## Spain FSAP (1espea2024004-print-pdf)

## Source details

**Canonical URL:** [Spain FSAP (1espea2024004-print-pdf)](https://www.imf.org/-/media/files/publications/cr/2024/english/1espea2024004-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/cr/2024/english/1espea2024004-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/cr/2024/english/1espea2024004-print-pdf.pdf.json)

---

### Executive summary
- Spain’s economy and bank-dominated financial system have shown resilience through the pandemic, rising global geo-political tensions and tighter financial conditions.
- Growth projection: average of 1.9 percent in 2024 (April 2024 WEO).
- Headline inflation rebounded to above 3 percent; core inflation expected to stay above the ECB’s target until early-to-mid 2025.
- Unemployment reached below 12 percent (post-GFC low).
- Adverse scenario (2024–2026): cumulative loss of over two standard deviations of real GDP growth over 2024-25; three-year horizon.
- Key systemic downside risks: escalation of geopolitical tensions; commodity (especially energy) price volatility; supply chain bottlenecks; higher inflation; greater likelihood of recession; tighter domestic monetary and financial conditions.
- Banks’ profile: deposit-funded credit intermediaries, low intra-financial exposures, significant foreign subsidiary business, strongly profitable, ample liquidity buffers.
- Capitalization: aggregate CET1 ratio for Significant Institutions (SIs) 12.6 percent as of June 2023; lower than Euro Area peers due to higher RWA density and lower voluntary buffers.
- FSAP bank stress test adverse scenario (2024-2026) headline:
  - Aggregate CET1 ratio of ten SIs declines to 9.6 percent in year 2 and recovers to just over 10 percent in year 3.
  - Two SIs’ CET1 capital fall below their Maximum Distributable Amount (MDA) trigger; one SI breaches the leverage ratio requirement.
  - Three-year cumulative credit impairments are 135 percent of starting CET1 capital by end-2026 in the adverse scenario (baseline: 54 percent).

### Real estate markets (RRE and CRE)
- Residential real estate (RRE) activity and prices:
  - Housing transaction volumes fell by 13.4 percent on average during the first three quarters of 2023.
  - Fixed-rate mortgage share: around 30 percent by 2023Q2 (see HH sector note).
  - Cumulative RRE price increase since 2019Q3 in line with euro area average; real RRE prices remain significantly below pre-GFC levels.
  - New dwellings: price appreciation exceeded 10 percent year-on-year in 2023Q3.
  - Box 1 finding: new dwelling prices show an unexplained component of 3.5 to 4 percentage points relative to the valuation model.
  - Model-based misalignment estimates:
    - Pure demand-based overvaluation: ~6 percent in 2022Q1 → 2.1 percent in 2023Q3.
    - With credit cycle and supply constraints: around 0.9 percent in recent two quarters.
- Commercial real estate (CRE):
  - CRE prices remain below pre-pandemic levels as of 2023Q3 across most segments; office buildings are the only segment with prices increasing.
  - CRE transaction volumes and new mortgage credit growing but slower than 2021–22.
  - Banking exposures to CRE small relative to European peers; trend decrease in banks’ CRE exposures since post-GFC.

### Public debt and sovereign exposures
- Pandemic fiscal measures increased public debt by around 25 percentage points.
- Debt-to-GDP ratio decreased to 107 percent in December 2023.
- As of June 2023:
  - Spanish banks held around 15 percent of Spanish sovereign debt securities.
  - Domestic NBFI sector held an additional 11 percent.
  - Banks’ public debt-to-assets ratio stable at 11 percent.
  - Sovereign debt shares by sector: Insurance firms 44 percent; Pension funds 30 percent; Open-ended funds (OEF) 30 percent.
- As of June 2023 (excluding BdE’s holdings): Spanish resident banks hold 14 percent of outstanding Spanish sovereign securities; foreign entities account for 42 percent of outstanding volume.
- FSAP finding: banks with initial government-debt-holdings-to-total-assets 10 percentage points higher experienced a decline of 0.7 percentage points in ROE after sovereign distress (CISS ≥ 0.48).

