## _cr06216

## Source details

**Canonical URL:** [_cr06216](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/scr/2006/_cr06216.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/scr/2006/_cr06216.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/scr/2006/_cr06216.pdf.json)

---

### Overview
- Purpose: Assess resilience of Spain’s banking and insurance systems through stress test exercises carried out in the context of the FSAP.
- Coverage: Consolidated group level, covering the most important risk exposures; shocks applied to domestic business, overseas business, or both.
- Reference date for financial information: end-2004.
- Types of tests:
  - Sensitivity analysis: one-off shocks to individual or multiple risk factors.
  - Scenario analysis: simultaneous, internally consistent changes in a group of macroeconomic variables over a two-year window (models developed by Banco de España).
- Four macroeconomic scenarios considered:
  - Permanent increase in oil prices to US$80 per barrel.
  - Fall in real house prices in Spain and the United Kingdom (Spain: cumulative fall of 30 percent over five years; translated into decreases of 17.3 percent in the first year and 4.6 percent in the second; United Kingdom: 37 percent fall translated into decreases of 20 percent and 8 percent in first and second years, respectively).
  - 30-percent depreciation of the U.S. dollar against other currencies.
  - Crisis in Latin America.
- Exercise participants:
  - Bottom-up: 7 largest financial groups, encompassing 37 credit institutions and roughly two-thirds of system assets (Total Sample: 66.9 percent of system assets; 65.9 percent of system liabilities).
  - Top-down: Complementary exercises by authorities (BE) and mission covering the entire banking system using supervisory data and off-site supervisory models.
  - Insurance: Sample of 27 insurers, covering 50 percent of the non-life market and 62 percent of the life market; parallel stress tests where applicable and results aggregated with parent credit institutions for cross-sector impact.

### Banking system — Key findings and conclusions
- Overall resilience: Banking system able to absorb losses associated with the shocks considered without systemic distress.
- Contributing factors:
  - Strong asset quality for the vast majority of credit institutions.
  - Comfortable loan-loss cushion induced by the dynamic provisioning system.
  - Sound risk management practices by systemically important institutions.
  - Supportive macroeconomic environment.
- Equity price risk (banking book, not-marked-to-market holdings):
  - Sample: 19 credit institutions representing roughly 80 percent of system assets.
  - A 30 percent across-the-board drop in equity prices would generate losses equivalent to 8.6 percent of regulatory capital (14.3 percent for the sampled cajas).
  - Large unrealized gains amounted to 12 percent of regulatory capital (24 percent for cajas), providing a substantial margin.
- Market risk in trading portfolios: Minimal due to relatively small size and active use of hedging strategies.
- Liquidity risk:
  - On average, the seven participating groups would be able to cope with a withdrawal of 39 percent of their maturing time deposits within a month, plus 11 percent of demand deposits, without cutting back normal operations, even if they lose access to interbank market financing.
  - Some institutions rely heavily on issuance of covered bonds in contingency liquidity plans, posing potential systemic risk if market sentiment/liquidity deteriorates.
- Credit risk (sensitivity analyses):
  - Shock applied: Increase of PDs of mortgage and commercial loans by 200 and 100 percent, respectively, from end-2004 levels.
  - Participating institutions’ estimates: average loss of 1.5 percent of Basel regulatory capital (in line with BE estimates for entire system).
  - Mission robustness check (similar shock to PDs with more conservative LGD): average losses of about 5 percent of Basel regulatory capital.
- Severe credit deterioration scenario:
  - Shock: Linear increase in PDs over three years to their average levels observed during 1993–2004 plus two standard deviations.
  - Estimated average losses equivalent to 13 percent of Basel regulatory capital, maximum of 27 percent.
  - Resulted in four credit institutions (representing 5 percent of system assets) going slightly below 8 percent Basel CAR.
- Dynamic panel + scenario results:
  - Fall in house prices: average losses equivalent to 11 percent of Basel CAR, maximum effect of 27 percent; effects larger for cajas (five institutions accounting for less than 3 percent of assets going slightly below 8 percent Basel CAR).
  - Oil price shock: average losses slightly less than 7 percent of regulatory capital.
  - Dollar depreciation: average losses of 5 percent of regulatory capital.
- Participating institutions vs BE top-down:
  - Participating credit institutions: largest losses associated with dollar depreciation, average impact of 5 percent of regulatory capital; fall in house prices in Spain and the United Kingdom represented slightly more than 2 percent of regulatory capital; crisis in Latin America below 1 percent.
  - BE exercises produced same relative ordering but even lower estimates.

### Market, trading and interest-rate risks (selected quantitative outcomes)
- Combined market/trading shock (specified for trading portfolios):
  - Interest rates: increase in USD and EUR: 30 b.p. short end and 100 b.p. long end; GBP: 100 b.p. short end and 30 b.p. long end.
  - Equity prices: across-the-board drop -30 percent.
  - Combined shocks chosen to maximize individual effects and be conservative.
- Trading portfolios:
  - Average trading portfolios roughly equivalent to one-third of regulatory capital and equivalent to 6.4 percent of credit portfolios.
  - Combined shocks produced average CAR change of -0.07 and worst of -1.35 (CAR before shock 12.67; CAR worst 9.99).
  - Example market-risk sensitivity results (average change in CAR; worst; CAR before shock; CAR worst):
    - Equity price risk (trading book, -30 percent): average change -0.04; worst -1.30; CAR before shock 12.68; CAR worst 9.99.
    - Increase in corporate spreads (15 b.p.; 300 b.p.): average change -0.09; worst -0.32; CAR before shock 12.67; CAR worst 10.12.
- Interest rate risk on structural positions:
  - Shocks: parallel shift in yield curve of 200 b.p.; slope increases of 0, 50, and 100 b.p. at 1-day, 3-year, 10-year maturities.
  - Example results (average CAR change; worst; CAR before shock; CAR worst):
    - Parallel increase in euro yield curve 200 b.p.: average CAR change 2.10; worst -3.30; CAR before shock 12.94; CAR worst 9.76.
    - Parallel decrease in euro yield curve -200 b.p.: average CAR change -2.80; worst -6.20; CAR before shock 12.32; CAR worst 9.38.
  - A 200 basis point parallel drop in the euro yield curve would cause average loss equivalent to 2.8 percent of Tier I and Tier II capital; maximum decline 6.2 percent.

### Equity risk in the banking book (detailed)
- Sample: 19 credit institutions (7 banks and 12 cajas), jointly accounting for 80 percent of system assets at end-2004.
- Average equity holdings at fair value equivalent to about one-third of Basel regulatory capital.
- Unrealized gains:
  - Roughly equivalent to 50 percent of book value.
  - Equivalent to 12 percent of regulatory capital for sampled institutions (24 percent for cajas).
- Adjustment effect:
  - Own resources adjusted upwards prior to stress test; average CAR increased from 12.7 to 14.4 percent after adjustment.
- Impact of a 30-percent drop in equity prices (from adjusted figures):
  - Average loss equivalent to 8.6 percent of Basel regulatory capital.
  - Maximum impact of 21.6 percent.
  - Six banks and four cajas (jointly accounting for 37 percent of assets in the sample) would incur a net loss even after unrealized gains; for these institutions average impact 5.4 percent of regulatory capital and maximum impact 13.3 percent.
  - No institution would fall below the minimum requirement.
- Historical VaR (99.5 percent, three-month) for equity portfolios:
  - Average VaR represented 8.2 percent of regulatory capital; maximum 24.1 percent.

### Liquidity risk exercises and results
- Exercise 1 (contingency plans; one-week and one-month windows):
  - Assumed withdrawals (average): 39 percent of time deposits maturing within a month; 11 percent of demand deposits.
  - Institutions applied precautionary measures, increasing asset liquidity especially in the first week.
  - Coverage ratios (sources to uses):
    - Cumulative one-week average: 1.98; minimum (unadjusted) 1.09.
    - Cumulative one-month average: 1.66; minimum (unadjusted) 1.01.
    - Adjusted (closing interbank market): one-week average 1.87; one-month average 1.46; minimums near 1.09 and 1.00.
  - Projected sources of funds (average):
    - Regular operations: 0.46.
    - Central Bank liquidity withdrawal: 0.05.
    - Sale of fixed-income securities: 0.24.
    - Sale of variable-income securities: 0.13.
    - Debt issuance: 0.79.
    - Average debt issuance needed to cover liquidity needs: 0.12.
  - Note: Bond issuance represented 79 percent of the uses of funds implied by the exercise; minimum issuance needed 12 percent.
- Exercise 2 (liquid assets vs maturing liabilities; demand deposit shocks):
  - Definitions of assets: broader (projected sources + all available fixed- and variable-income securities) and narrower (variable-income securities not liquidatable).
  - Liability shocks: Uses of Funds 1: maturing liabilities within one month; Uses of Funds 2: plus 10 percent of demand deposits; Uses of Funds 3: plus 20 percent of demand deposits.
  - System-average coverage ratios ranged from 1.27 (least demanding) to 1.08 (most demanding).
  - Some sampled institutions' minimum coverage ratios dropped slightly below one but overall dispersion was relatively small.
  - Memo: at end-2004, liabilities with residents were mainly short-term: demand deposits 50 percent, time deposits 35 percent, repos 15 percent.

