; -0.00004***; -0.00008*** - Commercial Real Estate Price Index (y/y): -0.00004*** - Market Volatility Index, level: -0.070382***; 0.00001; 0.00003***; 0.00004***; 0.00007***; 0.00001** - Business interest rates: 0.000508*** - d4(Consumer interest rates): 0.000598*** - d4(Federal funds rate): 0.000187 - Constant entries: 2.808341***; 2.584347***; -4.794070***; 0.000981***; 0.004841***; -0.003267***; -0.011038; 0.000731; 0.001724*** - R^2 entries: 0.39; 0.76; 0.84; 0.37; 0.56; 0.46; 0.28; 0.08; 0.56 - Number of observations entries: 2088; 2253; 1769; 2179; 2199; 2166; 2199; 2287; 2172 ### Banking Stress Test Results: IMF Top-Down (Appendix Figure 5) - CET1 ratio and contributions to CET1 ratio shown across 2014q3 through 2019 for the GSIB group. - Components in charts include: - Net interest income - Non-interest expense - Non-interest income - Trading income - Provisions for credit losses - Other (dividends, taxes, AOCI, other ded.) - RWAs change in CET1 - CET1 (%)-right axis - Regulatory threshold and GSIB CET1 charge indicators: - Regulatory threshold - Max GSIB CET1 charge - Min GSIB CET1 charge - Net income over the stress horizon, Recapitalization needs (share of 2014 net income), Recapitalization needs (share of nominal GDP) charts included. - Small-value axis examples preserved as in figure: 0.0%; 0.2%; 0.4%; 0.6%; 0.8%; 1.0%; 1.2%; and monetary axes listing negative and positive ranges (1,000,000,000 through 1,000,000,000 in steps shown). ### U.S. Cross-Border Assets, Liabilities, and Flows (Appendix Figure 6) - Charts present cross-border positions in Billions US$. - Source cited: IMF, BIS, IMF staff calculations ### Insurance Sector: Assets and Bond Portfolios (Appendix III, Figure 7) - Breakdown of assets (percentages shown visually): - Bonds: 44% - Stocks: 3% - Mortgage loans: 5% - Real estate: 0% - Cash, cash equivalents, other short-term investments: 2% - Derivatives: 0% - Other investment assets: 4% - Separate account assets: 37% - Other assets: 5% - Breakdown of bond portfolio shares (visual percentages): - U.S. sovereign; Non-U.S. sovereign; U.S. municipals and GSEs; Corporate bonds and hybrids; Non-GSE securitizations; with a 7% label appearing in the figure - Life / P&C / Health / Credit breakdowns shown across maturity buckets and ratings - Assumed average duration of bond portfolio by sector plotted with axis values 0,0 through 9,0 (life, P&C, health, credit) - Source cited: SNL Financial and IMF staff calculations ### Market-Based Tests: GAMLSS Overall U.S. Financial System Model (Appendix IV, Table 5) - Additive term structure coefficient estimates and P-values: - Intercept term: -5.55E+00 < 2e-16 *** - 5-year treasury yield: -4.21E-02 < 2e-16 *** - Dow Jones stock market index: free knot N/A; fully non-parametric; see term plot (Figure I) - VIX localized regression (LOESS): N/A; fully non-parametric; see term plot (Figure I) - House price index: -1.56E-03 < 2e-16 *** - CRE price index: -2.00E-03 < 2e-16 *** - USD/euro exchange rate: -2.59E-03 < 2e-16 *** - BBB corporate yield: 5.41E-02 < 2e-16 *** - Real GDP growth rate: free knot N/A; fully non-parametric; see term plot (Figure I) - Euro area inflation varying coefficient: 1.00E-02 1.57e-07 *** - Median 1-year default probability for Asset Manager sector: 6.14E+01 < 2e-16 *** - Median 1-year default probability for Bank sector: 3.71E+01 < 2e-16 *** - Median 1-year default probability for Insurer sector: 1.29E+01 4.58e-07 *** - Median 1-year default probability for NBFI sector: 9.79E+00 3.83e-13 *** - Median 1-year default probability for Foreign Bank sector: 3.37E+00 2.74e-11 *** - Median 1-year default probability for Foreign Insurer sector: 3.196E-00 0.000129 *** - Global clustering coefficient score first order orthogonal polynomial: 1.89E+00 < 2e-16 *** - Global clustering coefficient score second order orthogonal polynomial: -3.77E-01 < 2e-16 *** - Global clustering coefficient score third order orthogonal polynomial: 2.17E-01 < 2e-16 *** - Model fit diagnostics: - Number of observations in the fit: 1244 - Degrees of Freedom for the fit: 44.94656 - Residual Degrees of Freedom: 1199.053 - Global Deviance: -20500.05 - AIC: -20410.15 - SBC: -20279.75 - Distribution type: Inverse Gaussian - Link function for the mean: Log - Generalized R-squared: 0.9965047 - Dependent variable: Median 1-year default probability for Overall U.S. Financial System - Source cited: IMF staff calculations ### Overall U.S. Financial System Model: Non-Parametric Term Plots and Diagnostics (Appendix Figures 8–10) - Non-parametric additive