### Household sector vulnerabilities and methodology
- Deleveraging since GFC strengthened HH resilience to interest-rate shocks.
- Data and methodology:
  - Micro data from BdE’s Survey of Household Finances (EFF2020).
  - Vulnerability measures:
    - Debt-service-to-income (DSTI) >= 40.
    - Debt-service-and-essential-consumption-to-income (DSECTI) >= 70.
  - Simulation horizon: apply cumulative macro changes from 2020 to 2024 and 2020 to 2026.
  - Key macro assumption table (percent differences vs 2020; unemployment and interest-rate changes are percentage points):
    - 2022: Income +8.1; Unemployment rate -2.7; Mortgage interest rate 0; Other household loan interest rate 0; Food price +36.7; Energy price +226.6; CPI +11.4; Household debt +2.6.
    - 2024 baseline: Income +16.4; Unemployment rate -3.9; Mortgage interest rate +4.3; Other household loan interest rate +2.6; Food price +24.7; Energy price +91.7; CPI +18.7; Household debt +0.3.
    - 2024 adverse: Income +16.4; Unemployment rate +2.7; Mortgage interest rate +6.9; Other household loan interest rate +4.5; Food price +37.9; Energy price +131.5; CPI +21.7; Household debt -1.7.
    - 2026 baseline: Income +22.8; Unemployment rate -4.5; Mortgage interest rate +4.1; Other household loan interest rate +2.5; Food price +24.9; Energy price +74.1; CPI +22.5; Household debt +0.9.
    - 2026 adverse: Income +20.5; Unemployment rate +5.1; Mortgage interest rate +5; Other household loan interest rate +2.8; Food price +47.1; Energy price +136.8; CPI +25.8; Household debt -6.4.
- Household simulation results (selected exact figures):
  - Average DSTI for borrowers:
    - Baseline: 14.9 in 2024 and 14.3 in 2026.
    - Adverse: 15.6 in 2024 and 16.4 in 2026.
  - Share of HHs with DSTI >= 40: increases marginally by an average of 0.8 percentage points in 2024 and 2026 under the adverse scenario relative to baseline.
  - Cost-of-living adjusted vulnerability (DSECTI >= 70):
    - Baseline already-higher-than-peers level: 26 percent.
    - Adverse scenario: increases to 34 percent in 2024.
    - Share of vulnerable HHs would increase by 8.5 percentage points in 2024 and 13.7 percentage points in 2026 under the adverse scenario.
  - Distributional concentration: vulnerability concentrated in lowest-income quintile (HHs earning less than 1150 euro in 2019); this quintile holds only a small share of total household debt, limiting aggregate HH debt risk increase.

### Nonfinancial corporates (NFCs): vulnerabilities and stress-testing
- Data and coverage:
  - Central Balance Sheet Data Office Microdata; annual statements 1995–2022; around 950,000 firms per year; sample represents 51 percent of banks’ gross loan portfolio.
- Key metrics and trends:
  - NFC debt-to-GDP: fell to 82 percent in 2023Q3.
  - Interest Coverage Ratio (ICR) improved over the decade: ICR increased from 2.5 in 2012 to close to 6 for large firms and 7.5 for SMEs.
  - Cash-to-asset ratios end-2022: above 20 percent for large firms and SMEs in most sectors.
- Stress test outcomes:
  - Share of firms with ICR < 1 rises by around 6 percentage points in adverse vs baseline within a year; remains 5pp above baseline at end-2025.
  - Share of NFC debt with ICR < 1 increases by 0.9 percentage points vs baseline, reaching around 29 percent of total debt of these firms by end-2024.
  - Econometric evidence: primary driver of increased NFC debt distress is the fall in economic growth.
  - NFC liquidity problems peak by 2025 (liquidity metric = Cash and Cash Equivalents / Short-Term Liabilities).

### Banking sector structure, profitability, and vulnerabilities
- Size and business model:
  - Aggregate assets of Spanish credit institutions: 285 percent of GDP in June 2023.
  - SIs hold almost 95 percent of bank sector assets.
  - Loans account for 65.4 percent of total assets as of June 2023; debt securities 14.2 percent.
  - Household mortgages average maturity: 14.7 years.
  - As at Sept-2023, 63 percent of outstanding loans to HHs and 57 percent to NFCs were at floating rates.
- Funding composition (June 2023):
  - Deposits represent 77.5 percent of total non-equity liabilities for SIs.
  - 56 percent of liabilities are customer deposits from HHs and NFCs.
  - Almost half of deposits came from Spanish depositors; 59 percent of securities issued were held by Spanish counterparts.
- Profitability and capital:
  - ROE 12.1 percent and ROA 0.76 percent in June 2023.
  - CET1 ratio aggregate SIs: 12.6 percent as of June 2023.
  - Provisioning coverage of NPLs over 46 percent for resident private sector by June 2023.
  - System-wide NPL ratio: 3.4 percent at end-2023 H1.
  - Liquidity buffers: LCR for SIs 182 percent in June 2023; SIs with material international presence LCR 153 percent.
- Cross-border footprint:
  - Consolidated foreign claims of Spanish banks on top 10 counterparties: US$1.8 trillion in 2022Q4; US$1.2 trillion associated with Brazil, Mexico, the United Kingdom, and the United States.
  - Foreign banks’ exposures to Spain: US$ 450 billion in 2023Q2.