### Credit risk: sensitivity and scenario analysis (quantitative outcomes)
- Primary finding: credit risk main source of risk, but systemically important institutions able to absorb shocks with small effects on capital.
- Exercise 1 (sampled institutions):
  - PD shocks: mortgage +200 percent; commercial +100 percent from end-2004 levels.
  - Estimated losses: average 1.5 percent of regulatory capital; maximum 5.1 percent.
  - Profitability impact: on average losses represent 17 percent of after-tax profits; worst-case slightly less than 60 percent.
  - Accounting after statistical provisions: losses represent 10 percent of after-tax profits on average; worst-case slightly less than 14 percent.
  - Generic fund would fall by 13 percent on average; maximum around 25 percent for two sampled institutions.
- Top-down (BE) baseline: LGDs used 5 percent for mortgages and 25 percent for commercial credit; average drop in CAR 1.5 percent.
- Mission larger-sample robustness:
  - Exercise 2 (stressed LGDs 10 percent for mortgages and 45 percent for commercial): estimated losses 4.1 percent of regulatory capital on average for sampled institutions; 4.6 percent for entire system.
  - Exercise 3 (PDs = average 1992–2004 + 2 standard deviations; three-year window; stressed LGDs 10 percent mortgages, 45 percent commercial):
    - Stressed PDs: mortgages 6.31 percent; commercial 8.86 percent.
    - Average losses: 13 percent of Basel regulatory capital for sampled institutions; 14 percent for entire system.
    - Losses roughly 1.5 years of after-tax profits.
    - Larger for cajas and cooperatives (18 percent of their regulatory capital).
    - Maximum impact roughly one-quarter of regulatory capital.
    - Four credit institutions (about 5 percent of system assets) would fall slightly below minimum required CAR.
  - Scenario interpreted as unlikely under current circumstances; serves as medium-term lower-bound estimate.

### Scenario analysis and macro linkages (two-year macro scenarios)
- Baseline: European Commission spring 2005 forecast; baseline GDP growth 2.7 percent in 2005 and 2.8 percent in 2006.
- Scenario GDP growth deviations from baseline:
  - Oil Price Increase: -0.1 (2005), -0.7 (2006).
  - House Price: -0.4 (2005), -0.7 (2006).
  - Dollar Depreciation: -0.9 (2005), -0.6 (2006).
- Unemployment deviations:
  - Oil Price Increase: 0.1 (2005), 0.4 (2006).
  - House Price: 0.2 (2005), 0.6 (2006).
  - Dollar Depreciation: 0.4 (2005), 0.9 (2006).
- Real growth of housing prices (selected):
  - Oil Price Increase: -3.1 (2005), -2.4 (2006).
  - House Price scenario (assumption): -16.4 (2005), -4.6 (2006) — Spain shock: cumulative 30 percent decline over five years (17.3 percent in first year, 4.6 percent in second).
  - Dollar Depreciation: 0.8 (2005), 0.1 (2006).
- Participating institutions’ aggregated results (seven largest groups, ~two-thirds system assets):
  - Average CAR before shock: 12.7; Average CAR system: 11.6.
  - Dollar Depreciation: Average Impact on CAR -4.6; Maximum Impact on CAR -24.5; Average CAR After Shock 12.1; Worst CAR After Shock 9.6; Average Impact on Profits -8.3; Worst Impact on Profits -41.8.
  - Oil Price Increase: Average Impact on CAR 0.4; Maximum Impact on CAR -5.3; Average CAR After Shock 12.8; Worst CAR After Shock 12.0; Average Impact on Profits 1.0; Worst Impact on Profits -7.1.
  - Drop in House Prices: Average Impact on CAR -2.3; Maximum Impact on CAR -11.3; Average CAR After Shock 12.4; Worst CAR After Shock 11.2; Average Impact on Profits -4.9; Worst Impact on Profits -8.3.
  - Crisis in Latin America: Average Impact on CAR -0.7; Average CAR After Shock 12.6; Average Impact on Profits -11.6.
- BE system-wide results (aggregated top-down):
  - Dollar Depreciation: estimated losses 1.8 percent of regulatory capital.
  - Drop in House Prices: estimated losses 0.6 percent of regulatory capital.
  - Oil Price Increase: net effect a gain relative to baseline due to interest-margin effects.

### Dynamic panel models and loan-quality projections
- Database: panel covering roughly 90 percent of system assets for 1992–2004; 158 banks and cajas in sample after filters.
- Preferred explanatory variables: GDP growth, real growth of house prices, and long-term interest rate; NPLs logit-transformed.
- Key regression coefficients (selected; dependent variable: logit of NPLs):
  - Lagged autoregressive coefficients (LD. logit NPLs): Developers 0.582 [6.51]**; Commercial 0.660 [5.69]**; Mortgage 0.461 [3.49]**; Consumer 0.354 [3.71]**.
  - D. GDP Growth: Developers -0.091 [3.81]**; Commercial -0.110 [6.29]**; Mortgage -0.100 [4.28]**; Consumer -0.082 [5.21]**.
  - D. Real Growth of House Prices: Developers -0.065 [4.91]**; Mortgage -0.038 [2.45]*; Commercial -0.018 [1.79].
  - D. Long-term Interest: Mortgage 0.103 [4.44]**; Consumer 0.116 [5.70]**; Commercial 0.040 [2.56]*.
- Observations/sample sizes:
  - Developers: 942 observations; Number of banks 121.
  - Commercial: 986 observations; Number of banks 130.
  - Mortgage: 736 observations; Number of banks 98.
  - Consumer: 850 observations; Number of banks 114.
- Scenario-driven NPL projections (two-year horizon, mapping NPL changes to PD changes one-to-one):
  - Developers (house price scenario): NPLs from 0.4 percent in 2004 to 4.9 percent in 2006.
  - Consumer credit (oil scenario): NPLs from 2.2 percent in 2004 to 3.1 percent in 2006.
- Losses and capital impacts (two-year cumulative converted to yearly equivalents):
  - House price scenario: average losses equivalent to 11 percent of regulatory capital (maximum 27.4 percent); average CAR falling to 11.3 percent at end of two years.
  - Other scenarios: average losses between 4.9 and 6.6 percent of regulatory capital; maximum about 12–15 percent.
  - Small institutions: five institutions accounting for 2.5 percent of assets below minimum CAR under house price scenario; three institutions accounting for 2 percent of system assets below minimum CAR under oil price scenario.
  - Losses higher among cajas due to 26 percent exposure to real estate developers and construction companies at end-2004 (vs 20 percent for banks).
- Conclusion: banking system well prepared to absorb simulated scenarios without systemic threat to capital adequacy.

