## _081814a - EXECUTIVE SUMMARY ## Source details **Canonical URL:** [_081814a - EXECUTIVE SUMMARY](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/np/pp/eng/2014/_081814a.pdf) ## Other formats - [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/np/pp/eng/2014/_081814a.pdf.md) - [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/np/pp/eng/2014/_081814a.pdf.json) --- ### Purpose and scope - Standards assessments in FSAPs identify weaknesses in financial regulation and supervision and provide cross-country perspectives. - Focus of the paper: the three supervisory standards most frequently assessed in FSAPs—Basel Core Principles (BCP), Insurance Core Principles (ICP / IAIS), and IOSCO Principles on Securities Regulation (IOSCO). - Proposed approach applies only to supervisory standards assessments undertaken in the context of FSAPs or stability modules; full assessments remain available in other circumstances. ### Key problems identified - Integration and targeting: - Standards assessments are not well integrated into Fund surveillance: weaknesses identified are not specifically linked to risks and vulnerabilities facing the financial sector. - FSAP analysis of country-specific vulnerabilities does not guide targeting of standard assessments because formal assessments are exhaustive and must cover the entire standard. - Post-crisis complexity and resource intensity: - Revisions to core principles and methodologies extended the assessment process and the Detailed Assessment Report (DAR). - Human resource and dollar costs increased considerably, including assessors’ time in the field (for up to three weeks) and more extensive review. - Concerns about the Fund’s ability to continue delivering high-quality assessments and whether resources devoted are proportionate to net benefit for the FSAP and Fund surveillance. - Output and prioritization issues: - Formal assessments produce long, detailed DARs/ROSCs whose recommendations are not prioritized by financial stability relevance. - Limited use of targeted assessments: - The 2009 Revised Approach allowed targeted (partial) assessments but selection criteria remained compliance-based and narrow; since 2009 only two targeted assessments occurred: - Targeted BCP for Russia (IMF Report No. 11/336) - Targeted IOSCO for Canada (IMF Report No. 14/73) ### Evidence on assessment frequency (selected figures) - Average Number of ROSCs Conducted During FSAP Missions by Standard (FY08–FY14): - Overall: FY08 0.8; FY09 0.8; FY10 1.4; FY11 2.4; FY12 2.4; FY13 1.8; FY14 1.6 - BCP: FY08 0.5; FY09 0.6; FY10 0.8; FY11 0.9; FY12 0.9; FY13 0.7; FY14 0.9 - IAIS: FY08 0.0; FY09 0.1; FY10 0.2; FY11 0.5; FY12 0.7; FY13 0.5; FY14 0.5 - IOSCO: FY08 0.1; FY09 0.1; FY10 0.1; FY11 0.5; FY12 0.5; FY13 0.5; FY14 0.2 - CPSS: FY08 0.1; FY09 0.1; FY10 0.1; FY11 0.3; FY12 0.1; FY13 0.0; FY14 0.0 - CPSS/IOSCO: FY08 0.0; FY09 0.0; FY10 0.1; FY11 0.3; FY12 0.1; FY13 0.1; FY14 0.0 - MFPT: FY08 0.0; FY09 0.0; FY10 0.1; FY11 0.0; FY12 0.0; FY13 0.1; FY14 0.0 - Total number of ROSCs per fiscal year increased, peaking in FY11–FY12, with BCP, IAIS, and IOSCO accounting for most growth, driven mainly by S-25 countries. ### Proposed new approach (overview) - Objective: - (i) focus standards assessments in FSAPs on macrofinancially-relevant principles; and - (ii) incorporate findings systematically into the FSAP’s overall risk assessment. - Two core elements: - Focus assessments on macrofinancially-relevant principles: - A subset of principles is selected using empirical analysis and expert judgment, based on macrofinancial risks for banking, insurance, and securities sectors identified as relevant for financial stability. - Replaces compliance-based selection criteria with macro-relevant and risk-based criteria. - Identified as macrofinancially-relevant: - 11 BCP principles - 11 ICP principles - 17 IOSCO principles - Systemically incorporate assessment findings into FSAP risk assessment: - Based on compliance with selected principles, FSAP teams judge whether regulation and supervision mitigate or aggravate specific risks. - Findings reported into an expanded RAM (Risk Assessment Matrix), integrating supervisory findings into risk assessment and facilitating prioritization of policy recommendations. ### Implementation boundaries and continuity - Limited to supervisory standards assessments within FSAPs or stability modules; does not preclude full compliance assessments in other circumstances. - Full assessments remain appropriate when: - There has never been an assessment of the relevant standard. - Fundamental changes in supervisory architecture occurred since the last assessment. - The country requests technical assistance that would benefit from a prior full compliance assessment. - Full assessments can be stand-alone or combined with an FSAP if timing allows. ### The empirical methodology — three-stage overview - Stage 1: Identify proxies for macrofinancial risks relevant for banking, insurance, and securities sectors. - Stage 2: Prepare individual principles in each standard for empirical use; apply linear scoring 1–4 to qualitative grades (BCP: 1 compliant → 4 non-compliant; ICP: 1 observed → 4 not observed; IOSCO: 1 fully implemented → 4 not implemented). Missing principle ratings proxied by average score of rated principles in same group. - Stage 3: Use parametric and non-parametric econometric approaches (including EFA and PCA) to establish links between macrofinancial risks and individual principles (technical details in Appendix II). ### Macrofinancial risks and data coverage - Risk classifications: - Banking sector: credit, market, liquidity, and contagion risks. - Insurance sector: market/investment risks, insurance and underwriting risks, liquidity risks, and contagion risks. - Securities markets: credit risk, liquidity risk, and market risk; contagion risk across domestic markets and across borders included. - Data coverage and measures: - Bank and insurance risk measures from balance sheets/income statements of: - 1,426 individual banks in 83 countries - 848 insurance companies in 23 countries - Securities markets: market sources covering five market categories and market estimates of default rates for securities issuers in 77 countries. - Global financial market variables and a broad range of estimated risk premia included. ### Supervisory standards principles dataset — patterns and challenges - Number of principles per standard: - BCP: 29 BCPs adopted in 2012 - ICP: 26 ICPs issued in 2011 - IOSCO: 38 principles in 2010 (revised 2011); assessments generally cover 37 of 38 - Observed grade concentration and correlations: - Most grades concentrated in compliant and largely compliant categories; more advanced countries generally have better scores. - Many pairwise correlations of principles’ grades exceed 50 percent. - Unconditional average correlations among principles: - BCP: 34 percent - ICP: 39 percent - IOSCO: 26 percent - Example pairwise correlations: - Revised BCP 22 and BCP 23: 77 percent - Revised ICP 3 and ICP 25: 91 percent - Revised IOSCO P25 and P26: 66 percent - Time-varying methodologies: - Assessment methodologies evolved and became more conservative post-crisis; merging different versions can introduce noise. ### Empirical strategy (parametric and non-parametric) - Parametric approach: - Factor analysis to reduce dimensionality (PCA + EFA with Varimax rotation); four underlying factors retained explaining ~60 percent of variability. - Benchmark regression: risks Yijt (firm/market nested in country and year) with firm-specific effects μj, controls Xijt, macro MACROit; dynamic versions use Arellano-Bond where appropriate. - Expanded model includes interaction terms between macro fundamentals and individual principles/factors F; selection criteria require interaction sign and significance at the 10 percent level, robustness to additional interactions, and forecasting improvement for market data. - Factor interpretation (four factors per standard): - BCP: ‘Risk management’; ‘Legal’; ‘Information sharing’; ‘Asset quality’ - ICP: ‘Governance’; ‘Information sharing’; ‘Regulation’; ‘Prudential and Supervision’ - IOSCO: ‘Information sharing’; ‘Collective investment schemes’; ‘Market intermediaries’; ‘Issuers’ - Non-parametric approach: - Binary tree analysis to identify splitters (variables and thresholds) that best divide observations into high/low risk, weighting splits in adverse macro states. - Captures non-linear and compound effects; identifies principle-specific thresholds of compliance most relevant for mitigating risk. - Example: credit risk (NPL ratio) splits by interest rate thresholds and BCP grades (illustrative splits and effects provided). ### Data and estimation specifics - Temporal/frequency coverage: - BCP and ICP models estimated with annual data; BCP uses Bankscope firm-level and IMF FSI country aggregates; ICP uses SNL company-level and OECD/EIOPA aggregates. - IOSCO models estimated with monthly data; financial indicators from Bloomberg, Datastream, Consensus Forecasts, WEO, and Haver Analytics. - Risk proxies (selected): - BCP: Credit Risk → Loan losses to assets ratio / NPL ratio; Market Risk → Net gains or losses from securities to assets ratio; Liquidity Risk → Loans to deposits ratio; Contagion Risk → Banking sector EDF; Overall Risk → Z-score. - ICP: Market/Investment Risk → issuer, sovereign, equity risk constructs; Insurance and Underwriting Risk → leverage, combined ratio, retention ratio; Liquidity Risk → Cash assets to total assets; Contagion Risk → EDF of insurance sector at 50th percentile. - IOSCO: Credit Risk → EDF of corporate/ banking sectors at 50th percentile; Liquidity Risk → TED spread, equity volatility; Market Risk → turnover freezes, cumulated losses; Contagion Risk → EDF of financial sector at 50th percentile. ### Empirical results — parametric highlights - BCP