### FSAP solvency and market-risk stress test essentials
- Solvency stress test scope and cut-off:
  - Ten SIs included (three largest banks hold over 80 percent of banks assets; 10 SIs nearly 95 percent).
  - Cut-off date September 30, 2023.
  - Stress horizon: 2023 Q3–2026 Q4 (three years, one quarter).
- Adverse macro scenario (selected Spain values, percent change unless otherwise indicated):
  - 2024: Real GDP growth -4.09; Core Inflation 6.77; Unemployment 14.03; Short term sovereign yield 5.89; Long term sovereign yield 7.73; Sovereign spread over Germany 2.06; House Price -17.16; Credit Growth -2.06.
  - 2025: Real GDP growth -3.10; Core Inflation 2.46; Unemployment 18.61; Short term sovereign yield 4.80; Long term sovereign yield 6.35; Sovereign spread over Germany 1.71; House Price -3.29; Credit Growth -0.22.
  - 2026: Real GDP growth 1.00; Core Inflation 0.86; Unemployment 19.84; Short term sovereign yield 3.71; Long term sovereign yield 3.85; Sovereign spread over Germany 1.20; House Price 9.70; Credit Growth 1.32.
- Credit risk methodology:
  - PDs: domestic via panel Bayesian Model Averaging (BMA); foreign via Moody’s EDF anchoring.
  - LGDs: mortgages via structural LTV-based model; unsecured via Frye-Jacobs calibration; Vašíček linkage for advanced economies.
  - Transition matrices from BdE national credit registry for domestic portfolios.
- Interest rate pass-through (domestic betas):
  - Deposit sight Beta 0.06; Deposit term Beta 0.27; Mortgage loans Beta 0.78; Consumption loans Beta 0.6; Household sight deposits Beta 0.49; Household term deposits Beta 0.48; NFC loans (all) Beta 0.73.
- Market risk and securities valuations:
  - 40 percent of SIs’ sovereign debt portfolio held at fair value in June 2023.
  - Unrealized losses amounted to EUR 26 billion as of June 2023; reach EUR 78 billion in adverse scenario (triple the initial amount).
  - Majority of bond portfolio at amortized cost (close to 60 percent); estimated unrealized losses on Spanish sovereign exposures median EUR 2 billion initially, EUR 4 billion in adverse scenario.
- Interest rate effects on NII:
  - NII grows in first year of adverse scenario due to faster pass-through on assets vs liabilities; attenuates in later years as negative credit growth and rising NPEs dominate.
- Solvency stress test results (aggregate):
  - Baseline: aggregate CET1 ratio rises from 12.6 percent (Q3-2023) to 15.4 percent at end-2026.
  - Adverse: aggregate CET1 ratio declines up to 3 percentage points to 9.6 percent at end-2025, recovering to just over 10 percent by end-2026.
  - Three-year cumulative credit impairments: 135 percent of starting CET1 in adverse; 54 percent in baseline.

### Liquidity stress testing, cash-flow analysis, and reverse stress tests
- LCR and NSFR:
  - Aggregate LCR for 10 SIs falls below 100 percent under “outflow” and “aggressive” scenarios.
  - Number of banks below 100 percent LCR by scenario:
    - Basel III: 0
    - Market: 0
    - 25% Outflow: 0
    - 50% Outflow: 1
    - 75% Outflow: 2
    - 100% Outflow: 6
    - Aggressive: 7
  - Aggregate NSFR for 10 SIs: 129 percent (June 2023).
- Cash-flow (CF) analysis:
  - 25 scenario grid across haircuts, run-off rates, and roll-off rates; one-week and three-month horizons.
  - Under aggressive scenario: aggregate marginal liquidity surplus over one-week horizon; small shortfall over three-month horizon.
  - Three banks experience marginal one-week shortfall at aggressive severity; four banks for three-month horizon.
  - Distance to Liquidity Stress Indicator (DLSI / d):
    - Only two banks have negative DLSI for one-week horizon; one bank has negative DLSI for three-month horizon.
- Forced liquidation (if denied standing facilities and ELA):
  - Unencumbered EA sovereign securities available for sale: EUR 105 billion (39 percent of bond portfolio for subset of banks).
  - Median CET1 decline from potential sale losses: between 1.3 and 3.5 percentage points.
  - First quartile decline: 4.1 pp; third quartile decline: 1.7 pp (adverse scenario).
  - Partial liquidation (half portfolio): median CET1 decline 0.2 pp (initial) and 0.5 pp (adverse).
- Funding and counterbalancing:
  - Deposits 77 percent of bank liabilities (June 2023).
  - Stable retail funding 48 percent of deposit mix.
  - Counterbalancing capacity (CBC) major components: central bank reserves 31 percent; level 1 assets 32 percent.
  - Asset encumbrance 17 percent in 2023-Q2 (down from 23 percent end-2021).