### Insurance sector — valuation, stress tests and key findings
- Fair value revaluation:
  - Fair value estimation increased net assets for the whole industry by 82 percent; individual institution increases between 8 and 360 percent.
  - Revaluation of real estate assets had the largest impact.
- Valuation and stress-test guidelines:
  - Reporting date: December 31, 2004 (or December 31, 2003 if end-2004 data unavailable).
  - Liabilities to reflect market-consistent assessment; unit-linked liabilities valued by reference to assets covering them.
  - Life-liability valuation: discounting with full term structure (default) or duration approach (fallback).
  - Assets marked to market where possible; mark-to-model allowed for non-regularly traded assets.
  - Single-event tests: interest rate risk; credit risk; market risk (equity, real estate, FX); underwriting risk; lapse risk; non-life risk.
  - Interest-rate shocks: both rise and fall considered.
  - Underwriting risk: stress survival or sickness tables by 15 percent.
  - Lapse risk: increase in lapse rate of 50 percent.
- Aggregate insurance stress-test findings:
  - System-wide largest impact: fall in real estate prices of 17 percent reduced insurers’ capital by 5.6 percent.
  - Underwriting risk: capital losses of 4 percent for the entire system.
  - Interest-rate shocks: parallel fall in curve of 200 basis points reduces capital by 1.3 percent.
- By insurer type (selected):
  - Life insurers:
    - Downward parallel shift of yield curve by 200 basis points: capital losses of 24.8 percent.
    - Parallel increase of yield curve by 200 basis points: capital improves by 8.1 percent.
    - Credit shock reduces capital by 9.9 percent.
    - Real estate shock reduces capital by 3.9 percent.
  - Non-life insurers:
    - Parallel upward shift of 200 basis points reduces capital by 10 percent.
    - Fall in real estate prices reduces capital by 7.1 percent.
  - Mixed insurers:
    - More exposed to housing-market-related risks; fall in real estate prices reduces capital by 5.6 percent.
    - Interest-rate shocks have limited impact due to netting effects.
- Insurance-specific stress tests (underwriting, lapse, non-life):
  - Underwriting risk (system): fall of 4 percent in capital.
  - Life insurers: increase of 15 percent in mortality rates leads to decrease of 18 percent in capital.
  - Non-life insurers: 10-percent increase in average claim in “mass events” reduces capital by 7.3 percent.
  - Mixed companies: largest impact a decrease of 1.4 percent from average claim increase of peak events.
- Table 13 selected figures (Impact on Insurer’s Capital, December 2004; results in percent of insurer's capital; positive indicates reduction in capital):
  - All Institutions — Underwriting risk: Total Avg. 4.0; Std. Dev. 7.1; Max. 15.1; Min. -14.0.
  - Life Institutions — Underwriting risk: Total Avg. 18.2; Std. Dev. 18.2; Max. 23.0; Min. -62.2.
  - Non life institutions — Non life risk (group 1): Total Avg. 7.3; Std. Dev. 3.6; Max. 4.1; Min. 11.5.
  - Non life institutions — Non life risk (group 2, "mass" events): Total Avg. 4.7; Std. Dev. 4.6; Max. 3.3; Min. 11.4.
  - Non life institutions — Non life risk (group 2, "peak" events): Total Avg. 4.6; Std. Dev. 3.2; Max. 2.0; Min. 5.7.
  - Mixed institutions — Non life risk (group 1): Total Avg. -1.4; Std. Dev. -1.6; Max. 9.4; Min. 2.8.
- Cross-sector exposure:
  - Losses from insurance stress tests amounted to a negligible part of the regulatory capital of their parent credit institutions.
- Overall conclusion: Insurance system resilient to a wide range of shocks, with heterogeneous effects across life, non-life, and mixed insurers.

*Source — _cr06216 - 1.   List of Groups Participating in the Stress Test Exercises (PDF chapter/section).*

### 1.   List of Groups Participating in the Stress Test Exercises ....................................................4

### _cr06216 - 1.   List of Groups Participating in the Stress Test Exercises ....................................................4

### Overview
- Purpose: Assess resilience of Spain’s banking and insurance systems through stress test exercises carried out in the context of the FSAP.
- Coverage: Consolidated group level, covering the most important risk exposures; shocks applied to domestic business, overseas business, or both.
- Reference date for financial information: end-2004.
- Types of tests:
  - Sensitivity analysis: one-off shocks to individual or multiple risk factors.
  - Scenario analysis: simultaneous, internally consistent changes in a group of macroeconomic variables over a two-year window (models developed by Banco de España).
- Four macroeconomic scenarios considered:
  - Permanent increase in oil prices to US$80 per barrel.
  - Fall in real house prices in Spain and the United Kingdom (Spain: cumulative fall of 30 percent over five years; translated into decreases of 17.3 percent in the first year and 4.6 percent in the second; United Kingdom: 37 percent fall translated into decreases of 20 percent and 8 percent in first and second years, respectively).
  - 30-percent depreciation of the U.S. dollar against other currencies.
  - Crisis in Latin America.
- Exercise participants:
  - Bottom-up: 7 largest financial groups, encompassing 37 credit institutions and roughly two-thirds of system assets (Table 1 indicates Total Sample: 66.9 percent of system assets; 65.9 percent of system liabilities).
  - Top-down: Complementary exercises by authorities (BE) and mission covering the entire banking system using supervisory data and off-site supervisory models.
  - Insurance: Sample of 27 insurers, covering 50 percent of the non-life market and 62 percent of the life market; parallel stress tests where applicable and results aggregated with parent credit institutions for cross-sector impact.

### Banking system — Key findings and conclusions
- Overall resilience: Banking system able to absorb losses associated with the shocks considered without systemic distress.
- Contributing factors:
  - Strong asset quality for the vast majority of credit institutions.
  - Comfortable loan-loss cushion induced by the dynamic provisioning system.
  - Sound risk management practices by systemically important institutions.
  - Supportive macroeconomic environment.
- Equity price risk (not-marked-to-market holdings):
  - Sample: 19 credit institutions representing roughly 80 percent of system assets.
  - A 30 percent across-the-board drop in equity prices would generate losses equivalent to 8.6 percent of regulatory capital (14.3 percent for the sampled cajas).
  - Large unrealized gains amounted to 12 percent of regulatory capital (24 percent for cajas), providing a substantial margin.
- Market risk in trading portfolios: Minimal due to relatively small size and active use of hedging strategies.
- Liquidity risk:
  - On average, the seven participating groups would be able to cope with a withdrawal of 39 percent of their maturing time deposits within a month, plus 11 percent of demand deposits, without cutting back normal operations, even if they lose access to interbank market financing.
  - Some institutions rely heavily on issuance of covered bonds in contingency liquidity plans, posing potential systemic risk if market sentiment/liquidity deteriorates.
- Credit risk (sensitivity analyses):
  - Shock applied: Increase of PDs of mortgage and commercial loans by 200 and 100 percent, respectively, from end-2004 levels (replicating the largest yearly increase observed during 1990-2004).
  - Participating institutions’ estimates: average loss of 1.5 percent of Basel regulatory capital (in line with BE estimates for entire system).
  - Mission robustness check (similar shock to PDs with more conservative LGD): average losses of about 5 percent of Basel regulatory capital.
- Severe credit deterioration scenario:
  - Shock: Linear increase in PDs over three years to their average levels observed during 1993–2004 plus two standard deviations.
  - Estimated average losses equivalent to 13 percent of Basel regulatory capital, maximum of 27 percent.
  - Resulted in four credit institutions (representing 5 percent of system assets) going slightly below 8 percent Basel CAR.
- Dynamic panel + scenario results:
  - Fall in house prices: average losses equivalent to 11 percent of Basel CAR, maximum effect of 27 percent; effects larger for cajas (five institutions accounting for less than 3 percent of assets going slightly below 8 percent Basel CAR).
  - Oil price shock: average losses slightly less than 7 percent of regulatory capital.
  - Dollar depreciation: average losses of 5 percent of regulatory capital.
- Scenario analysis by participating institutions and BE:
  - Participating credit institutions: largest losses associated with dollar depreciation, average impact of 5 percent of regulatory capital; fall in house prices in Spain and the United Kingdom represented slightly more than 2 percent of regulatory capital; crisis in Latin America below 1 percent.
  - BE exercises produced same relative ordering but even lower estimates.

### Insurance sector — Key findings and conclusions
- Overall resilience: Insurance system resilient to a wide range of shocks, with heterogeneous effects across life, non-life, and mixed insurers.
- Largest sectoral impact:
  - Fall in real estate prices: 17 percent price drop entailed capital losses of 5.6 percent (for the sector as a whole).
  - Underwriting risk: capital losses of 4 percent for the entire system.
- Segment-specific impacts:
  - Life insurers: downward parallel shift of the yield curve by 200 basis points entailed capital losses of 24.8 percent.
  - Non-life insurers: upward parallel shift of the yield curve entailed capital losses of 10 percent.
  - Mixed insurers: relatively more exposed to housing-market-related risks.
- Cross-sector exposure: Losses from insurance stress tests amounted to a negligible part of the regulatory capital of their parent credit institutions.

### Methodology — Approaches and risk categories
- Combined bottom-up and top-down approaches:
  - Bottom-up: Participating systemically important credit institutions used their own datasets and internal models applying agreed shocks and scenarios.
  - Top-down: BE and mission used supervisory data and off-site supervisory models to validate and generalize bottom-up results.
- Risk categories in sensitivity analysis: (a) market risk; (b) interest rate risk on all assets and liabilities; (c) credit risk; (d) liquidity risk.
- Calibration: Shocks calibrated using historical information spanning at least an entire economic cycle or episodes of financial distress; shocks often set to the highest monthly variation during 1980–2004. Monthly volatilities measured by the standard deviation of daily-data series.
- Market risk shocks applied to trading portfolios; separate interest rate risk exercise applied to all assets and liabilities of participating credit institutions.