econometric findings (selected): - Credit risk: increases in lending rates, lower inflation, and lower growth exacerbate credit losses; larger capital inflows associated with more borrowing and higher credit losses to assets. - Market risk (interest rate in banking book): higher interest rates and faster past lending growth associated with higher net interest income to assets; net interest income increases with higher real GDP growth. - Liquidity risk: loans to deposits ratio increases (liquidity decreases) with higher economic growth and intermediation spread; higher lending rates and lower deposit rates deepen intermediation. - Contagion risk: domestic banking sector EDFs increase with global stock market volatility (VIX), higher domestic corporate EDFs, and slower domestic growth. - Overall risk (z-score): higher economic growth increases the z-score (reducing risk); higher bank funding costs and currency depreciation exacerbate overall risk. - ICP econometric findings (selected): - Market/investment risk: default risk declines when long term bond yields or equity returns increase; higher domestic equity volatility associated with higher corporate/financial default risk; stronger domestic growth mitigates default risk. - Sovereign risk: higher GDP growth and improved fiscal balance mitigate sovereign default risks and lower CDS spreads. - Insurance underwriting: higher leverage ratio signals higher risk appetite; combined ratio declines with higher growth; retention ratio declines (more reinsurance) in recessions. - Liquidity: higher long term bond yields reduce insurance liquidity ratio; liquidity ratio declines during economic expansions. - Contagion: insurers vulnerable to distress in domestic banking sector and sovereign due to fixed-income holdings. - IOSCO econometric findings (selected): - Credit risk: driven by country-specific fundamentals and global market factors (VIX, corporate yield spreads, commodity prices); global factors mattered more historically, country-specific factors dominated later windows. - Liquidity risk: increases with domestic inflation and low policy rates; US TED spread exacerbates liquidity risk across countries. - Market risk: cumulated equity losses increase with low growth and high volatility; results robust to alternative measures including market freezes. - Contagion risk: corporate issuer distress, sovereign CDS spikes, and increased equity volatility are key contagion mechanisms; cross-border contagion from foreign intermediaries and domestic money market volatility became more prominent over time. ### Empirical results — non-parametric (binary tree) highlights - Baseline binary tree for credit risk (25 advanced economies, 1998-2012): - 8 terminal nodes; in-sample R-square 0.37; out-of-sample R-square 0.17. - Parent node: interest rate (IRDEP); IRDEP > 13.8 percent → NPL ratio increases from 3.5 percent to 11.9 percent. - Example illustrative splits and node averages provided (terminal Node 8 Avg = 11.9; Node 6 Avg = 2.1). - Variable importance (baseline scaled to Interest rate = 100.00): - Baseline: Interest rate 100.00; Fiscal balance 47.79; Growth 24.51; Inflation 23.39; Current account balance 12.77. - With BCP gradings included, top contributors: - Interest rate 100.00; Fiscal balance 48.85; Inflation 39.08; Growth 13.78; Current account balance 15.64 - BCP principles: CP11 33.33; CP18 30.92; CP20 28.99; CP6 15.40; CP5 13.58; SP12 12.10; CP12 11.27; ... down to CP21 0.03. - Diagnostics (example RMSE/R^2 for NPL ratio): - Baseline: In-sample RMSE 3.22; Out-of-sample RMSE 3.69; In-sample R^2 0.37; Out-of-sample R^2 0.17. - With Gradings: In-sample RMSE 1.98; Out-of-sample RMSE 2.98; In-sample R^2 0.76; Out-of-sample R^2 0.46. ### Principle selection and combined rule - Data reduction and factor identification: - PCA + EFA (Varimax orthogonal rotation); four factors retained; factors explain around 60 percent of variability. - Factor loadings threshold: > 0.50 absolute value and fast decline used to select principles loading on factors. - Some principles fail to load significantly; complemented by principle-by-principle econometric analysis. - Selection rule: - Macrofinancially-relevant principles chosen as the union of principles identified in parametric (factor-based) and non-parametric (individual-principle) analyses. - Outcomes from initial empirical union: 19 BCP principles, 15 ICP principles, and 21 IOSCO principles. - Final set determined by combining empirical results and expert judgment, yielding: - 11 BCPs - 11 ICPs - 17 IOSCO Principles - Categorization by empirical strength: “Strong empirical evidence” (significant across most approaches) vs “Weak empirical evidence” (significant in some approaches). - Some principles excluded from empirical work due to insufficient data (e.g., recently added principles). ### Role of expert judgment and expert-identified priorities - Expert judgment supplements empirical work to capture: - Frequent revisions of standards and principles without historical data. - Contextual supervisory/regulatory experience not captured by econometrics. - Examples of expert rationale (selected): - BCP highlights: - Foundational/gateway: BCPs 1 and 2; BCP5 (licensing). - Corrective powers: BCP11. - Supervisory understanding and intervention: BCP8. - Capital adequacy: BCP16. - Risk governance and specific risks: BCP15, BCP17, BCP18, BCP19, BCP24; BCP14 (corporate governance) noted as new with no historical data. - Disclosure/transparency: BCP28. - ICP highlights: - Legal powers and corrective authority: ICP2. - Licensing: ICP4. - Governance and risk management: ICP7 and ICP8. - Supervisory review and macroprudential surveillance: ICP9, ICP11, ICP24 (macroprudential cornerstone). - IOSCO highlights: - Regulator independence/resources/powers: P1-P3 and P10-P11. - Collective investment schemes and client asset protection/disclosure/valuation: P24-P27. - Issuer disclosures and hedge fund risks: P16, P18, P28. - Market intermediaries regulation and failure procedures: P30-P32, P36-P37. ### Integration into financial stability analysis (Enhanced RAM) - Post-assessment reporting: - Findings on macrofinancially-relevant principles are summarized in an expanded/enhanced RAM and integrated into FSAP overall risk assessment. - Assessors and FSAP teams judge whether supervision/regulation quality: Aggravates, Mitigates, or is Neutral regarding probability and/or impact of specific risks; judgments explicitly included in RAM with brief explanations. - Enhanced RAM explicitly links each macrofinancial shock to likely effects on financial sector/sub-sectors and how supervision/regulation quality affects probability and expected impact. - Illustrative Enhanced RAM scenario (loss of confidence and currency depreciation): - Scenario: Loss of confidence → decline in money demand, capital outflows, currency depreciation. - Quantitative illustrative impacts: - A 30 percent depreciation, with 0.2 pass-through, increases inflation by 6 percentage points. - NPL ratios increase by about ½ percentage point on average. - 4 banks would fall below the required minimum. - Supervisory assessment summary: - Banks: Aggravating. - Securities: Aggravating. - Insurance: Neutral. ### Risks and implementation challenges of the targeted, risk-based approach - Benefits: - Improved focus, resource-efficiency, and effectiveness by concentrating assessment effort on principles most relevant to country risks. - Identified risks: - Incomplete coverage due to misidentification of risks: - Requires thorough understanding of country financial system, sector sizes, business models, and interconnectedness. - Necessitates stronger pre-mission preparation: scoping, data requests, and FSAP-assessor interaction. - Residual risk that important principles may be missed. - Fragmented perspective: - Many principles are interrelated; selective assessment risks losing holistic perspective. - Mitigation option: treat certain principles (independence, resources, powers) as essential, though this may increase number of principles and dilute benefits. - Staff capacity and implementation: - Economists need deeper understanding of non-bank financial sector issues. - Supervisory experts need better understanding of linkages between regulation/supervision and financial stability. - Mitigations: training, more on- and off-site interaction between FSAP teams and assessors, and ensuring efficiency when relying on external experts. ### Sample construction and methodological notes - Final sample used in empirical work: - 85 BCP assessments, 45 ICP assessments, and 77 IOSCO assessments. - Excluded offshore jurisdictions and ROSCs under 1997 BCP and 1999 IAIS methodologies. - Observed proportions under newer methodologies: - Less than 10 percent of assessments under revised 2012 BCP methodology. - Over 30 percent of ICP assessments under new 2011 ICP methodology. - Close to 20 percent of IOSCO assessments under new 2010 IOSCO methodology. - Mapping across methodologies: - Conversion matrix informed by expert judgment; principles 100 percent new excluded from mapping. ### Summary conclusions - The paper proposes focusing FSAP supervisory standards assessments on a defined subset of macrofinancially-relevant principles selected using a union of empirical methods and expert judgment. - Selected principles (final set): 11 BCPs, 11 ICPs, and 17 IOSCO Principles as the starting point for FSAP-focused assessments, with flexibility to include additional principles as warranted by country-specific financial stability considerations. - Assessment findings to be systematically integrated into the FSAP risk assessment via an enhanced RAM, with supervisory quality judged as Aggravating, Mitigating, or Neutral for probability and expected impact of identified risks. *Prepared by an MCM team supervised by Dimitri G Demekas and Michaela Erbenova, led by Sònia Muñoz, and comprising Timo Broszeit, Mario Catalán, Christina Daniel, Eija Holttinen, Fabiana Melo, Katharine Seal, Nobuyasu Sugimoto, and Laura Valderrama; Approved by José Viñals. August 18, 2014.