### Interconnectedness, contagion analysis, and domestic interbank channels
- Cross-border contagion (network model):
  - Model applied across 11 banking centers; BIS consolidated statistics to 2023-Q2.
  - Spanish banks’ vulnerability to cross-border banking distress modest: average potential loss ≈ 1.2 percent of current total capital if cross-border claims fully forfeited under assumptions.
  - Losses from distress in some European countries could reach up to 4 percent of bank capital; post-shock capital adequacy remains above 14 percent.
- Domestic interbank contagion:
  - Domestic interbank losses concentrated within SIs: SIs→SIs account for 72.3 percent of total losses.
  - Credit channel accounts for 3/4 of total losses; funding channel the remainder.
  - Cross-border intragroup share: ~44 percent of Spanish banks’ cross-border claims (2023Q2), lower than British (71.5 percent) and German (67 percent).

### Key risks (RAM highlights)
- Intensification of regional conflict(s) and geo-economic fragmentation: Overall Level of Concern High; Relative Likelihood Medium.
- Abrupt global slowdown or recession leading to systemic financial instability: Overall Level of Concern Medium; Relative Likelihood High.
- Commodity price volatility: Overall Level of Concern High; Relative Likelihood Medium.
- Monetary policy miscalibration: Overall Level of Concern Medium; Relative Likelihood High.
- Weak implementation of fiscal commitments or reassessment of sovereign risk: Overall Level of Concern Medium; Relative Likelihood High.
- Extreme climate events and Cyberthreats: Overall Level of Concern Medium; Relative Likelihood Medium (climate) / High (cyber).

### Policy recommendations (summary of Table 1)
- Enhance data collection and monitoring on foreign investments in the real estate market. Timing: NT. Agency: BdE, CNMV.
- Create infrastructure for more granular cash-flow analysis (as designed by the FSAP) and report regular stress testing results for financial stability analysis and monitoring purposes. Timing: NT. Agency: BdE.
- Integrate interconnectedness and contagion analysis more systematically into BdE’s financial stability toolkit. Timing: MT. Agency: BdE.
- Timing notation: “Immediate” = within one year; “NT-near-term” = 1–3 years; “MT-medium-term” = 3–5 years.

*Source: EXECUTIVE SUMMARY and chapters (IMF staff analysis) of 1espea2024004-print-pdf.*

### EXECUTIVE SUMMARY __________________________________________________________________________ 7

### EXECUTIVE SUMMARY

### Overview
- Spain’s economy and its well-developed, bank-dominated financial system have shown resilience through the pandemic, rising global geo-political tensions and tighter financial conditions.
- The economy remains near potential and growth is projected to continue its robust performance in the coming quarters.
- Long running deleveraging by households (HHs) and nonfinancial corporates (NFCs) continues; housing market very moderate overvaluation is receding; commercial real estate valuations remain below pre-pandemic levels; foreign investments in the real estate market are on the rise.

### Macroeconomic outlook and major risks
- Growth projection: average of 1.9 percent in 2024 (April 2024 WEO).
- Headline inflation rebounded to above 3 percent; core inflation expected to stay above the ECB’s target until early-to-mid 2025.
- Labor market: unemployment rate reached a post-GFC low at below 12 percent.
- Downside risks: further escalation of elevated global geopolitical tensions, renewed volatility in commodity (especially energy) markets, supply chain bottlenecks, higher inflation, greater likelihood of recession, and tighter domestic monetary and financial conditions.
- If downside risks materialize globally, banks’ international diversification benefits may be unavailable and banks may be compelled to deleverage to protect solvency.
- Adverse scenario scope: three-year horizon (2024-2026) with a significant and persistent recession entailing a cumulative loss of over two standard deviations of real GDP growth over 2024-25 and gradual recovery in 2026.

### Household (HH) sector vulnerabilities
- Deleveraging since the GFC strengthened HH resilience against the surge in interest rates.
- Fixed-rate mortgage share: around 30 percent by 2023Q2.
- Vulnerability concentrated among low-income households when assessing debt servicing costs; these HHs hold a small share of overall HH debt, limiting aggregate impact.
- When combining the impact of rising interest and essential consumption expenses, low-to-middle-income HHs could find it challenging to meet rising borrowing costs.
- The limited take-up so far of end-2022 government support measures for families affected by the increase in the Euribor points to resilience of HH balance sheets.

### Nonfinancial corporates (NFC) vulnerabilities
- NFC debt-to-GDP: fell to 82 percent in 2023Q3.
- NFCs rank among the least indebted corporate sectors compared to European peers.
- Firms’ interest coverage ratio (ICR) has increased through most of the previous decade, rebounding after a partial reversal during the pandemic.
- FSAP adverse scenario projects a moderate rise in debt-at-risk for NFCs: the share of firms that may face challenges meeting debt servicing costs out of earnings is projected to increase materially compared to the baseline by 2026.
- Firms with low initial debt servicing capacity are more susceptible; the share of such firms rises by over 6 percentage points in the adverse scenario relative to the baseline by end-2024.