### Selected shocks used in sensitivity analysis to market risks (as presented)
- Interest rates (euro, U.S. dollar, and British pound exposures):
  - Interest rate increase; negative tilt short : 100 b.p.; long: 30 b.p.
  - Interest rate increase; positive tilt (*)short : 30 b.p.; long: 100 b.p.
  - Interest rate decrease; negative tilt short : -30 b.p.; long: -100 b.p.
  - Interest rate decrease; positive tilt short : -100 b.p.; long: -30 b.p.
  - Increase in volatility of short-term interest rates25 b.p.
  - Increase in volatility of long-term interest rates 2/15 b.p.
  - Increase in corporate bond spreads 3/15 b.p. to 300 b.p.
- Equity price:
  - Across the board drop in equity prices (*)-30 percent
  - Increase in the volatility of market indexes200 b.p.
- Exchange rate:
  - Appreciation 10 percent U.S. dollar/euro-10 percent
  - Depreciation 10 percent U.S. dollar/euro10 percent
  - Appreciation 7 percent British pound/euro-7 percent
  - Depreciation 7 percent British pound/euro7 percent
  - Increase in the volatility U.S. dollar/euro170 b.p.
  - Increase in the volatility British pound/euro120 b.p.
- Notes extracted from table:
  - For bonds rated BBB; and 300 b.p. for bonds rated below BBB.
  - For British pound exposures, the shock to volatility of long term interest rates was set to 25 b.p.
  - The tests were applied to 7 banking groups covering 37 credit institutions and 67 percent of system assets.
  - Sensitivity analysis to multiple risk factors entailed combinations of the shocks marked with (*).
  - The increase in corporate bond spreads was as follows: 15 b.p. for corporate bonds rated AAA through A; 100

*Italic: Source — _cr06216 - 1.   List of Groups Participating in the Stress Test Exercises (PDF chapter/section)._

### 20.      In addition to the shocks to single risk factors, the participating credit

### _cr06216 - 20.      In addition to the shocks to single risk factors, the participating credit

### Market and trading portfolio combined shock
- Participating credit institutions estimated effects of a combined shock to interest rates and equity prices on trading portfolios.
- Combined shock specification:
  - Increase in interest rates in U.S. dollars and euros: 30 b.p. at the short end and 100 b.p. at the long end.
  - Increase in interest rates in British pounds: 100 b.p. at the short end and 30 b.p. at the long end.
  - Across-the-board drop in equity prices: 30 percent.
- Combined shocks chosen to maximize individual effects and be conservative.

### Equity price risk in not-for-sale portfolios (banking book)
- Sample: 19 credit institutions (7 banks and 12 cajas), jointly accounting for 80 percent of system assets at end-2004.
- Shock: across-the-board 30-percent drop in equity prices applied as in trading-book exercise.
- Unquoted participations accounting for 18 percent of the portfolio were subjected to a similar 30-percent shock.

### Interest rate risk on structural positions
- Shocks entailed:
  - Parallel shift in the yield curve of 200 b.p.
  - Increase in slope by 0, 50, and 100 b.p. at the one-day, three-year, and ten-year maturities, with linear interpolation for intermediate maturities.
- Shocks applied directly to all assets and liabilities by participating credit institutions.

### Credit risk baseline exercises
- Focused on two loan categories: mortgage credit and commercial credit, representing about 90 percent of total loans at end-2004.
- Shock sizes based on credit registry data (Central de Riesgo) for 1990–2004:
  - Increase in PDs: mortgage credit +200 percent from 2004 levels; commercial credit +100 percent from 2004 levels.
- Application:
  - Shocks applied to PDs estimated by individual banks using their own datasets and models; effects computed over a one-year window.
  - Losses estimated by combining stressed PDs with point-in-time LGDs estimated by each participating credit institution.
  - BE conducted parallel aggregate exercise using supervisory PDs and LGDs.
- Point-in-time LGD estimates used by BE: 5 percent for mortgage loans and 25 percent for commercial loans.

### Credit risk sensitivity (robustness) exercises
- Coverage: sample of 86 credit institutions (~94 percent of system assets) and aggregate system data.
- First exercise:
  - PD shocks similar to baseline combined with larger LGD estimates to reflect downturn deterioration.
  - Stressed LGD values: 10 percent for mortgage loans (Basel floor) and 45 percent for commercial loans.
- Second exercise (more severe deterioration):
  - Shocks to PDs set to average values during 1992–2004 plus two standard deviations.
  - Conducted over a three-year window with linear interpolation to increase PDs from current values at beginning of year 1 to target at beginning of year 3.
  - For mortgages: PD increased from 0.5 percent (point-in-time estimate in 2004) to 6.3 percent at end of three-year window.
  - For commercial loans: PD increased from 1.8 percent to 8.9 percent over same period.
  - Stressed LGDs used: 10 percent for mortgage and 45 percent for commercial loans.

### Liquidity risk exercises
- First exercise (systemically important institutions):
  - Negative shock to liabilities of an otherwise solvent institution (idiosyncratic, not systemic run).
  - Institutions required to apply liquidity contingency plans, continue normal business, and not use extraordinary central bank financing.
  - Institutions could select shock sizes on a pre-specified subset of liabilities and report sources and uses of funds over one-week and one-month windows using common reporting formats.
  - Baseline results computed showing relative evolution of sources and uses across sub-periods.
- More stringent liquidity scenario (two modifications):
  - Minimum shock of 10 percent to demand deposits (4 percent for the first week) applied to institutions that used a lower value; shock calibrated to ninety-ninth percentile of monthly changes in demand deposits for 1994–2004.
  - Contingency plans modified to close access to any proposed interbank market financing.
- Second liquidity exercise (ratios):
  - Authorities compared various liquidity ratios for seven sampled credit institutions and for the entire banking system.
  - Baseline liquidity ratio: broad liquid assets (projected sources within one month + available fixed- and variable-income securities) versus liabilities maturing within one month.
  - Liquidity ratio recalculated combining shocks to demand deposits in the range of 10–20 percent with increasingly stringent assumptions on asset liquidation capacity.

### Scenario analysis: macroeconomic two-year scenarios
- Scenarios projected over a two-year horizon using extreme but plausible shocks to key exogenous variables; projections estimated by BE with econometric models comprising four components: (a) exogenous variables, (b) effects on international variables via NiGEM, (c) effects on key Latin American economies via OEF, (d) effects on Spanish macro variables using BE models.
- Target macro variables for Spain: GDP growth, credit growth, inflation, house prices, unemployment, interest rates, stock market prices, exchange rate, among others.

### Four scenario descriptions and calibrations
- Decline in property prices in Spain and the U.K.:
  - Spain shock: cumulative 30 percent decline in real terms over five years—17.3 percent during the first year and 4.6 percent during the second.
  - Shock consistent with estimated overvaluation 24–35 percent and exceeded historic 1992–96 correction.
  - U.K. (Abbey exposure): cumulative 37 drop in real house prices over five years—20 percent during the first and 8 percent during the second; implied correction of a 60 percent overvaluation in U.K. house prices.
- Permanent increase in oil prices:
  - Shock calibrated to US$80 per barrel (highest historic real level reached in 1980).
  - Effects modeled by BE and NiGEM; effects on Latin America considered comparatively smaller and not included.
- Crisis in Latin America:
  - Large Spanish banks' end-2004 exposures concentrated: México 53 percent, Chile 14 percent, Brazil 12 percent, Argentina 4 percent.
  - Scenario: political crisis in México transmitted regionally via capital market turmoil; focus on México and Brazil.
  - México shocks:
    - Permanent increase in country risk premium by 400 b.p.
    - 20-percent depreciation in equilibrium exchange rate and increase in exchange rate risk premium by 150 b.p.
    - Permanent 50 percent drop in capital inflows relative to 2004.
    - Gradual reduction of external competitiveness by 25 percent.
  - Brazil shocks:
    - Permanent increase in country risk premium by 800 b.p.
    - 30-percent depreciation in equilibrium exchange rate and increase in exchange rate risk premium by 800 b.p.
    - Permanent drop in capital inflows by 50 percent.
    - Gradual reduction in external competitiveness by 30 percent.
- Depreciation of the U.S. dollar:
  - Sudden and permanent 30-percent nominal depreciation of the U.S. dollar against other main currencies, including the euro.
  - Slightly above largest yearly depreciation observed in last 25 years (25 percent in 1985).
  - Effects in U.S. and euro area simulated with NiGEM; Latin American effects via OEF and BE adjustments for increased country risk premiums, currency depreciation, and permanent drop in capital flows.
  - BE modeled effects on Spanish economy directly.