* *Source: IMF. A Macrofinancial Approach to Supervisory Standards Assessments (selected excerpts).* ### EXECUTIVE SUMMARY ### _081814a - EXECUTIVE SUMMARY ### Purpose and scope - Standards assessments in FSAPs are used to identify weaknesses in financial regulation and supervision and provide cross-country perspectives. - This paper focuses on the three supervisory standards most frequently assessed in FSAPs: Basel Core Principles (BCP), Insurance Core Principles (ICP / IAIS), and IOSCO Principles on Securities Regulation (IOSCO). - The proposed approach applies only to supervisory standards assessments undertaken in the context of FSAPs or stability modules; full assessments remain available in other circumstances. ### Key problems identified - Standards assessments are not well integrated into Fund surveillance: - Weaknesses identified in assessments are not specifically linked to the risks and vulnerabilities facing the financial sector. - FSAP analysis of country-specific vulnerabilities does not guide the targeting of standard assessments because formal assessments are exhaustive and must cover the entire standard. - Post-crisis revisions have increased complexity and resource intensity: - Revisions to core principles and methodologies have extended the length of the assessment process and the Detailed Assessment Report (DAR). - Human resource and dollar cost of producing assessments has increased considerably, including additional assessors’ time in the field (for up to three weeks) and more extensive review. - Concerns arise about the Fund’s ability to continue delivering high-quality assessments in the current resource environment and whether resources devoted are proportionate to their net benefit for the FSAP and Fund surveillance. - Formal assessments produce long, detailed DARs/ROSCs whose recommendations are not prioritized by financial stability relevance. - The 2009 “Revised Approach to Regulation and Supervision Standards Assessments” allowed targeted (partial) assessments but selection criteria remained compliance-based and narrow; since 2009 only two targeted assessments occurred: - Targeted BCP for Russia (IMF Report No. 11/336) - Targeted IOSCO for Canada (IMF Report No. 14/73) ### Evidence on assessment frequency (selected figures) - Average Number of ROSCs Conducted During FSAP Missions by Standard (FY08–FY14): - Overall: FY08 0.8; FY09 0.8; FY10 1.4; FY11 2.4; FY12 2.4; FY13 1.8; FY14 1.6 - BCP: FY08 0.5; FY09 0.6; FY10 0.8; FY11 0.9; FY12 0.9; FY13 0.7; FY14 0.9 - IAIS: FY08 0.0; FY09 0.1; FY10 0.2; FY11 0.5; FY12 0.7; FY13 0.5; FY14 0.5 - IOSCO: FY08 0.1; FY09 0.1; FY10 0.1; FY11 0.5; FY12 0.5; FY13 0.5; FY14 0.2 - CPSS: FY08 0.1; FY09 0.1; FY10 0.1; FY11 0.3; FY12 0.1; FY13 0.0; FY14 0.0 - CPSS/IOSCO: FY08 0.0; FY09 0.0; FY10 0.1; FY11 0.3; FY12 0.1; FY13 0.1; FY14 0.0 - MFPT: FY08 0.0; FY09 0.0; FY10 0.1; FY11 0.0; FY12 0.0; FY13 0.1; FY14 0.0 - Total number of ROSCs per fiscal year increased, peaking in FY11–FY12, with BCP, IAIS, and IOSCO accounting for most growth, driven mainly by S-25 countries. ### Proposed new approach (overview) - Objective: (i) focus standards assessments in FSAPs on macrofinancially-relevant principles; and (ii) incorporate findings systematically into the FSAP’s overall risk assessment. - Two core elements: - Focus assessments on macrofinancially-relevant principles: - A subset of principles is selected using empirical analysis and expert judgment, based on macrofinancial risks for banking, insurance, and securities sectors identified as relevant for financial stability. - This replaces the compliance-based selection criteria of the existing “targeted ROSC” model with macro-relevant and risk-based criteria. - On this basis, the proposed approach identifies: - 11 BCP principles as macrofinancially-relevant. - 11 ICP principles as macrofinancially-relevant. - 17 IOSCO principles as macrofinancially-relevant. - Systemically incorporate assessment findings into FSAP risk assessment: - Based on assessment of compliance with the selected principles, FSAP teams will judge whether the quality of regulation and supervision in these areas mitigates or aggravates specific risks. - Findings will be reported systemically into an expanded version of the RAM (Risk Assessment Matrix), guaranteeing full integration of supervisory findings into the risk assessment and facilitating prioritization of policy recommendations. ### Implementation boundaries and continuity - The approach is limited to supervisory standards assessments within FSAPs or stability modules and does not preclude full compliance assessments in other circumstances. - Full assessments would remain appropriate and available when: - There has never been an assessment of the relevant standard. - There have been fundamental changes in the supervisory architecture since the last assessment. - The country requests technical assistance that would benefit from a prior full compliance assessment. - Full assessments would be undertaken on a stand-alone basis or combined with an FSAP if timing allows. ### Empirical and methodological work (as described) - The paper includes: - A case for mapping macrofinancial risks into supervisory standards. - An empirical methodology to identify macrofinancially-relevant principles, including: - Definition and measurement of macrofinancial risks. - Construction of a supervisory standards principles dataset. - Empirical methods (including EFA and PCA references in contents). - Empirical results. - A process for combining empirical results and expert judgment to select principles. - Guidance on integrating assessment results into the financial stability analysis via an expanded RAM. - Annexes and appendices cover technical methodology, merging supervisory standards under different methodologies, and an Enhanced RAM. *Prepared by an MCM team supervised by Dimitri G Demekas and Michaela Erbenova, led by Sònia Muñoz, and comprising Timo Broszeit, Mario Catalán, Christina Daniel, Eija Holttinen, Fabiana Melo, Katharine Seal, Nobuyasu Sugimoto, and Laura Valderrama; Approved by José Viñals. August 18, 2014.* ### 9. The proposed approach could be considered by the Board in the context of the next ### A MACROFINANCIAL APPROACH TO SUPERVISORY STANDARDS ASSESSMENTS ### The Empirical Methodology — overview - The methodology for mapping macrofinancial risks into supervisory standards is a three-stage process: - Stage 1: Identify proxies for macrofinancial risks relevant for the banking, insurance, and securities sectors. - Stage 2: Prepare individual principles in each of the three standards for use in the empirical exercise. - Stage 3: Use parametric and non-parametric econometric approaches to establish links between macrofinancial risks and individual principles (detailed in Appendix II). ### A. Macrofinancial Risks - Definition: Macrofinancial risks originate from or are transmitted to the financial sector in adverse macroeconomic scenarios and can threaten the viability of individual institutions or the system as a whole. - Risk classifications by sector: - Banking sector: credit, market, liquidity, and contagion risks. - Insurance sector: market/investment risks, insurance and underwriting risks, liquidity risks, and contagion risks. - Securities markets: credit risk, liquidity risk, and market risk; contagion risk across domestic markets and across borders is also included. - Data coverage and measures: - Bank and insurance risk measures are constructed from balance sheets and income statements of: - 1,426 individual banks in 83 countries - 848 insurance companies in 23 countries - Securities markets: market sources covering five market categories (money markets, debt markets, equity markets, financial intermediaries, and foreign exchange markets), and market estimates of default rates for securities issuers in the corporate, banking, and non-banking financial sectors for 77 countries. - Global financial market variables and a broad range of estimated risk premia are included. ### B. The Supervisory Standards Principles Dataset - Scoring function: - A linear scoring function assigns numerical values of 1 through 4 to compliance grades (see footnote mapping): - BCP: 1 (compliant), 2 (largely compliant), 3 (materially non-compliant), 4 (non-compliant) - ICP: 1 (observed), 2 (largely observed), 3 (partly observed), 4 (not observed) - IOSCO: 1 (fully implemented), 2 (broadly implemented), 3 (partly implemented), 4 (not implemented) - When a principle is not rated, the proxy used is the average score on the rated principles belonging to the same group of principles. - Proportionality principle: - Establishes a moving hurdle rate so jurisdictions hosting many SIBs face a higher hurdle to obtain a compliant grading. - Concentration and correlation of grades: - Most grades are concentrated in the compliant and largely compliant categories in the BCP standards; more advanced countries generally have better scores across standards. - Many pairwise correlations of principles’ grades exceed 50 percent, indicating limited differentiation across principles. - Unconditional average correlations among principles: - BCP: 34 percent - ICP: 39 percent - IOSCO: 26 percent - Example pairwise correlations: - Revised BCP 22 (market risk) and BCP 23 (interest rate risk in the banking book): 77 percent - Revised ICP 3 (information and confidentiality requirements) and ICP 25 (supervisory cooperation and coordination): 91 percent - Revised IOSCO P25 (segregation and protection of client assets in collective investment schemes) and IOSCO P26 (disclosure in collective investment schemes): 66 percent - Number of principles per standard: - BCBS adopted 29 BCPs in 