### Banking sector structure, profitability, and vulnerabilities
- Spanish banking sector: global imprint, traditional business model, strongly profitable.
- Banks are deposit-funded credit intermediaries with low intra-financial system, and limited intra-group, exposures.
- Significant share of business and earnings of the largest banks emanates from foreign subsidiaries.
- Profitability surged post-pandemic as financial impairments receded and lending margins rose with interest rates; banks increased dividend payouts and accelerated share buybacks while holding solvency buffers steady.
- Capitalization of Spain’s significant banking institutions (SIs) is lower than Euro Area (EA) peers owing to lower voluntary buffers and higher density of risk weighted assets.
- Liquidity buffers are ample.

### FSAP bank stress test results (adverse scenario, 2024-2026)
- Adverse global scenario features energy price spikes, monetary tightening, and lower economic activity with negative trade and financial spillovers.
- Aggregate SI solvency outcome:
  - Two SIs’ CET1 capital fall below their Maximum Distributable Amount (MDA) trigger; one SI breaches the leverage ratio requirement.
  - Total loss of capital across the ten SIs is large: the aggregate CET1 ratio of the ten SIs declines to 9.6 percent in year 2, 1.2 percentage points above regulatory requirements, and recovers to just over 10 percent in year 3.
  - On aggregate-average basis, the 10 SIs’ capital levels remain above regulatory requirements, reflecting substantial and rapid credit deleveraging for domestic exposures undertaken by banks under the scenario.
- Credit losses persist through the three years of the adverse scenario, with reduced severity as recovery begins in 2026.
- Large credit losses and low incumbent voluntary capital buffers are important factors explaining breaches of regulatory hurdles under stress.
- Evidence of a weak tail of SIs despite aggregate resilience.

### Liquidity stress testing
- SIs can cope comfortably with market valuation shocks.
- SIs would face cash flow challenges under large withdrawals of retail deposits.
- Resilience bolstered by substantial share of retail deposits and large reserves of high-quality liquid assets (HQLA).
- Starting positions of short- and long-term stable funding are strong.
- LCRs are relatively insulated from market-valuation shocks and are primarily susceptible to significant (retail) deposit outflows which, beyond a high threshold, would outstrip cash inflow and counterbalancing capacity.

### Interconnectedness and cross-border contagion
- Interconnectedness analysis does not reveal significant vulnerabilities of Spanish banks to cross-border contagion of foreign banking distress.
- The capital ratio of Spanish banks remains, in the aggregate, comfortably above regulatory thresholds under the full range of adverse shocks, reflecting low intra-financial and interbank exposures of Spanish banks.

### Policy recommendations (Table 1 summary)
- Enhance data collection and monitoring on foreign investments in the real estate market. Timing: NT. Agency: BdE, CNMV.
- Create the infrastructure for a more granular cash-flow analysis (as designed by the FSAP) and report regular stress testing results for financial stability analysis and monitoring purposes. Timing: NT. Agency: BdE.
- Integrate the interconnectedness and contagion analysis more systematically into the toolkit of BdE’s financial stability framework. Timing: MT. Agency: BdE.
- Timing notation: “Immediate” is within one year; “NT-near-term” is 1–3 years; “MT-medium-term” is 3–5 years.

*Source: EXECUTIVE SUMMARY, 1espea2024004-print-pdf.*

### 3.      Prices are growing steadily in the residential real estate (RRE) market even as activity

### 3.      Prices are growing steadily in the residential real estate (RRE) market even as activity

### Residential real estate activity and prices
- Housing transaction volumes fell by 13.4 percent on average during the first three quarters of 2023.
- Unlike other euro area countries, Spain did not experience a large house price boom in 2020-21.
- Nominal and real RRE prices in Spain have continued to register small increases, supported by large immigration flows and supply shortages.
- The cumulative increase of RRE prices in Spain since 2019Q3 is in line with the average in the euro area.
- RRE prices in real terms remain significantly below pre-GFC levels.
- Lending standards have remained prudent:
  - Relatively low average LTV ratios.
  - Declining shares of loans with LTV ratios higher than 80 percent.
- There has been an increasing share of housing transactions that involve non-mortgage buyers, which warrants close monitoring.
- The new dwellings segment showed price appreciation that exceeded 10 percent year-on-year in 2023Q3.

### Box 1 — New dwellings: valuation and unexplained price component
- Applying the same house price valuation model (including the same demand- and supply-side variables) to new dwelling prices results in an unexplained component of 3.5 to 4 percentage points.
- Interpretation:
  - This suggests a greater extent of price misalignment for new houses compared to the overall market.
  - The large unexplained component could be driven by the model not incorporating foreign buyers, or by aggregate indicators not adequately capturing new-housing-specific supply constraints.
- Recommendation:
  - Continue analyzing and monitoring the new dwellings segment.