### Implementation of scenario analysis
- Participating credit institutions used their own models (bottom-up) to estimate effects on financial positions given each scenario.
- BE conducted parallel top-down exercise with off-site supervisory models for aggregate banking system, simulating impact on aggregated balance sheet and income statement components.
- Both approaches measured scenario impacts against baseline.

### Macro stress tests based on dynamic panel models
- Developed by mission and BE using panel database covering roughly 90 percent of system assets for 1992–2004.
- Dynamic panel techniques allowed separate estimation across banks and cajas, and dispersion within institutional groups.
- Designed to complement scenario-based stress tests and assess credit risk from domestic housing exposures.

### Reporting formats and metrics
- Two primary metrics:
  - Impact on ratio of regulatory capital to risk-weighted assets (CAR), using Basel definition.
  - Impact on after-tax profits.
- Results typically included average and maximum losses in CAR and after-tax profits, and average and minimum CAR after shocks.
- Reporting conveyed distribution of results across institutional types and within groups, indicating number of institutions and percent of system assets within CAR intervals after shocks.
- Results on capital measured against Basel own resources; impacts on profits excluded cushioning effect of Spanish dynamic provisioning.
- Note: Spanish solvency regulations stricter than Basel and do not count statistical loan loss provision as own resources.
  - Using Basel definition of own resources: CAR for entire system was 11.6 at end-December 2004, compared with 10.3 according to Spanish definition.

### Treatment of insurance company losses
- Losses experienced by insurance companies in shocks were added to those experienced by their parent credit institutions for cross-sector risk assessment, since insurance companies were not included in consolidated financial statements of parent credit institutions.

### Stress test results — sensitivity analysis (selected findings)
- Market risks for systemically important credit institutions were negligible due to small trading portfolios and active risk controls.
- Average trading portfolios of participating credit institutions:
  - Roughly equivalent to one-third of their regulatory capital.
  - Equivalent to 6.4 percent of their credit portfolios.
- Estimated losses on market-risk stress tests accounted for a negligible part of regulatory capital.
- Combined shocks (interest rates increase and equity price drop) produced larger losses but represented on average 0.07 percent of Basel CAR, with a maximum impact of 1.35 percent.
- Table 3 selected results (high-level reproductions of reported figures):
  - Average CAR and ROA before shock: 12.68 and 0.9, respectively.
  - Market risks (bank estimates) examples:
    - Interest rate (Euro exposures), increase; negative tilt (short; long) (100 b.p.; 30 b.p.): average change in CAR -0.05; worst -0.13; CAR before shock 12.68; CAR worst 10.14; percent of regulatory capital in profits impact -0.45 to -0.79.
    - Interest rate (Euro exposures), increase; positive tilt (30 b.p.; 100 b.p.): average CAR change -0.04; worst -0.08; CAR before shock 12.68; CAR worst 10.14; percent of regulatory capital in profits impact -0.32 to -1.01.
    - Interest rate (US dollar exposures), positive tilt (30 b.p.; 100 b.p.): average CAR change 0.00; worst -0.03; CAR before shock 12.68; CAR worst 10.14; percent of regulatory capital in profits impact -0.08 to 0.37.
    - Increase in corporate spreads (15 b.p.; 300 b.p.): average CAR change -0.09; worst -0.32; CAR before shock 12.67; CAR worst 10.12; percent of regulatory capital in profits impact -0.61 to -1.70.
    - Equity price risk in trading book, across-the-board drop in equity prices -30 percent: average CAR change -0.04; worst -1.30; CAR before shock 12.68; CAR worst 9.99; percent of regulatory capital in profits impact -0.36 to -15.15.
    - Multiple factors (combination of shocks marked with (*)): average CAR change -0.07; worst -1.35; CAR before shock 12.67; CAR worst 9.99; percent of regulatory capital in profits impact -0.75 to -15.73.
  - Interest rate risk (bank estimates) examples:
    - Parallel increase in the euro yield curve 200 b.p.: average CAR change 2.10; worst -3.30; CAR before shock 12.94; CAR worst 9.76.
    - Parallel decrease in the euro yield curve -200 b.p.: average CAR change -2.80; worst -6.20; CAR before shock 12.32; CAR worst 9.38.
    - Increase and positive tilt in yield curve (3 years; 10 years) 50 b.p.; 100 b.p.: average CAR change 1.30; worst -2.10; CAR before shock 12.84; CAR worst 9.86.
    - Decrease and negative tilt in yield curve (3 years; 10 years) -50 b.p.; -100 b.p.: average CAR change -1.10; worst -3.10; CAR before shock 12.54; CAR worst 9.77.
- Notes from Table 3:
  - Average change on CAR and profits is weighted by the own funds of the sampled credit institutions.
  - Worst computed by applying the largest variation in CAR to the lowest CAR in sampled institutions.
  - Increase in corporate spreads specification: 15 b.p. for corporate bonds rated AAA through A; 100 b.p. for bonds rated BBB; and 300 b.p. for bonds rated below BBB.
- Sensitivity tests applied to seven banking groups covering 37 credit institutions and 67 percent of consolidated system assets.

*Source: IMF staff and Banco de España materials in the provided content.*

### 38.      Equity risk in the banking book was found to be substantial, albeit large

### 38.      Equity risk in the banking book was found to be substantial, albeit large unrealized gains offered a cushion against adverse equity price movements.

### Equity holdings and sensitivity to a 30-percent equity price shock
- Sample: 19 credit institutions accounting for roughly 80 percent of system assets.
- Average equity holdings in the banking book, at fair value, were equivalent to about one-third of the Basel regulatory capital.
- Unrealized gains on these investments:
  - Roughly equivalent to 50 percent of their book value.
  - Equivalent to 12 percent of the regulatory capital of the sampled institutions (24 percent for the cajas).
- Adjustment for unrealized gains:
  - Own resources adjusted upwards prior to the stress test.
  - Average CAR increased from 12.7 to 14.4 percent after adjustment.
- Impact of a 30-percent drop in equity prices (from adjusted figures):
  - Average loss equivalent to 8.6 percent of Basel regulatory capital.
  - Maximum impact of 21.6 percent.
  - Effect substantially larger for sampled cajas.
- Net-loss outcome:
  - Six banks and four cajas (jointly accounting for 37 percent of assets in the sample) would incur a net loss even after unrealized gains.
  - For these institutions: average impact of 5.4 percent of regulatory capital and a maximum impact of 13.3 percent.
  - No institution would fall below the minimum requirement.
- Table 4 reference: the exercise based on a 30 percent across-the-board drop in equity prices.

### Historical VaR (99.5 percent, three-month) exercise
- Scope: fair value of equity investments of the same 19 credit institutions; portfolio composition as of end-December 2004.
- Methodology:
  - Quoted companies (84 percent of portfolios): individual stock market prices.
  - Non-quoted companies: mapped into six main sectors using Madrid stock exchange sector indexes.
  - Price volatilities computed at daily frequency with a three-month rolling window from February 2002 to September 2005.
  - VaR computed at the 99.5 confidence level over a three-month holding period (Basel II recommendation).
- Results:
  - Average VaR represented 8.2 percent of regulatory capital.
  - Maximum of 24.1 percent.
  - Average VaR reflected low stock market volatility in recent years; substantially larger estimates observed after credit and corporate events of 2002.

### Insurance market-risk losses and interest rate risk
- Insurance companies:
  - Losses from parallel market-risk exercises were immaterial for capital adequacy of parent credit institutions.
  - Losses represented a negligible part of Basel regulatory capital of parent banks.
- Interest rate risk (entire balance sheet and off-balance sheet positions):
  - Found to be small.
  - Credit institutions more sensitive to decreases in interest rates (predominance of loans at floating rates and increased reliance on debt financing).
  - A 200 basis point parallel drop in the euro yield curve would cause:
    - Average loss equivalent to 2.8 percent of Tier I and Tier II capital.
    - Maximum decline reaching 6.2 percent.
    - Both figures well below the reference 20 percent suggested by the Basel Committee.
  - Inclusion of insurance companies’ results did not materially change outcomes.