2012 - IAIS issued 26 ICPs in 2011 - IOSCO issued 38 principles in 2010 (revised in 2011); as a general practice, IOSCO assessments cover only 37 of these 38 Principles - Time-varying assessment methodologies: - Assessment methodologies and application have evolved and become more conservative post-crisis. - Merging different versions of standards (e.g., 1999 BCP methodology with 2006 and 2012) may introduce noise and time-varying factors that are difficult to capture empirically. ### C. The Methodology — empirical strategy - Third-stage empirical steps: - Factor analysis to reduce dimensionality of the standards dataset. - Both parametric and non-parametric approaches to establish empirical links between macrofinancial risks and compliance with individual principles. - Technical details are discussed in Appendix II. - Motivation and past evidence: - Previous studies that aggregate principles into overall compliance indexes (sum or mean) find little or no consistent evidence of a connection between those compliance indexes and financial stability; evidence is mixed and often limited to specific risks, countries, or time periods. - Evidence on IOSCO and ICP effects on financial stability is sparse. - Shortcomings of typical compliance measures in the literature: - Reliance on summary statistics (sum or mean) disregards granular information and prevents identifying principle-level relevance. - Linear scoring assumes equal distance across qualitative grades, which may not hold in practice. - Asymmetric effects and compound weaknesses across principles are typically ignored. - Advantages of the proposed econometric methodology: - Provides a rigorous analytical basis for selecting macrofinancially-relevant principles and controls for different development levels. - Addresses multicollinearity by: - Constructing a battery of regressions to conduct a "horse-race" comparing principles' contributions. - Using factor analysis to reduce explanatory factors while preserving information. - Employing both individual principles and a reduced number of unobserved factors. - Captures non-linear effects: - Evaluates principles with non-parametric as well as parametric econometric techniques. - Considers both direct and indirect effects (interactions with macrofinancial risks). - The non-parametric approach endogenizes principle-specific thresholds of compliance most relevant for mitigating financial risks and shows compound effects from weak compliance across principles. - Expands sample coverage by merging countries assessed under the last two methodologies for each standard using expert-judgment mapping; noise is reduced by focusing on the most recent methodological revisions (Appendix I). ### Literature review and identified pitfalls (Box 1) - Empirical findings are mixed and inconsistent across country groups and time periods: - The 2011 Review of Standards and Codes found BCP compliance significant in explaining financial stress only in 2009 (not 2008) and only for selected advanced economies; IOSCO significance was limited to a sub-index of financial stress in securities markets of advanced economies. - Studies produce varying results on bank stability measures (Z-score, NPLs, net interest margins) and the interaction between compliance and macroeconomic or liquidity factors. - Identified methodological pitfalls in existing studies: - Disregard for distribution of grades and principle-level granularity. - Treatment of categorical grades as having equal distances via linear scoring. - Ignoring asymmetric effects and potential compound effects from combined weaknesses across several principles. *Source: IMF staff estimates and analysis in the chapter “The Empirical Methodology” from the document titled “A Macrofinancial Approach to Supervisory Standards Assessments.”* ### 19. We use an innovative data reduction procedure to choose subsets of principles, while ### 19. We use an innovative data reduction procedure to choose subsets of principles, while ### Data reduction and factor identification - Approach: combination of principal component analysis (PCA) and exploratory factor analysis (EFA) to identify coherent subsets (factors) of principles that are relatively independent of one another (Appendix II). - Rationale: - PCA groups principles that are correlated with one another but largely independent of other subsets. - EFA permits factor rotation to sharpen distinctions in each underlying core principle’s contribution to factors. - Note: Factor analysis addresses the multidimensional nature of standards assessments but may exclude individual principles that do not co-move with others. ### Parametric approach — benchmark model and extensions - Benchmark model (banks, insurers, or markets nested within countries): - Equation (1): ' ,,ijt Y denotes the specific risk analyzed; i = country; j = individual bank/insurance company/market; t = time (year). - Model includes: - bank- or insurance company-specific effects j μ - time-varying controls ,,ijt X - country-specific macro-financial determinants , MACRO it - Benchmark (1) does not incorporate individual principles or factors. - Dynamic vs. static specification: dynamic versions estimated using the Arellano-Bond approach when appropriate. - Benchmark regression results are in Appendix II. - Expanded model to assess compliance effects: - Model (2) includes interaction terms between macro-fundamentals and individual principles or factors F: (2)  ''' ,,,, 1,,,,,, MACROMACRO F. ijtjijtititiijtijt YYX      - F is time-invariant and country-specific. - Model (2) estimated including either single or multiple interaction terms. - Principle/factor selection criteria (from multi-regression specification search): - Sign and significance of the interaction effects (parameter ): - Higher score in a principle or factor indicates a lower degree of compliance. - In regressions with single macro-principle or macro-factor interactions, the risk should be exacerbated by a higher score in the principle/factor in “adverse” macroeconomic scenarios. - Interaction must be statistically significant at the 10 percent level. - Robustness to inclusion of additional interactions: - Sign and significance of the interaction effect must be robust to adding other interaction terms between principles and macro variables. - Ability to improve forecasts of looming financial risks for market data: - For IOSCO, panel regressions estimated using rolling windows on monthly market data. - Principles are ranked in a “horse race” by their capacity to minimize forecasting errors. ### Non-parametric approach — binary tree analysis and splitters - Motivation: capture non-linear interactions and perverse feedback loops highlighted by the global financial crisis; relax parametric assumptions. - Method: - Include both macrofinancial determinants and principles’ grades in the panel specification and let the data reveal the functional form. - Use binary tree analysis to identify combinations of economic factors and supervisory quality (compliance grades) that predict financial risk levels. - At each node, identify the variable and its threshold that best divides observations into high-level and low-level financial risk — the “splitter”. - Construct a score for each variable based on its ability to act as a splitter at every node, weighting more heavily nodes representing adverse macroeconomic conditions. - Advantages: - Captures potential non-linearities between compliance with individual principles and financial stability not possible in parametric estimation. - Identifies specific thresholds of supervisory quality effective in mitigating risk. - Principles with highest splitter scores are those where better (worse) compliance has substantial mitigating (aggravating) impacts on financial stability, especially in stressful macroeconomic conditions. - Illustrative example (Figure 3): - Risk proxy: NPL ratio (credit risk). - Example splits: among observations with high fiscal deficits: - If interest rates are low (below 6.5 percent), compliance with CP12 matters: average NPL ratio is higher by some 5 percentage points when CP12 is graded “Materially Non-Compliant” or “Non Compliant.” - If interest rates are higher than 6.5 percent (though lower than 13.8 percent), compliance with CP13 matters: average NPL ratio goes from 5.3 percent to 8.7 percent when the grade is less than “Compliant.” - Application: repeated across all principles and macro variables; principles assigned scores emphasizing splits in stressed macro states. - Example dataset for binary tree: 25 advanced economies assessed against 2006 and 2012 BCP principles over 1995-2012, total of 319 observations; the illustrated optimal binary tree for credit risk shows the upper part (17 nodes) of a complete tree containing 35 nodes. - Notation used in figure: IRDEP (interest rate of deposits), FB (fiscal balance), INFL (inflation). - Interest rates used in nominal terms as they show more explanatory power than real interest rates. ### Empirical results — union of parametric and non-parametric findings - Selection rule: macrofinancially-relevant principles chosen as the union of principles identified in parametric (factor-based) and non-parametric (individual-principle) analyses. - Rationale: factor analysis reduces dimensionality but may exclude individual principles that do not co-move; inclusion requires significance in both approaches. - Outcomes: - Identified principles: 19 BCP principles, 15 ICP principles, and 21 IOSCO principles. - Selected principles are categorized by strength of empirical evidence: - Strong empirical evidence: statistically significant correlation in all or most parametric and non-parametric approaches. - Weak empirical evidence: statistically significant in some parametric or non-parametric approaches. - Some principles not included in empirical investigation due to insufficient