### House price valuation and misalignment estimates
- Price-to-income ratio has been stable since end-2019 and declined moderately recently following robust household income growth.
- Price-to-rent ratio has been continuously increasing, reflecting:
  - Rent index limitations (does not fully capture new rental contracts).
  - Limits on rent increases at contract renewals arising from authorities’ anti-inflation measures.
- Both ratios remain significantly lower compared to pre-GFC levels.
- Model-based misalignment estimates:
  - Pure demand-based assessment suggests house price overvaluation relative to fundamentals dropped from around 6 percent in 2022Q1 to 2.1 percent in 2023Q3.
  - Incorporating the credit cycle and supply constraints reduces price misalignment further to around 0.9 percent in the recent two quarters.
- Implication:
  - Looking forward, nominal house price growth in the baseline should be moderately lower than headline inflation.

### Commercial real estate (CRE)
- CRE prices remain below pre-pandemic levels as of 2023Q3 across most segments; office buildings are the only segment with prices increasing.
- CRE transaction volumes and new mortgage credit are growing, but at a much slower pace compared to 2021–22.
- Tightening lending standards and accelerated digitalization could impose further downward pressure on CRE prices.
- Banking sector exposures and resilience:
  - Downside risks are expected to have limited impact on banks given:
    - The small size of the Spanish CRE market relative to European peers.
    - A trend decrease in banks’ CRE exposures following stringent post-GFC lending standards.
    - Relatively high NPL ratios on banks’ CRE portfolios are partly driven by legacies from previous crises.
- Data and monitoring recommendation:
  - Increased reliance on foreign funding by the CRE sector points to the importance of enhancing data collection on foreign investment in the CRE market, potentially in collaboration with European institutions and foreign jurisdictions.

### Public debt and sovereign exposures
- Public debt dynamics:
  - Fiscal measures during the pandemic increased public debt by around 25 percentage points.
  - Debt-to-GDP ratio decreased over the next two years, reaching 107 percent in December 2023.
- Financial sector sovereign exposures:
  - Spanish banks held around 15 percent of Spanish sovereign debt securities in June 2023.
  - The domestic NBFI sector held an additional 11 percent.
  - Banks’ public debt-to-assets ratios have been stable at 11 percent but remain among the most exposed to sovereign risk across European peers.
  - Sovereign debt shares of other financial sectors:
    - Insurance firms: 44 percent.
    - Pension funds: 30 percent.
    - Open-ended funds (OEF): 30 percent.
- Risk implication:
  - Adverse macro-financial implications could be higher if market sentiment shifts against highly indebted euro area countries.

### Banking sector structure and vulnerabilities
- System size and orientation:
  - Aggregate assets of Spanish credit institutions amounted to 285 percent of GDP in June 2023.
  - Spanish banks operate a universal model with a strong retail orientation.
  - Almost 90 percent of bank lending is channeled towards lending to mortgages and NFCs.
  - A significant share of household mortgage lending is at floating rates (70 percent in case of household mortgages).
  - Significant Institutions (SIs) hold almost 95 percent of bank sector assets.
- Funding composition (June 2023):
  - Deposits represent 77.5 percent of total non-equity liabilities for SIs.
  - 56 percent of liabilities are customer deposits coming from HHs and NFCs.
  - Central bank funding share dropped by 6.2 percentage points as a share of total assets between June 2022 and June 2023.
  - Geographic composition of SIs’ deposits and securities (June 2023):
    - Almost half of deposits came from Spanish depositors.
    - 59 percent of securities issued were held by Spanish counterparts.
- Asset side and loan composition:
  - As of June 2023, loans account for 65.4 percent of total assets of all banks; debt securities represent 14.2 percent.
  - Household mortgages have the highest share of loans, with an average maturity of 14.7 years.
  - As at Sept-2023, 63 (57) percent of the outstanding loans to HHs (NFCs) were at floating rates.
- Cross-border activities and subsidiaries:
  - The largest banks, including one G-SIB, have significant presence in Brazil, Mexico, Türkiye, the United Kingdom, and the United States.
  - International subsidiaries fund liquidity via local deposits and their own issuance programs, resulting in relatively low intragroup claims and liabilities compared to other major banking systems.
  - Note: intragroup reliance could increase in a crisis.
- Profitability and near-term risks:
  - Return on equity (ROE) stood at 12.1 percent, and return on assets (ROA) was 0.76 percent in June 2023.
  - Factors that could reduce profitability in coming quarters:
    - Rising funding costs (e.g., deposit betas, migration from sight to term deposits, or to alternative instruments).
    - Economic slowdown or materialization of downside risks leading to increased financial impairments.
  - These downside pressures could more than offset repricing benefits from existing variable rate loans and income from new loan production.
  - Internationally, Spanish banks have rebounded to pre-pandemic profitability levels in terms of ROA and ROE.