### Liquidity risk: Exercise 1 (contingency plans and sources of funds)
- Assumed shocks (average across institutions):
  - Withdrawal of 39 percent of the time deposits maturing within a month.
  - Withdrawal of 11 percent of demand deposits.
- Institutions’ reaction: precautionary measures to increase asset liquidity, particularly during the first week.
- Coverage ratios (ratio of sources to uses of funds):
  - Cumulative one-week average: 1.98.
  - Cumulative one-month average: 1.66.
  - Minimums (unadjusted): 1.09 (one-week) and 1.01 (one-month).
- Adjusted figures (closing interbank market):
  - Average one-week: 1.87.
  - Average one-month: 1.46.
  - Minimums remained near 1.09 and 1.00.
- Proposed sources of funds (average detail as provided by institutions):
  - Projected sources from regular operations: 0.46.
  - Liquidity withdrawal from Central Bank: 0.05.
  - Sale of fixed-income securities: 0.24.
  - Sale of variable-income securities: 0.13.
  - Debt issuance: 0.79.
  - Average debt issuance needed to cover liquidity needs: 0.12.
- Notes on debt issuance:
  - Bond issuance represented 79 percent of the uses of funds implied by the exercise.
  - Minimum issuance needed to cover liquidity needs was 12 percent; the difference used to increase precautionary liquidity.
  - Some institutions showed substantial reliance on debt issuance and covered bonds, which could entail risks if system-wide liquidity conditions deteriorate.

### Liquidity risk: Exercise 2 (liquid assets vs maturing liabilities)
- Additional BE exercise comparing definitions of liquid assets against maturing liabilities plus drops in demand deposits.
- Two definitions:
  - Broader: projected sources from regular operations within a month + all available fixed- and variable-income securities.
  - Narrower: assumes variable income securities could not be liquidated on short notice.
- Liability shocks:
  - Uses of Funds 1: maturing liabilities within one month.
  - Uses of Funds 2: maturing liabilities within one month, plus 10 percent of demand deposits.
  - Uses of Funds 3: maturing liabilities within one month, plus 20 percent of demand deposits.
- Results:
  - System-average coverage ratios ranged from 1.27 under the least demanding scenario to 1.08 at the most demanding scenario.
  - Sampled institutions produced roughly similar average results; in some cases minimum coverage ratios dropped slightly below one but overall dispersion was relatively small.
- Memo: at end-2004, liabilities with residents were mainly short-term: demand deposits 50 percent, time deposits 35 percent, repos 15 percent.

### Credit risk: sensitivity and scenario analysis
- Primary finding: credit risk was the main source of risk, but systemically important institutions would be able to absorb the shocks considered with small effects on capital.
- Exercise 1 (sensitivity by sampled credit institutions):
  - Shock: increase estimated PDs of mortgage and commercial loans by 200 percent and 100 percent, respectively, from end-2004 levels.
  - Estimated losses: 1.5 percent of regulatory capital on average, maximum impact of 5.1 percent (Table 7).
  - Profitability impact:
    - These losses represent 17 percent of after-tax profits, on average, and slightly less than 60 percent in the worst case.
    - Accounting effects on profitability after statistical provisions: 10 percent of after-tax profits, on average, and slightly less than 14 percent in the worst case.
  - Resulting change in generic provisions/fund:
    - Generic fund would fall by 13 percent on average, with a maximum of around 25 percent for two sampled institutions.
- Top-down approach (aggregate system data):
  - Expected losses computed using LGDs of 5 percent for mortgages and 25 percent for commercial credit.
  - Results roughly similar to participating institutions: average drop in CAR of 1.5 percent.
  - Limitation: lack of distributional information across institutions.
- Parallel mission exercises (larger sample of 86 credit institutions covering 94 percent of system assets):
  - Exercise 2 (larger LGDs to account for procyclicality): estimated losses accounted for 4.1 percent of regulatory capital on average for sampled institutions, and 4.6 percent for the entire system.
  - Losses slightly larger for cajas and cooperatives; none fell below minimum regulatory capital.
- Exercise 3 (more extreme deterioration: PDs = average 1992–2004 + 2 standard deviations; stressed LGDs):
  - Stressed PDs used: 6.31 percent for mortgages and 8.86 percent for commercial credit.
  - Average losses:
    - 13 percent of Basel regulatory capital for sampled credit institutions.
    - 14 percent for the entire system.
    - Approximately 1.5 years of after-tax profits.
    - Larger for cajas and cooperatives (18 percent of their regulatory capital).
  - Maximum impact roughly one-quarter of regulatory capital.
  - Four credit institutions (about 5 percent of system assets) would fall slightly below minimum required CAR in this scenario.
  - Scenario viewed as unlikely under current circumstances; interpreted as a medium-term lower-bound estimate.

*Source: IMF staff summary of stress test exercises and Bank of Spain data as presented in the referenced chapter.*

### 49.      The scenarios considered did not translate into a sharp deceleration of GDP

### 49.      The scenarios considered did not translate into a sharp deceleration of GDP 

### Macroeconomic baseline and scenarios
- Baseline projections used the European Commission spring 2005 economic forecast, consistent with the 1995 Spanish National Accounts; subsequent base year 2000 National Accounts released after the exercise entailed an upward revision of both the level and growth of GDP.
- Baseline GDP growth projections: 2.7 percent in 2005 and 2.8 percent in 2006.
- Models and complementary BE simulations were used for some variables, such as house prices and credit to the private sector.
- The stressed scenarios considered included: Oil Price Increase, House Price fall, Dollar Depreciation, and Crisis in Latin America.

### Scenario impacts on GDP, unemployment, inflation, interest rates, and house prices (deviations from baseline)
- GDP Growth (deviations and baseline combined as shown in Table 8):
  - Baseline: 2.7 (2005), 2.8 (2006)
  - Oil Price Increase: -0.1 (2005), -0.7 (2006)
  - House Price: -0.4 (2005), -0.7 (2006)
  - Dollar Depreciation: -0.9 (2005), -0.6 (2006)
- Unemployment Rate (yearly averages, deviations from baseline):
  - Baseline: 10.4 (2005), 10.3 (2006)
  - Oil Price Increase: 0.1 (2005), 0.4 (2006)
  - House Price: 0.2 (2005), 0.6 (2006)
  - Dollar Depreciation: 0.4 (2005), 0.9 (2006)
- Household Debt to Gross Disposable Income:
  - Baseline: 104.1 (2005), 103.5 (2006)
  - Oil Price Increase: -0.2 (2005), -1.7 (2006)
  - House Price: -2.5 (2005), -4.3 (2006)
  - Dollar Depreciation: 0.3 (2005), 0.9 (2006)
- Firm Debt to Profits:
  - Baseline: 470.1 (2005), 465.9 (2006)
  - Oil Price Increase: -3.3 (2005), -8.4 (2006)
  - House Price: 0.2 (2005), -0.2 (2006)
  - Dollar Depreciation: 3.8 (2005), 5.8 (2006)
- Inflation:
  - Baseline: 2.9 (2005), 2.6 (2006)
  - Oil Price Increase: 0.7 (2005), 0.6 (2006)
  - House Price: -0.1 (2005), -0.3 (2006)
  - Dollar Depreciation: -0.5 (2005), -1.1 (2006)
- Short-term Interest:
  - Baseline: 2.2 (2005), 2.7 (2006)
  - Oil Price Increase: 1.0 (2005), 1.0 (2006)
  - House Price: 0.0 (2005), 0.0 (2006)
  - Dollar Depreciation: -0.9 (2005), -0.7 (2006)
- Long-term Interest:
  - Baseline: 3.8 (2005), 4.3 (2006)
  - Oil Price Increase: 0.8 (2005), 0.8 (2006)
  - House Price: 0.0 (2005), 0.0 (2006)
  - Dollar Depreciation: -0.3 (2005), -0.3 (2006)
- Real Growth of Housing Prices:
  - Baseline: -0.9 (2005), 0.0 (2006)
  - Oil Price Increase: -3.1 (2005), -2.4 (2006)
  - House Price scenario (assumption): -16.4 (2005), -4.6 (2006)
  - Dollar Depreciation: 0.8 (2005), 0.1 (2006)
- Narrative highlights:
  - Dollar depreciation scenario produced the largest negative impact on Spanish GDP relative to baseline: a drop of 0.9 percent in 2005 and 0.6 percent in 2006.
  - Real house prices: under the oil scenario, growth rate of real house prices fell by 4 percent in 2005 and 2.4 percent in 2006; under the house price scenario by assumption, house prices dropped 17.3 percent in 2005 and 4.6 percent in 2006.
  - Simulated inflation and interest-rate evolution incorporated assumptions of no euro-area monetary reaction to Spain-specific house-price drops, tightening in response to oil shock, and easing following dollar depreciation.