data (e.g., principles recently added to the standard). - Conclusion: a large number of regulatory principles across all three standards are not macrofinancially-relevant. - Association with risks: - Methodology permits linking selected principles with specific macrofinancial risks (Figure 4); many principles are relevant for more than one risk. ### Combining empirical results and expert judgment - Role of expert judgment: - Important to identify individual principles that are particularly important for financial stability given frequent revisions of standards and limits of empirical approaches. - Supervisory/regulatory expert input provides perspectives and contextual experience that econometrics cannot capture. - Though judgmental and partly subjective, expert input is crucial for integrity of the exercise. - Expert-identified subsets: - Experts in the Monetary and Capital Markets Department (MCM) identified subsets of principles deemed relevant for financial stability (not associated with individual risks but for financial stability writ large). - Basel Core Principles (selected expert rationale highlights): - Foundational/gateway principles: BCPs 1 and 2 (foundational) and BCP5 (licensing) — adequacy of resources, legal powers, clarity of objectives, responsibilities, independence, and legal protection are essential for supervisors to act. - Corrective powers: BCP11 examines whether the supervisor acts to forestall emerging issues. - Supervisory understanding and intervention capacity: BCP8 tests whether authorities have comprehensive, detailed, forward-looking understanding of banks and the system and can intervene and resolve banks in an orderly manner. - Capital adequacy: BCP16 examines capital adequacy; especially relevant for advanced economies where internal models are more used; weaknesses in model review/approval can lead to unreliable capital assessments. - Risk governance and specific risks: - BCP15 (risk management process) — overarching. - BCP17 (credit risk), BCP18 (problem assets and provisions), BCP19 (concentration risk), BCP24 (liquidity risk) — deep examinations of management and control of these risks. - BCP14 (corporate governance) — insight into governance and management quality; new principle with no historical assessment data. - Disclosure and transparency: BCP28 — meaningful disclosure reduces information asymmetry and promotes comparability; regulatory disclosure (e.g., Pillar 3) helps market discipline and financial stability. - Insurance Core Principles (selected expert rationale highlights): - Legal powers and corrective authority: ICP2 — adequacy of legal powers for prudential and business supervision; adequacy of resources and independence from undue interference are essential. - Licensing: ICP4 — controls entry of entities that contribute to financial stability and can meet obligations to policyholders. - Governance and risk management: ICP7 and ICP8 — core to checks and balances; lack of governance/controls can lead to macrofinancial consequences. - Supervisory review and macroprudential surveillance: - ICP9 tests supervisory review and reporting mechanisms; risk-based approach to evaluating risk profile, conduct, governance, compliance. - ICP11 addresses enforcement of corrective actions. - ICP24 is cornerstone of macroprudential surveillance for insurance: forward-looking analysis of financial market developments and exogenous factors, processes to assess systemic importance of insurers and adequate supervisory responses. *Source: IMF staff text (excerpts as provided in the content unit).* ### 37. An effective and timely resolution framework is critical for the stability of the ### 37. An effective and timely resolution framework is critical for the stability of the ### Insurance resolution, solvency, and valuation - An effective resolution framework is the primary tool to mitigate contagion risks from troubled insurance company failures where other safety nets—such as liquidity facility of central banks and deposit insurance schemes—do not exist in many jurisdictions (paragraph 37). - ICP12 (on winding-up and exit from the market) stipulates essential elements of insurance resolution: - policyholder priority - timely provision of benefits to policyholders - determination point of resolution - Although there is no common global capital standard for insurers at the present time, the IAIS has been working on: - a basic capital requirement for G-SIIs - higher loss absorbency requirements for internationally active insurance groups (paragraph 38). - Until a global insurance capital standard is in place, solvency-related ICPs (ICP14 to 17) include important requirements for: - valuation of assets and liabilities - investment activities - capital standards for solvency purposes - These standards underpin: - (i) risk identification and measurement - (ii) adequacy of technical provisions - (iii) availability of capital to address all relevant and material risks for legal entity and group-wide solvency assessment (paragraph 38). ### Market conduct and consumer protection - Market conduct is integral to insurance supervision to ensure consumers are treated fairly before entering into a contract and through to the point at which obligations have been satisfied (ICP19) (paragraph 39). - ICP19 objectives: - strengthen trust and consumer confidence in the insurance sector - contribute to overall financial stability - create a level playing field where insurers can compete while maintaining acceptable business practices with respect to fair treatment of customers (paragraph 39). ### Group-wide supervision and cross-border risks - Global interconnectedness and increasing presence of insurance groups and financial conglomerates increase the importance of group-wide supervision for financial stability (paragraph 40). - ICP23 (Group-wide Supervision) applies to both the legal entity and insurance group or financial conglomerate to ensure all relevant group-wide risks impacting insurance entities are addressed appropriately. - Failures of group-wide risk management and oversight have led to distressed insurers, resolution, and winding-up and exit from markets (paragraph 40). ### IOSCO Principles and securities-sector risks - Expert assessments for IOSCO Principles build on ensuring the regulator has sufficient independence, resources and powers, and clearly defined responsibilities (P1-P3 and P10-P11) before assessing supervisory and enforcement responsibilities (P12) (paragraph 41). - Collective investment schemes are important for retail and institutional investors, including banks, insurers, and pension funds; robust regulation and supervision (P24) and stringent requirements on: - client asset protection (P25) - disclosure (P26) - asset valuation (P27) are essential to avoid transmission of financial shocks (paragraph 42). - Issuer disclosures and financial reporting quality (P16 and P18) matter; hedge funds in some jurisdictions can be highly leveraged and interconnected and should be assessed as potential sources of systemic risk (P28). Securities regulator’s ability to monitor, mitigate and manage systemic risk is a key focus (P6) (paragraph 42). - Market intermediaries require robust supervision and prudential requirements (P30-P31) to control risks; procedures for failure and containment of contagion are important (P32 and P37) (paragraph 43). ### Combined empirical and expert approaches: overlap and final set - Two approaches used to identify macrofinancially-relevant principles: - Econometric approach: "context-free", data-driven, principle-by-principle empirical investigation. - Expert approach: relies on assessors’ judgment and experience from compliance assessments (paragraph 44). - Overlap results: - 70-90 percent of the principles identified as relevant by either approach in each of the three standards overlap with those identified by the other approach and are part of the final set (paragraph 44). - The final set is the intersection of empirical and expert approaches; it includes principles: - found macrofinancially-relevant in econometric analysis and considered important by experts - considered important by experts but not included in econometric analysis due to lack of data or the need to be assessed in conjunction with other principles (paragraph 45). - The final list includes: - 11 BCPs - 11 ICPs - 17 IOSCO Principles (paragraph 45). ### Integration into financial stability analysis - Once macrofinancially-relevant principles have been identified and compliance assessment completed, findings are summarized in an expanded version of the RAM, becoming an integral part of the FSAP’s overall risk assessment (final paragraph). *Source: IMF staff estimates as presented in the chapter content.