*Source: IMF staff analysis and figures contained in the chapter.*

### 12.      Monetary policy normalization has impacted, and may impact, SIs’ profitability and

### 12.      Monetary policy normalization has impacted, and may impact, SIs’ profitability and

### Channels of impact from monetary policy normalization
- Interest rate passthrough:
  - Pass-through on the asset side has been faster and, on average, more pronounced than on the liability side, contributing to increasing net interest margins.
  - More recent catch-up of deposit rates on all but sight contracts indicates this factor is now less significant.
  - A descriptive finding: "Interest rates have increased significantly since end-2021."
- Sovereign debt valuation:
  - A significant share of SIs’ sovereign debt portfolio is held at fair value: 40 percent in June 2023.
  - Higher interest rates have led to realized losses in assets valued at fair value.
- Credit quality channel:
  - Resilient economic performance has ensured that credit quality of SIs’ loans has not worsened.
  - Nonperforming loans (NPL) ratio: 3.4 percent at end-2023 H1.
  - Downside risks could change this trajectory.

### Profitability and capital of Significant Institutions (SIs)
- Profitability indicators and trends:
  - Net interest income and net income series shown for 2008-2023 indicate widening net interest margins (see referenced Figures 8 and 9).
  - Recent catch-up of deposit rates for new term deposits; passthrough lags for outstanding term and especially sight deposits.
- Capitalization:
  - CET1 ratio for aggregate Spanish SIs: 12.6 percent as of June 2023.
  - Improvement factors: retained earnings (after dividends and share buy backs) more than compensated for increase in RWA from foreign exposures and revisions of internal ratings models.
  - Spanish SIs’ CET1 ratios remain lower than EA peers owing to higher risk-weight density and lower voluntary buffers; banks favor high dividend payouts and buy-backs over retaining earnings.
  - Leverage ratios: comparable to most EA peers.
  - Capital depletion in the 2023 EBA stress test adverse scenario: Spanish SIs faced the least amount of capital depletion relative to peers.

### Asset quality, NPLs, and stage migration
- System-wide NPL and Stage 2 metrics:
  - NPL ratios have decreased since the end of the global financial crisis, falling by 3 percentage points in the five years through June 2023.
  - Highest sectoral NPL ratios by June 2023:
    - Hospitality, restaurant, and leisure sector: 7.5 percent.
    - Construction and real estate activities: 4.6 percent.
  - By company size (June 2023):
    - SMEs segment NPL ratio: 6.6 percent.
    - Microenterprises segment NPL ratio: 8 percent.
  - Share of Stage 2 loans (leading indicator of future NPLs): average 7.6 percent of lending to the private sector by 2023Q3.
  - Pre-pandemic comparison: Stage 2 share was 6.2 percent as of Dec 2019.
- Pandemic-related guarantee loans:
  - Outstanding loans under the ICO program exhibit a higher-than-average share of stage 2, stage 3, and NPLs (see Box 2).

### ICO (Official Credit Institute) loans – Box 2 summary
- Program scale and coverage:
  - Banks issued over €80 billion worth of loans to SMEs and autonomous institutions under ICO loss sharing (about 6 percent of total private sector credit).
  - ICO loss coverage in event of nonperformance:
    - SMEs and entrepreneurs: ICO covers 80 percent of the loss on outstanding principal of new and renewed loans.
    - Other NFCs: ICO covers 70 percent for new loans and 60 percent for renewed loans.
  - ICO does not cover loss of interest, fees, commissions, and recovery costs.
- Recent developments and impact:
  - Volume of ICO-backed credit declined by 10 percent in 2023Q3.
  - Share of stage 2 and stage 3 loans rose to 32 percent of total ICO credit.
  - The impact on banks is significantly attenuated by the ICO guarantee and the program is now closed.

### Provisioning and coverage
- Provisioning coverage:
  - SIs’ provisioning coverage of NPLs grew in the twelve months prior to June 2023, reaching over 46 percent for the resident private sector as a whole.
  - Coverage ratio patterns:
    - Coverage remains higher for corporate exposures than for households.
    - Coverage ratios are lower than the EU average for households (HH).
    - Coverage ratios started exceeding the EU average for nonfinancial corporations (NFCs) in 2023Q2.
  - Coverage ratio indicators shown for households and NFCs across 2019Q2–2023Q2 (referenced Figures).