### Bank stress test exercises and participating institutions' results
- Participating credit institutions were provided macro projections under the four scenarios and required to estimate losses using their internal models; BE conducted parallel system-wide stress tests and the mission and BE conducted additional domestic-market credit-risk exercises focused on housing exposures.
- Participating institutions assessed losses accumulated over two years and reported deviations relative to the baseline.
- Key aggregated results (Table 9, sample estimates covering seven largest financial groups, roughly two-thirds of system assets):
  - Average CAR participating institutions before shock: 12.7
  - Average CAR system: 11.6
  - Dollar Depreciation: Average Impact on CAR -4.6; Maximum Impact on CAR -24.5; Average CAR After Shock 12.1; Worst CAR After Shock 9.6; Average Impact on Profits -8.3; Worst Impact on Profits -41.8
  - Oil Price Increase: Average Impact on CAR 0.4; Maximum Impact on CAR -5.3; Average CAR After Shock 12.8; Worst CAR After Shock 12.0; Average Impact on Profits 1.0; Worst Impact on Profits -7.1
  - Drop in House Prices: Average Impact on CAR -2.3; Maximum Impact on CAR -11.3; Average CAR After Shock 12.4; Worst CAR After Shock 11.2; Average Impact on Profits -4.9; Worst Impact on Profits -8.3
  - Crisis in Latin America: Average Impact on CAR -0.7; Average CAR After Shock 12.6; Average Impact on Profits -11.6
- After the dollar-depreciation shock, average CAR of participating institutions would fall to 12.1 percent, with a minimum of 9.6 percent.
- Estimated losses under the fall in house prices in Spain and the U.K. were 2.3 percent of regulatory capital on average; crisis in Latin America losses were below 1 percent on average.
- Conclusion from internal-model exercises: none of the scenarios posed a threat to participating institutions' capital adequacy; systemically important institutions considered well prepared to absorb losses from scenarios considered.

### Bank of Spain (BE) system-wide exercise results
- BE aggregated-system simulations produced the same ordering of scenarios by losses but with lower estimated impacts versus participating institutions' internal models.
- Estimated system-wide impacts (deviations from baseline):
  - Dollar Depreciation: estimated losses amounted to 1.8 percent of regulatory capital; BE noted decrease in euro-area interest rates plus Spanish activity decline reduced interest margins and commission fees and worsened loan quality.
  - Drop in House Prices: estimated losses amounted to 0.6 percent of regulatory capital.
  - Oil Price Increase: net effect was a gain relative to baseline as tighter euro-area monetary policy increased interest margins that counteracted loan losses.
- Caveat emphasized: results reflect soundness of system but also relatively mild macro projections for Spain under the scenarios; scenarios were low-probability but severe, yet produced macro projections for Spain relatively close to baseline.

### Exercises combining dynamic panel models with scenario analysis (mission and BE)
- Purpose: provide complementary estimates focused on credit risk from domestic exposures, using dynamic panel models combining cross-section and time-series data.
- Advantages of approach: 
  - obtain separate results across institutional types;
  - obtain measures of dispersion of results;
  - differentiate short- and long-term effects;
  - conduct robustness checks with slightly modified scenarios.
- Data and sample:
  - Comprehensive BE database including 158 banks and cajas, accounting for roughly 90 percent of system assets during 1992-2004.
  - Filter excluded institutions with total assets below 0.01 percent of system assets.
  - Mergers and acquisitions treated by keeping pre- and post-merger institutions as separate cases.
  - Loan portfolios and NPLs classified into four categories: (a) mortgages to households, (b) credit to real estate developers and construction companies, (c) commercial credit to other sectors, and (d) consumer and other credit.
- Data construction and limitations:
  - Database combined credit registry (CIR) information with accounting information from individual credit institutions.
  - CIR provided historical coverage but lower precision in loan classification; loans to households with real collateral were classified as mortgages if maturity exceeded three years.
  - Models used unconsolidated (solo) institution-level data mapped to parent groups for aggregation to assess capital adequacy effects.
- Historical loan-quality patterns (1992–2004):
  - Strong procyclicality: NPLs peaked during early-1990s downturn.
  - Most cycle-sensitive category: credit to real estate developers and construction companies (e.g., NPLs reached 14.2 percent for cajas in 1993 and 11.9 percent for banks in 1994).
  - Since mid-1990s, loan quality improved across-the-board, with substantial dispersion across individual institutions, especially in consumer and commercial credit.
- Links between mortgage NPLs and macro variables:
  - Mortgage portfolio quality improved with: number of workers per household, real household income, and real house prices.
  - Mortgage portfolio quality deteriorated with higher interest rates or higher unemployment.
  - Slight unemployment increase in 2001-2002 was not accompanied by a change in NPL trend, possibly due to a supportive overall economic environment.
- Model specification used for sensitivity analysis of NPLs:
  - General form estimated: NPL_{i,t} = α_i + Σ_{s=0}^k β_s MACRO_{F,t-s} + ... (logit-transformed NPL, fixed effects α_i, MACRO factors with lags).
  - Specifications excluded bank-specific time-varying explanatory variables (other than lagged NPLs) to maintain parsimony and facilitate projections.
  - Lag structure explored with one and two lags given annual data frequency.
  - Estimation: first-differences using the Arellano-Bond estimator; macroeconomic variables treated as strictly exogenous from the perspective of individual credit institutions.

*Source: European Commission, Bank of Spain and IMF staff calculations (material in content unit 49–59).*

### 60.      The preferred specifications included GDP growth, the real growth of house

### Spain: Credit Portfolio Panel Models and Insurance Stress Tests (content unit)

### Bank loan-quality model specifications and estimation
- Preferred explanatory variables: GDP growth, the real growth of house prices, and the long-term interest rate.  
- Other macro variables considered: unemployment, household indebtedness, real disposable income per household, short- and long-term interest rates, and the real growth of house prices.  
- Variable transform: NPLs were logit-transformed as log(NPL/(1-NPL)).  
- Models: Panel estimations with differing explanatory-variable sets across loan-category equations to reflect idiosyncratic sensitivities.

### Regression results (dependent variable: logit of NPLs) — key coefficients and diagnostics (Table 11)
- Lagged autoregressive coefficients (LD. logit NPLs):
  - Developers: 0.582 [6.51]**
  - Commercial (Firm Loans): 0.660 [5.69]**
  - Mortgage Loans: 0.461 [3.49]**
  - Consumer Loans: 0.354 [3.71]**
- D. GDP Growth coefficients:
  - Developers: -0.091 [3.81]**
  - Commercial: -0.110 [6.29]**
  - Mortgage: -0.100 [4.28]**
  - Consumer: -0.082 [5.21]**
- D. Real Growth of House Prices coefficients:
  - Developers: -0.065 [4.91]**
  - Commercial: -0.018 [1.79]
  - Mortgage: -0.038 [2.45]*
  - (No entry shown for Consumer)
- D. Long-term Interest coefficients:
  - Developers: 0.028 [1.65]
  - Commercial: 0.040 [2.56]*
  - Mortgage: 0.103 [4.44]**
  - Consumer: 0.116 [5.70]**
- Observations and sample:
  - Developers: 942 observations; Number of banks 121
  - Commercial: 986 observations; Number of banks 130
  - Mortgage: 736 observations; Number of banks 98
  - Consumer: 850 observations; Number of banks 114
- Arellano-Bond tests:
  - AR(1) z = -4.9, -3.66, -2.22, -3.15 (for Developers, Commercial, Mortgage, Consumer respectively)
  - AR(2) z = -0.31, 0.44, -1.13, -1.37
- Significance notation: * significant at 5%; ** significant at 1%.  
- Source: Bank of Spain and IMF staff calculations.