* ### 46. The set of macrofinancially-relevant principles in each standard should be seen as the ### The set of macrofinancially-relevant principles in each standard should be seen as the starting point for the assessment process in individual FSAPs ### Selection of macrofinancially-relevant principles and assessment scope - The set of macrofinancially-relevant principles in each standard is a starting point and a guide for focusing assessment effort; it is not a hard rule. - If additional principles are necessary from a financial stability point of view for a particular country FSAP, those principles should be included. - Selection must be guided by financial stability considerations whether the assessment is: - A formal compliance assessment (summarized in a DAR/ROSC), or - An informal assessment (summarized in an FSAP Technical Note). ### Integration with the RAM (risk assessment matrix) - Results of supervisory assessments in FSAPs will be integrated into the risk assessment through an “enhanced” RAM. - On the basis of (formal or informal) standards assessment, assessors and the FSAP team judge whether the overall quality of supervision and regulation: - Aggravates, mitigates, or is neutral with respect to the probability of realization and/or impact of specific risks. - The assessment of how supervision/regulation quality affects “probability of realization” and “expected impact” will be explicitly included in the RAM and summarized (e.g., “Aggravating”, “Mitigating”, or “Neutral”) with brief explanation. - The enhanced RAM structure will explicitly include: - How each macrofinancial shock identified in the RAM is likely to affect the financial sector or relevant sub-sectors (banking, insurance, securities market). - The team’s assessment of how supervision and regulation quality affects probability and expected impact. ### Risks and implementation challenges of a targeted, risk-based approach - Benefits of focusing assessment effort on a subset of principles most relevant to country risks: - Improved focus, resource-efficiency, and effectiveness of assessment. - Identified risks: - Incomplete coverage due to incomplete identification of risks: - Teams must have thorough understanding of the country’s financial system, sector sizes, business models, and interconnectedness. - Requires substantial strengthening of pre-mission preparation: scoping, data requests, interaction between FSAP team and assessors. - Residual risk that important principles may be “missed” remains. - Fragmented view of the role of regulation and supervision in financial stability: - Many principles are interrelated; selective assessment risks losing holistic perspective. - One mitigation option: treat certain principles (independence, resources, powers) as essential for any risk mapping — but this may dilute benefits by increasing the number of principles. - Implementation risks for staff: - Economists need better understanding of financial sector issues, especially non-bank sectors. - Experts need better understanding of linkages between regulation/supervision and financial stability risks and vulnerabilities. - Mitigations: training and more on- and off-site interaction time between FSAP team and assessors; ensure efficiency where external experts are relied upon. ### Merging supervisory standards under different methodologies (sample and benchmark choices) - Sample construction and exclusions: - Final sample includes 85 BCP assessments, 45 ICP assessments, and 77 IOSCO assessments. - Assessments exclude offshore jurisdictions due to lack of macroeconomic data. - Excluded ROSCs conducted under the 1997 BCP methodology and the 1999 IAIS methodology. - Benchmark methodology choices guided by: - Relative country coverage under methodologies. - Specifics of the conversion matrix: preference for methodologies that reduce noise when inputs merge rather than expand. - Observed proportions under newer methodologies: - Less than 10 percent of assessments were conducted under the revised 2012 BCP methodology. - Over 30 percent of ICP assessments were carried out under the new 2011 ICP methodology. - Close to 20 percent of IOSCO assessments were performed under the new 2010 IOSCO methodology. - Note on mapping: - The conversion matrix was informed by expert judgment. - Principles that are 100 percent new were excluded from the mapping matrix. ### Factor analysis methodology for principle selection and dimensionality reduction - Purpose: - Reduce number of principles and address correlation across grades to enable robust econometric analysis. - Three-pronged approach: - Combine principal component analysis (PCA) and exploratory factor analysis (EFA) to reduce dimensionality and orthogonalize the dataset. - Rotate factors to sharpen distinctions and interpret factor meanings via loadings. - Estimate factor scores for each country using the Anderson method; factor scores computed as linear combinations of centered data for each principle and used as regressors. - Technical details and criteria: - PCA analyzes all variance (common and unique) but does not allow rotation. - EFA accounts only for variance shared with other principles and allows rotation. - Two-stage procedure: use PCA to generate a matrix exploring patterns; use that matrix to perform EFA allowing rotation. - Rotation method: Varimax orthogonal rotation (angle between factors maintained at 90 degrees). Results robust to other orthogonal rotations. - Criteria for number of factors: minimum average partial (MAP) method, “eigenvalues greater than one” rule, and the scree test. - Factor retention result: four underlying factors selected. - The four factors explain around 60 percent of total variability for the three standards. - Identification of significant loadings uses criteria including: - Factor loadings should be greater than 0.50 in absolute value. - The rate of change in relative magnitude of loadings should reflect fast decline at the point the last principle is selected. - Some principles fail to load significantly on any factor; factor analysis is complemented with a principle-by-principle approach in econometric analysis. - Estimation properties and validation: - Multiple correlations for all four factors are around 0.90. - Validity coefficients are close to 0.9 for all four factors — above thresholds advocated for using estimated scores as replacements for original variables. ### Factor analysis results — interpretation of the four factors for each standard - BCP four underlying factors and associated principles: - ‘Risk management’: CP7, CP12, CP13, CP14, and CP15 - ‘Legal’: SP11, SP13, and SP14 - ‘Information sharing’: SP16, CP3, CP24, and CP25 - ‘Asset quality’: SP15, CP8, CP10, and CP22 - ICP four underlying categories and associated principles: - ‘Governance’: ICP5, ICP8, ICP16, and ICP19 - ‘Information sharing’: ICP3, ICP23, and ICP25 - ‘Regulation’: ICP4, ICP10, and ICP15 - ‘Prudential and Supervision’: ICP9, ICP14, ICP17, and ICP24 - IOSCO four categories and associated principles: - ‘Information sharing’: P1, P13, P14, P15, and P34 - ‘Collective investment schemes’: P24, P25, P26, and P27 - ‘Market intermediaries’: P30, P31, P32, and P36 - ‘Issuers’: P5, P16, P17, and P18 *Source: _081814a - The set of macrofinancially-relevant principles in each standard should be seen as the starting point for the assessment process in individual FSAPs* ### 58. All models for BCP and ICP are estimated using annual data. For BCP we use firm-level ### _081814a - 58. All models for BCP and ICP are estimated using annual data. For BCP we use firm-level ### Data and estimation approach - All models for BCP and ICP are estimated using annual data. - BCP data sources: - Firm-level data from Bankscope. - Aggregate country data from IMF FSI statistics. - ICP data sources: - Company-level data from the SNL database. - OECD and EIOPA data for aggregate country variables. - IOSCO models are estimated using monthly data. - Financial indicators constructed from Bloomberg, Datastream, Consensus Forecasts, WEO, and Haver Analytics. - Benchmark model specification: - Includes macro-financial determinants, fixed effects, and time-varying controls. - Dynamic versions estimated using the Arellano-Bond approach when appropriate. - Expanded to assess effects of compliance with core principles or factors via interaction terms between macro-fundamentals and individual principles/factors. - Principles/factors selected when interaction coefficient is significant and has the expected sign given the dependent-variable interpretation of vulnerability. - Non-parametric approach illustrated by a benchmark binary tree for credit risk in the banking sector. - Rolling-window panel regression for IOSCO over the sample 2000m1-2013m12 on a panel of 77 countries; focus on 1 month ahead forecasts and conditioning the dependent variable on observed realizations of regressors. ### Risk definitions (Appendix II. Table 1) - BCP (Type of Risk and proxy dependent variables): - Credit Risk: Loan losses to assets ratio / NPL ratio - Market Risk: - Interest rate risk in the banking book: Net interest income to assets ratio - Market risk from securities: Net gains or losses from securities to assets ratio / Non-net interest income to assets ratio - Liquidity Risk: Loans to deposits ratio - Contagion Risk: Banking sector EDF - Overall Risk: Z-score - ICP: - Market/Investment Risk: - Issuer Risk: Non-government bonds to asset ratio * EDF of corporate and financial sector at 50th percentile - Sovereign Risk: Government bonds to asset ratio * CDS of sovereign - Equity Risk: Equity to assets ratio * PTB ratio - Overall Market Risk: Net income to assets ratio - Insurance and Underwriting Risk: - Leverage Ratio: Net premiums written to total provisions - Combined Ratio: Net claims and administrative exposures to net provisions written - Retention Ratio: Net premiums written to gross premium written - Liquidity Risk: Cash assets to total assets - Contagion Risk: EDF of insurance sector at the 50th percentile - IOSCO: - Credit Risk: - Non-financial: EDF of corporate sector at the 50th percentile - Financial: EDF of banking sector at the 50th percentile - Liquidity Risk: - Funding Cost: TED spread - Market Value: Equity volatility - Market Risk: - Freezes in turnover: Change in turnover value of traded equity securities - Cumulated losses: Inverse of the CMAX measure - Contagion Risk: EDF of financial sector at the 50th percentile - Note: EDF = implied expected default frequency from Moody's KMV database. CMAX measure calculated as x /max[x (x | j 0,1, T)] t where x denotes the MSCI index closing price and T = 24 for monthly data. ### Econometric results for BCP (Appendix II. Table 2) - Credit risk: - Increases in lending rates, lower inflation, and lower growth rates exacerbate credit losses. - Higher nominal lending rates and lower inflation increase ex-post real interest rates; lower economic growth increases unemployment. - Larger capital inflows are associated with more borrowing and a