term plots: red lines denote the fit of a GAMLSS non-parametric additive term for the response variable (median 1-year ahead U.S. financial system default probabilities) against given explanatory variables. - Sector Default Probability Forecasts (Appendix Figure 9): - Blue lines indicate 25th and 75th percentile values of historical distribution of U.S. institution 1-year ahead default probabilities. - Dashed black line: median of historical distribution. - Solid red and black lines: median 1-year ahead default probabilities projected by the CCA stress tests under the stress and baseline scenarios, respectively. - Note on axis truncation: Blue 75th percentile reached maximum values of 4 %, 6.5 %, 2 %, and 16 % for domestic banks, insurers, asset managers, and NBFIs, respectively, in 2008-09. - Only projections for the overall U.S. financial system model explicitly account for changes in estimated default probabilities of other sectors; individual sector projections use macroeconomic and connectivity factors only. - Model diagnostics (Appendix Figure 10): - Worms plot of normalized quantile residuals: first two moments of the response distribution satisfactorily modeled; curved dotted lines denote 95% CIs; fitted central red line fairly straight. - Correlation matrix of predictors: shows low correlation among predictors; modeling assumption of uncorrelated predictors not grossly violated due to orthogonalized additive terms. - Model normalized quantile residuals closely follow a standard normal distribution; Q-Q plot supports appropriateness of chosen distribution. ### Significance of GAMLSS Sector Model Spillover Effects (Appendix Table 6) - Table reports statistical significance of positive credit risk shocks across recipient and donor sectors. Significance coding: 0 ***, 0.001**, 0.01*, 0.05'', 0.1''' - Generalized R-squared by recipient model: - Banks: 0.9731212 - Insurers: 0.9886623 - Corporates: 0.9840041 - Asset Managers: 0.9640468 - NBFIs: 0.9128003 - Foreign Banks: 0.9841397 - Foreign Insurers: 0.9904947 - Note: Cells indicate statistical significance; distribution type denotes assumed response distribution for each sector model. Overall U.S. model only considered effect of foreign sectors; response variable values for overall model drawn from pool of domestic sector observations. ### Financial Account Net Exposures: Sector and Instrument Definitions (Appendix Table 7) - Sector/Instrument mappings (examples preserved exactly): - Households: Households and Nonprofit Organizations - Corporates: Non-Financial Business - Government: General Government; Monetary Authority Domestic - GSEs: Government Sponsored Enterprises; Agency- and GSE-Backed Mortgage Pools - Banks: Private Depository Institutions - Insurers: Property-Causality Insurance Companies; Life Insurance Companies; - Pensions: Private and Public Pension Funds - Asset Managers: Money Market Mutual Funds; Mutual Funds; Closed End and Exchange Traded Funds; Real Estate Investment Trusts - NBFIs: Issuers of Asset-Backed Securities; Finance Companies; Security Brokers and Dealers; Holding Companies; Funding Corporations - Rest of World: Rest of World - Money Market: Net Interbank Transactions; Checkable Deposits and Currency; Time and Savings Deposits; Federal Funds and Security Repurchase Agreements; Open Market Paper - Fund Shares: Money Market Mutual Fund Shares; Mutual Fund Shares - Treasury Securities: Treasury Securities - Other Government Securities: Treasury Securities; Agency- and GSE-Backed Securities; Municipal Securities and Loans - Corporate Bonds: Corporate and Foreign Bonds - Corporate Equities: Corporate Equities - Deposits, Loans, Mortgages and Interbank Transactions: U.S. Deposits in Foreign Countries; Other Loans and Advances; Total Mortgages; Consumer Credit; Security Credit; Depository Institution Loans Not Elsewhere Classified; Net Interbank Transactions - Other Instruments: Special Drawing Rights (SDRs) Certificates and Treasury Currency; Trade Credit; Life Insurance Reserves; Pension Entitlements; Taxes Payable by Businesses; Proprietors' Equity in Noncorporate Business; Direct Investment; Total Miscellaneous Financial Claims - CDS: *BIS official statistics on credit-default swap notional amounts outstanding as of 2014Q2, adjusted for double-counting - Source cited: FRB’s Financial Accounts of the United States; Haver; BIS; IMF staff estimates. *Source: IMF Staff calculations and cited data sources as shown in the PDF content.* --- _Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/scr/2015/_cr15173.pdf_