### Projected loan-quality outcomes (two-year horizon, scenarios and mapping)
- Horizon and method:
  - Loan quality projected over a two-year horizon using panel results combined with macro scenarios.
  - NPL projections for each loan category and institution were mapped into PDs assuming a one-to-one relationship between percent changes in NPLs and percent changes in PDs.
  - Losses computed combining resulting PDs with stressed LGDs; losses measured incrementally in two ways: (a) vs PDs and LGDs observed in 2004; (b) vs projected baseline scenario.
- Scenario-driven NPL projections (selected outcomes):
  - Developers and construction (house price scenario): NPLs increased from 0.4 percent in 2004 to 4.9 percent in 2006.  
    - Note: 4.9 percent in 2006 is "still less than half the ratio observed in 1994."
  - Mortgages and commercial credit: main deterioration associated with the house price scenario, with more moderate effects than developers.
  - Consumer credit (oil scenario due to higher interest rates): NPLs grew from 2.2 percent in 2004 to 3.1 percent in 2006.

### Losses, profits comparison, and capital impacts
- Yearly-equivalent losses: cumulative two-year losses were converted to yearly equivalents (dividing cumulative losses by two) before comparison with after-tax profits.
- Average-system ability to absorb losses: On average, yearly losses under the three scenarios were below after-tax profits for sampled institutions.
- Distributional variation:
  - Under the house price scenario:
    - Yearly average losses represented 70.1 percent of after-tax profits for the entire sample.
    - Yearly average losses represented 116.2 percent of after-tax profits for the cajas segment.
    - For one small bank (one percent of assets in the sample) yearly losses represented 459.5 percent of profits.
  - Parallel but less severe results under the other two stressed scenarios.
- Capital adequacy (CAR) impacts:
  - House price scenario: average losses equivalent to 11 percent of regulatory capital (maximum impact 27.4 percent); average CAR falling to 11.3 percent at end of the two years.
  - Other scenarios: average losses ranging between 4.9 and 6.6 percent of regulatory capital; maximum values about 12–15 percent of regulatory capital.
  - A few small institutions (each less than one percent of sampled assets) would fall slightly below minimum CAR:
    - Five institutions accounting for 2.5 percent of assets below minimum CAR under the house price scenario.
    - Three institutions accounting for 2 percent of system assets below minimum CAR under the oil price scenario.
  - Losses higher among cajas, reflecting 26 percent exposure to real estate developers and construction companies at end-2004 (vs 20 percent for banks).
- Conclusion: The banking system appeared well prepared to undergo the simulated scenarios without a systemic threat to capital adequacy.
- Table 12 (stress-test results summary): results reported in percent (detailed table entries and subgroup splits reported in the source).

### Insurance sector: valuation, stress-test design, and summary results
- Fair value revaluation:
  - Fair value estimation of all assets and liabilities increased net assets for the whole industry by 82 percent.
  - Individual institution increases ranged between 8 and 360 percent.
  - Revaluation of real estate assets had the largest impact.
- Guidelines for valuation and stress testing (selected rules):
  - Reporting date: values based on December 31, 2004 (or December 31, 2003 if end-2004 data unavailable).
  - Liabilities should reflect a market-consistent assessment; unit-linked liabilities valued by reference to assets covering them.
  - Two valuation methods for life-insurer liabilities: discounting with the full term structure (default) or the duration approach (fallback).
  - Investment assets reported on market-value basis where possible; mark-to-model allowed for non-regularly traded assets.
  - Single events to consider (in isolation): (a) interest rate risk; (b) credit risk; (c) market risk (equity, real estate, foreign exchange); (d) underwriting risk; (e) lapse risk; (f) non-life risk.
  - Interest-rate shocks: both rise and fall in the term structure considered.
  - Credit-risk default method broadly aligned with the standardized approach of Basel II; credit spreads allowed as an alternative.
  - Underwriting risk: stress by changing survival or sickness tables by 15 percent.
  - Lapse risk: calculate effect of an increase in lapse rate of 50 percent.
- Aggregate stress-test findings (system-wide and by insurer type):
  - System-wide:
    - Largest impact: fall in real estate prices of 17 percent, reducing insurers’ capital by 5.6 percent.
    - Life underwriting risk: second largest impact, a fall of 4 percent.
    - Largest interest-rate shock effect: parallel fall in the curve of 200 basis points reduces capital by 1.3 percent.
    - Substantial variation across institutions (dispersion measures reported in Table 13).
  - Life insurers:
    - Downward parallel shift of yield curve by 200 basis points reduces capital by 24.8 percent.
    - Parallel increase of yield curve by 200 basis points improves capital by 8.1 percent.
    - Credit shock reduces capital by 9.9 percent.
    - Real estate shock reduces capital by 3.9 percent.
  - Non-life insurers:
    - Parallel upward shift of 200 basis points reduces capital by 10 percent (explained by large share of fixed-income securities).
    - Fall in real estate prices reduces capital by 7.1 percent.
  - Mixed insurance companies:
    - Fall in real estate prices reduces capital by 5.6 percent.
    - Interest-rate shocks have limited impact (netting effect on assets and liabilities).
    - Credit shock reduces capital by 1.5 percent.
- Overall conclusion: The insurance system is resilient to a wide range of shocks, supported by improved asset-liability management and a conservative investment portfolio.

*Source: Bank of Spain and IMF staff calculations; Dirección General de Seguros y Fondos de Pensiones (DGSFP) note as summarized in the source document.*

### 84.      The second set of stress tests dealt with insurance-specific risk factors. These

### _cr06216 - 84.      The second set of stress tests dealt with insurance-specific risk factors. These

### Insurance-specific stress tests: design and shocks
- Tests included:
  - “Underwriting risk.” Firms evaluated the more severe effect for each risk group of either an increase or decrease of the factors in the survival or sickness tables by 15 percent.
  - “Lapse risk.” Firms applied a 50-percent increase in their historical surrender rates.
  - “Non-life risk” organization:
    - Group 1: “mass events” (high probability, relatively small claims; includes all risks not in Group 2).
    - Group 2: risks with a “significant” probability of large claims (marine, aviation and transport; fire; third party motor liability; third party liability; credit; surety; and industrial multi-risk).
    - Group 2 subcategories: claims below €30,000 and claims equal to or above €30,000.
  - Shocks applied by group:
    - Increase of 10 percent in the average cost of claims applied to Group 1 and Group 2 claims below €30,000.
    - For Group 2 claims equal to or above €30,000, an increase of 15 percent in the frequency of claims.

### Key stress-test results and impacts on insurer capital
- Aggregate and sector-level impacts:
  - For the sector as a whole, the largest impact from the insurance-specific shocks would come from underwriting risk, which would result in a fall of 4 percent in capital.
- By insurer type:
  - Life insurers: largest impact on capital is a decrease of 18 percent resulting from an increase of 15 percent in the mortality rates.
  - Non-life insurers: largest impact on capital is a decrease of 7.3 percent resulting from a 10-percent increase in the average claim in the case of “mass events”.
  - Mixed insurance companies: largest impact on capital is a decrease of 1.4 percent, coming from an increase in the average claim of peak events.
- Table 13 selected figures (Impact on Insurer’s Capital, December 2004; results presented in percent of insurer's capital; a positive number indicates a reduction in capital):
  - All Institutions — Underwriting risk: Total Avg. 4.0; Std. Dev. 7.1; Max. 15.1; Min. -14.0.
  - Life Institutions — Underwriting risk: Total Avg. 18.2; Std. Dev. 18.2; Max. 23.0; Min. -62.2.
  - Non life institutions — Non life risk (group 1): Total Avg. 7.3; Std. Dev. 3.6; Max. 4.1; Min. 11.5.
  - Non life institutions — Non life risk (group 2, "mass" events): Total Avg. 4.7; Std. Dev. 4.6; Max. 3.3; Min. 11.4.
  - Non life institutions — Non life risk (group 2, "peak" events): Total Avg. 4.6; Std. Dev. 3.2; Max. 2.0; Min. 5.7.
  - Mixed institutions — Non life risk (group 1): Total Avg. -1.4; Std. Dev. -1.6; Max. 9.4; Min. 2.8.
- Note included in table: The results are presented in percent of insurer's capital (a positive number indicates a reduction in capital).
- Footnotes in table:
  - 1/ Group 1 refers to those claims related to "mass events" (i.e. high frequency but low claims).
  - 2/ Group 2 covers the following risk categories: marine, aviation and transport; fire; third party motor liability; third party liability; credit; surety; and industrial multi-risk. Group 2 is divided into two subcategories: group 2 “mass events”, which corresponds to claims under 30,000 Euro, and group 2 “peak events”, which corresponds to claims equal to or above 30,000 Euro.

*Source: _cr06216 - excerpt on insurance-specific stress tests and Table 13 (Spain: Summary of Stress Test Results for Insurance Firms, Impact on Insurer’s Capital, December 2004).*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/scr/2006/_cr06216.pdf_