higher ratio of credit losses to total assets. - Market risk: - Interest rate risk in the banking book: - Higher interest rates and faster past lending growth are associated with a higher ratio of net interest income to assets. - Net interest income increases in response to higher real GDP growth rates. - Market risk from securities (including sovereign): - Higher lending rates reduce net income from securities relative to assets. - Higher long-term government bond yields are associated with higher net income from securities relative to assets. - Net income from securities to assets ratio tends to increase in periods of high economic growth and large current account balances. - Liquidity risk: - Measured by loans to deposits ratio; as the loans to deposits ratio increases, liquidity decreases. - Banks hold less liquid assets and increase lending when economic growth and the intermediation spread are higher. - Higher lending rates and lower deposit rates deepen intermediation, increasing the loans to deposit ratio and reducing liquidity. - Contagion risk: - Changes in EDFs corresponding to global banks are transmitted into EDFs of domestic banks, controlling for other determinants. - Domestic banking sector EDFs increase with global stock market volatility (VIX), higher domestic corporate sector EDFs, and slower domestic economic growth. - Contagion risk analysis used country-specific monthly data; real GDP growth expectations captured by monthly consensus forecasts. - Overall risk: - Overall risk is inversely related to the z-score measure. - Higher economic growth increases the z-score, reducing overall risk. - An increase in bank funding costs (measured by the deposit rate) and a depreciation of the domestic currency exacerbate overall risk. - Overall risk analyzed using country-aggregate data. - Note on z-score: measures distance to bankruptcy incorporating capitalization, profitability, and profit variability; based on past variability of profits it measures how many standard deviations future profits have to fall to deplete all the capital of a bank. ### Econometric results for ICP (Appendix II. Table 3) - Context: - Insurance companies collect premiums and invest in cash, bonds, equity, real estate, and other assets; liabilities mainly technical provisions. Main solvency risk is a decline in asset values or insufficiency of technical provisions. - Market/investment risk: - Main solvency risk is decline in value of assets, with bond holdings usually the largest investment; risks associated with sudden changes in bond yields and credit risks in sovereign, financial, and non-financial securities. - Issuer risk: - Default risk declines when either long term bond yields or equity returns increase. - For a given domestic equity return, an increase in international stock returns increases corporate/financial default risk. - Higher domestic equity volatility is associated with higher corporate/financial default risk. - An increase in VIX—holding domestic volatility constant—reduces domestic corporate/financial default risk. - Stronger domestic growth mitigates default risk. - Sovereign risk: - Higher domestic equity return reduces probability of sovereign default. - Higher GDP growth and strengthening of the fiscal balance mitigate sovereign default risks and are associated with lower CDS spreads. - Higher domestic equity volatility and a lower VIX also increase sovereign risk. - Equity risk: - Lower price-to-book (PTB) values associate with lower future earnings growth and higher risk. - Higher GDP growth and return on equity lead to higher market-to-book values, higher future earnings, and lower risk (and expected return). - Overall market risk: - Insurance companies benefit from an increase in interest rates (higher interest income on bonds held to maturity; actuarial value of future claims declines). - Overall profitability increases with higher stock market returns and stronger GDP growth. - Insurance and underwriting risks: - Leverage ratio: - Increases often signal higher risk appetite; net premium written outpaces growth of technical provisions. - Variations are cyclically associated with business cycle; net premium written rises during economic expansions. - Combined ratio: - An increase reduces profitability; higher growth associated with a lower combined ratio. - Other determinants: size (total assets), past profitability (net income ratio), and long term bond yields. - Retention ratio: - Declines in retention (more reinsurance) signal distress; declines tend to occur during recessions. - Larger and more capitalized insurers tend to retain a larger share of written premium. - Liquidity risk: - Considered less important than in banks because insurers do not perform liquidity transformation. - Higher long term bond yields reduce the liquidity ratio (opportunity cost of holding cash increases). - During economic expansions (higher GDP growth and equity returns), liquidity ratio declines as cost of holding liquid assets increases. - Contagion risk: - Distress propagation examined from key issuers of fixed income including banking sector, corporate sector, and sovereign. - Insurance firms vulnerable to distress in domestic banking sector and own sovereign due to large holdings of fixed income from these entities. - Some evidence of contagion from global insurance firms has subsided over time. - Evidence of flight-to-quality effects at the height of the financial crisis from distress in the global banking sector into the insurance sector. ### Econometric results for IOSCO (Appendix II. Table 4) - Purpose: - Securities regulators aim to enable market participants to manage and price risk appropriately; chosen indicators: credit risk (issuers’ default risk), liquidity risk (liquidity squeeze), market risk (distortions in secondary markets), and contagion effects. - Estimation setup: - Panel regression using monthly data, sample 2000m1-2013m12, panel of 77 countries with rolling windows; focus on 1 month ahead forecasts. - Credit risk: - Regressed on local economic variables (growth, inflation, monetary policy, sovereign stress), market liquidity (volatility in money markets), global risk premium measures (VIX, change in investment grade corporate yield spreads, term premium), and global economic variables (change in commodity prices, global growth outlook). - Credit risk for corporates driven by combination of country-specific fundamentals and global market factors. - Recursive regressions show weights of determinants changed over time: global factors more significant in early windows; country-specific factors dominate in later windows. - Global factors exacerbating credit risk: realized volatility (VIX), corporate yield spreads in investment-grade segment, commodity prices (Standard and Poor’s Goldman Sachs Commodity Index). - Credit risk eases with growth in major emerging market economies. - Country-specific drivers: deceleration of economic growth, higher sovereign stress, accommodative monetary policy prompting higher leverage. - Results robust across different percentiles of credit distress. - Credit risk for financial intermediaries rises mainly with sovereign CDS spreads and corporate stress. - Liquidity risk: - Measured as interest rate spread between the 3-month interbank rate and the 3-month T-bill at monthly frequency. - Regressed on local economic variables (growth, inflation, monetary policy stance, distress in financial sector, policy rate, sovereign CDS spreads) and global liquidity variables (US liquidity spread, VIX, itraxx). - Relevance of country-specific determinants increased over time as global liquidity strains eased. - Liquidity risk increases with domestic inflation. - Low policy rates associated with heightened liquidity risk. - Slowdown of GDP growth exacerbates liquidity risk although not significant. - Higher EDF in overall financial sector and a tight monetary policy (deviation from Taylor rule) tend to increase liquidity risk but effects not statistically significant. - A rise in sovereign CDS spreads lowers the interest rate spread between interbank rate and T-bill rate. - Global factors: US TED spread exacerbates liquidity risk across countries; VIX and itraxx found insignificant for this measure. - Determinants of equity volatility: exacerbated by higher inflation, rising VIX, and spikes in equity premia. - Market risk: - Cumulated equity losses increase with low economic growth and high inflation (inflation effect not statistically significant). - Equity losses rise with volatility risk premium (VIX) and equity premium proxied by monthly changes in price-to-earnings ratio for world MSCI index. - Cumulated equity losses decrease amid buoyant world equity markets but are exacerbated by the US stock market excess return (monthly value-weighted return on all NYSE, AMEX, and NASDAQ stocks minus the one-month T-bill return) — supporting a flight-to-quality effect into US stocks. - Results robust to alternative measures including market freezes proxied by a decline in turnover. - Contagion risk: - Two contagion channels considered: (i) contagion across domestic financial system (corporate issuers, sovereign issuers, money markets, equity markets), and (ii) cross-border contagion from stress in foreign financial intermediaries weighted by GDP. - Contagion from corporate issuers is the most significant channel squeezing balance sheets of financial intermediaries. - Spike in sovereign default risk and increased equity market volatility are key contagion mechanisms propelling systemic risk. - These contagion channels remained broadly stable over the forecast horizon. - Contagion from volatility in domestic money markets raising funding costs and cross-border contagion from foreign intermediaries exacerbating counterparty risk have become more prominent over time. - Among control variables, only inflation has become increasingly significant over time; slowdown in economic growth shows negative sign but not statistically significant. *Italic source attribution: IMF. A Macrofinancial Approach to Supervisory Standards Assessments (selected excerpts).* ### Appendix II.Table 2. BCP Be ### Appendix II.Table 2. BCP Benchmark Estimation Results ### Parametric estimation — key coefficients and significance - Dependent risk columns shown: Interest Rate Risk in the Banking Book; Market Risk from Securities; Overall Risk; Market Risk; Credit Risk; Liquidity Risk; Contagion Risk (table shows coefficients by risk). - Selected coefficients (value and significance): - Loan losses to assets ratio: 0.29100 *** - Bank lending interest rate: 4.79500 *** ; 2.61100 *** - Inflation rate: -0.03170 ; -0.02470 * ; -0.03370 *** - Real GDP growth rate: -0.03310 ** ; 0.00287 ; 0.01550 ** ; 0.00311 ** ; 0.00600 * - GDP per capita: 0.00014 *** - Fiscal balance to GDP ratio: 0.05030 *** - Current account balance to GDP ratio: -0.04520 *** ; 0.04330 *** - Bank deposit rate: 3.11800 ** - Bank rate spread: 3.10500 ** - Bank operating costs to assets ratio: 5.71800 * - Bank credit growth rate: 0.00198 ** - Bank non-performing loan ratio: -1.90300 ** - Long term government bond yield: 0.06740 *** - Real GDP growth rate squared: -0.00313 *** - Loans to deposits ratio: 0.83200 *** - Real bank lending interest rate: 0.14900 ** - Real bank deposit interest rate: -0.13800 ** - Real GDP growth forecast: -0.04060 * - Global banking sector EDF: 0.54100 *** - VIX: 0.00255 * - Corporate sector EDF: 0.44700 ** - Deposit rate: -0.00560 * - Real exchange rate depreciation: -0.00280 * - Significance legend: ***p<0.01, **p<0.05, *p<0.1 ### Estimation methods (notes) - Estimation methods used under each risk: - Two-step Arellano-Bond for credit risk and liquidity risk. - Fixed effects for interest rate risk in the banking book, market risk from securities, contagion risk and overall risk. ### Cross-sector (insurance/ICP) and IOSCO benchmark highlights - Appendix II.Table 3 (ICP benchmark) selected coefficients and significance: - Treasury bill rate: 0.05670 *** - Treasury bill rate (one year lag): 0.00210 *** - Total return on equity: -0.00200 *** ; -0.39600 *** ; 0.01450 *** - SP 500 total return: 0.00120 *** ; -0.13600 *** ; -0.00220 *** ; 0.00030 *** - Equity return volatility: 0.12400 *** 10.3 7000 ** - Real GDP growth rate: -0.00400 *** ; -1.37000 *** ; 0.01740 *** ; 0.00200 *** ; 0.00770 *** ; -0.8080 *** - Real exchange rate depreciation: 0.51200 *** - Long term bond yield: -0.0640 *** - Fiscal balance to GDP ratio: -2.02900 *** - Equity to assets ratio: 0.31700 ** - EDF banking sector at 50th: 0.40300 ** ; 0.35600 ** - Spread investment grade: -19.1600 *** - VIX: -0.00780 *** ; -0.92300 *** - Appendix II.Table 4 (IOSCO benchmark) recursive forecasts — selected coefficients: - Growth: values include -0.00964 ; -0.152** ; -0.0138 ; -0.0560 ; -0.0729 ; -0.0249 ; 0.0162* ; -0.00320 ; 0.0149*** ; 0.0192*** ; -0.00403 ; -0.00731 - Inflation: values include -0.0101 ; 0.0731 ; -0.0379 ; -0.0322 ; 0.168** ; 0.237** ; 0.0183*** ; 0.0181*** ; 0.000911 ; -0.000788 ; -0.0504 ; -0.0684* - VIX: 0.00872*** ; 0.00724 ; 0.00474 ; 0.00337 ; 0.00846* ; 0.00376 ; 0.0324*** ; 0.0310*** ; -0.0117*** ; -0.0108*** ; -0.540*** ; -0.337*** - World MSCI equity return: 0.0362*** ; 0.0314*** ; 0.0196*** ; 0.00819*** ; 4.176*** ; 1.846*** ; 0.00760** ; 0.00397 - US equity excess return: -0.0776*** ; -0.0704*** ; -0.0281*** ; -0.0174*** ; -6.178*** ; -2.985*** - Observations and fit statistics shown (example rows from Table 4): - Observations: 1,882 ; 3,863 ; 1,569 ; 3,401 ; 1,761 ; 3,583 ; 4,610 ; 7,417 ; 4,607 ; 7,413 ; 1,421 ; 2,904 ; 1,609 ; 3,420 - R-squared: 0.374 ; 0.165 ; 0.418 ; 0.496 ; 0.755 ; 1.000 ; 0.459 ; 0.412 ; 0.552 ; 0.426 ; 0.172 ; 0.054 ; 0.441 ; 0.543 ### Notes on IOSCO recursive estimation setup - The estimation method is recursive 1-period ahead forecasts on monthly data; the first rolling window spans 2000m1 to the last observation in the recursive window. - Monetary policy stance measured as deviation of the policy rate from the Taylor rule; Taylor (2010) weights/parameters used: 1 percent for real interest rate, 0.5 for growth, and 1.5 for inflation. A positive sign shows too lax MP, a negative sign shows too tight MP. - Term premium based on Cochrane-Piazzesi (2005). - Change in corporate credit yields: monthly changes in basis-point yield spread between BBB and AAA industrial bond indexes (Bloomberg fair market curve). - Cross-border EDF banking sector at the 50th percentile: average implied EDF for other countries weighted by GDP, excluding the specific country. - Liquidity spread: difference between the US 3m repo rate and the US 3m T-bill rate. - World MSCI equity premium: monthly changes in price-earnings ratio for MSCI world index. - US equity excess return: monthly value-weighted return on all NYSE, AMEX, and NASDAQ stocks minus the one-month Treasury-bill return. --- ### Non-parametric (binary tree) analysis — credit risk (BCP assessments) ### Methodological description - Binary tree (classification) analysis identifies threshold values of candidate predictors that best split the sample into high and low risk sets. - For credit risk, algorithm computes at every threshold of each regressor the average loan loss provisioning ratio at each child node recursively until single-observation terminal nodes. - Compliance with Core Principles is coded on a four-grade scale with scoring rule: ‘compliant’ (1), ‘largely compliant’ (2), ‘materially non-compliant’ (3), and ‘non-compliant’ (4). - Sample exercises performed for all countries and subsets by income level (advanced, emerging, LICs). - Determinants included in non-parametric regressions are the same as parametric regressions (Appendix II. Table 2-4). ### Baseline binary tree findings for credit risk (25 advanced economies, 1998-2012) - Baseline tree details: - 8 terminal nodes. - In-sample R-square: 0.37. - Out-of-sample R-square: 0.17. - Parent node (first split): interest rate (IRDEP). - If IRDEP > 13.8 percent, the NPL ratio increases from 3.5 percent to 11.9 percent. - Example node statistics (from Appendix II. Figure 1): - Node 1: STD = 4.1 ; Avg = 3.6 ; N = 319 - Node 2: STD = 3.6 ; Avg = 3.5 ; N = 314 - Node 3: STD = 4.6 ; Avg = 5.3 ; N = 123 - Terminal Node 8: STD = 13.1 ; Avg = 11.9 ; N = 5 - Terminal Node 6: STD = 1.7 ; Avg = 2.1 ; N = 181 - Splits include thresholds: IRDEP>13.8 ; IRDEP<=13.8 ; FB>-2.9 ; FB<=-2.9 ; IRDEP<=4.4 ; INFL<=3.8 ; GROWTH<=2.4 ; FB<=-13.0 ; IRDEP<=6.5 ; IRDEP>6.5 ; IRDEP>4.4 ; INFL>3.8 ; GROWTH>2.4 ; FB>-13.0 ### Variable importance for credit risk (baseline vs with BCP gradings) - Baseline relative importance scores (scaled to Interest rate = 100.00): - Interest rate: 100.00 - Fiscal balance: 47.79 - Growth: 24.51 - Inflation: 23.39 - Current account balance: 12.77 - With gradings against BCP principles, top contributors (scores shown where provided): - Interest rate: 100.00 - Fiscal balance: 48.85 - Inflation: 39.08 - Growth: 13.78 - Current account balance: 15.64 - CP11 (exposures to related parties): 33.33 - CP18 (abuse of financial services): 30.92 - CP20 (supervisory techniques): 28.99 - CP6: 15.40 - CP5: 13.58 - SP12: 12.10 - CP12: 11.27 - (Additional CP/SP items listed with scores down to CP21: 0.03) - Interpretation: - Under baseline, best predictors: interest rate, fiscal balance, growth, inflation, current account balance. - When BCP compliance grades included: interest rate, fiscal balance, inflation remain main determinants; compliance against CP11, CP18, CP20 outperform growth and current account balance. - These rankings guide which regulatory/supervisory principles are most relevant to mitigate credit risk conditional on macroeconomic environment. ### Binary tree model diagnostics (Appendix II. Table 5 sample metrics) - Baseline vs With Gradings (examples): - In-sample RMSE: 3.22 ; Out-of-sample RMSE: 3.69 (Baseline) - In-sample RMSE: 1.98 ; Out-of-sample RMSE: 2.98 (With Gradings) - In-sample R^2: 0.37 ; Out-of-sample R^2: 0.17 (Baseline) - In-sample R^2: 0.76 ; Out-of-sample R^2: 0.46 (With Gradings) - Note: Credit risk measured by NPL ratio. Sample covers 25 advanced economies over 1998-2012. --- ### Appendix III. Enhanced RAM — illustrative scenario (loss of confidence and currency depreciation) - Overall scenario: Loss of confidence → decline in money demand, capital outflows, currency depreciation. - Nature/source of main threats: - Banks: liquidity shock via deposit withdrawals; solvency pressures via (i) declining net interest margins as deposit rates increase faster than lending rates (loan-deposit maturity gap); (ii) increases in NPLs; (iii) declines in value of locally-denominated, long-duration nominal bonds. - Securities dealers: liquidity squeeze and solvency pressures as locally-denominated repos may not roll over. - Insurance companies (esp. life): affected by decline in locally-denominated bond prices; potential large policy withdrawals causing solvency and liquidity problems. - Risk assessment: Medium - Likelihood of severe realization in next 1–3 years: Medium - Expected impact on financial stability if realized: Medium - Quantitative illustrative impacts: - A 30 percent depreciation, partially transmitted to domestic prices, would increase inflation by 6 percentage points—under a 0.2 pass-through assumption. - NPL ratios would increase by about ½ percentage point on average due to credit, market, and sovereign risks, with variation across banks. - 4 banks would fall below the required minimum. - Supervisory assessment summary: - Banks: Aggravating. - Securities: Aggravating. - Insurance: Neutral. --- *Appendix II.Table 2. BCP Benchmark Estimation Results — content unit from the provided IMF document* --- _Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/np/pp/eng/2014/_081814a.pdf_