### Liquidity and Less Significant Institutions (LSIs)
- System-wide liquidity buffers:
  - Liquidity Coverage Ratio (LCR) of SIs: 182 percent in June 2023 (well above minimum required threshold).
  - Spanish SIs with material international presence: LCR 153 percent in June 2023 (in line with European peers; still well above minimum requirements).
- LSIs profile and metrics:
  - Number of LSIs: 57 (including 35 credit cooperatives, 20 banks, 2 savings banks).
  - Share of Spanish banking assets held by LSIs: 5.5 percent.
  - Domestic financing orientation: 92 percent of counterparties are Spanish.
  - Aggregate fully loaded CET1 ratio (LSIs): improved y-o-y from 19.2 percent to 21 percent by June 2023.
  - Liquidity metrics (LSIs): LCR of 305 percent, NSFR of 162 percent as of June 2023.
  - Profitability: strong increase; ROA close to doubling y-o-y through 2023H1 due to widening interest margins.
  - LSIs exhibited low NPL ratios and performed well in the BDE’s 2023Q3 supervisory stress tests.
  - LSIs maintain conservative liquidity profiles, supported by stable retail deposits; low loan-to-deposit ratios and high volumes of liquid assets eligible for ECB monetary operations.
  - Overall liquidity position (in terms of LCR and NSFR) is well above regulatory ratios.

### Cross-border banking claims and interconnectedness
- Spanish SIs’ consolidated foreign claims:
  - Total consolidated foreign claims of Spanish banks on top 10 counterparties amounted to US$1.8 trillion in 2022Q4.
  - Of this, US$1.2 trillion foreign claims are associated with four countries: Brazil, Mexico, the United Kingdom, and the United States.
  - Distribution of counterparties (consolidated foreign claims): predominant counterparty type is NFCs at 58 percent, followed by public sector 25 percent, banks 10.5 percent, and NBFI 6.5 percent.
- Foreign banks’ exposures to Spain:
  - Foreign banks' exposures to Spain: US$ 450 billion in 2023Q2.
  - Around 26 percent of these foreign claims are directed towards the Spanish nonbank private sector; remaining claims distributed between public sector (41 percent) and banks (28 percent).
- Trends since crises:
  - International claims of Spanish SIs: steady increase since the EA sovereign debt crisis.
  - Foreign banks’ claims on Spain: steady decrease since the EA sovereign debt crisis, primarily due to reduced claims from other European countries.
- Business model and intragroup transfers:
  - Spanish SIs primarily use a subsidiary business model; foreign claims primarily take the form of loans by foreign subsidiaries (local claims).
  - Degree of intragroup transfers relatively limited compared to peers.
  - Cross-border intragroup share: approximately 44 percent of cross-border bank claims for Spanish banks are intragroup (2023Q2), lower than British banks at 71.5 percent and German banks at 67 percent.

*Source: IMF staff.*

### 22.      The key risks to financial stability in the current situation are geopolitical risks, a

### 1espea2024004-print-pdf - 22.      The key risks to financial stability in the current situation are geopolitical risks, a

### Key risks to financial stability
- Main risks: geopolitical risks, a higher and more persistent inflation, weaker exports and investment growth, and tighter financial conditions (Appendix I).
- Channels and potential consequences:
  - Further escalation of geo-political tensions, higher-for-longer interest rates, renewed volatility in commodity, especially energy, markets could:
    - create shortages in critical supply chain components,
    - further raise inflation,
    - increase the likelihood of recession,
    - lead to even tighter domestic financial conditions.
  - Increased risk aversion could generate negative macro-financial consequences, weighing on households, NFCs, the property market, and banks’ asset quality.
  - Domestic policy miscalibration or exogenous events, such as cyber threats, could amplify financial stability impacts.

### FSAP stress tests: adverse scenario and scope
- Scenario timing and alignment:
  - Scenarios span three years (2024-2026).
  - Baseline scenario underlying banking solvency stress tests aligned with the October 2023 World Economic Outlook projections.
- Adverse scenario characterization:
  - Two consecutive years of recession entailing a cumulative loss of 7 percentage points of real GDP over two years:
    - -4 p.p. in 2024 and -3 p.p. in 2025.
  - Modest recovery in 2026.
  - Shock magnitude: equivalent to 2.1 standard deviations in cumulative terms relative to the baseline over 2024-25, and 2.3 standard deviation from historical mean.
  - Scenario drivers: significant and persistent recession, negative spillovers across trade and financial channels, rising geopolitical concerns and related energy price spikes increase inflation, monetary tightening precipitates lower economic activity.
- Coverage:
  - Scenarios are common to all banks and cover all major jurisdictions where Spanish banks are active: Latin America (LatAm), Türkiye, the United Kingdom, and the United States.
  - Latin American countries in the sample: Argentina, Brazil, Colombia, Chile, and Mexico.
  - Scenario based on the Global Macro-Financial Model (GFM), disaggregated into forty national economies.

### Household (HH) vulnerability analysis: scope and methodology
- Data source:
  - Micro data from the BdE’s Survey of Household Finances (EFF).
  - Representative sample of Spanish HHs from EFF2020.
- Income and expenditure considerations:
  - Income side: labor and non-labor income.
  - Expenditure side: essential expenses on food, utilities, and rents for HHs who do not own their primary residence.
- Vulnerability measures (two alternative measures):
  - Debt-servicing based measure:
    - HH defined as vulnerable if debt-service-to-income (DSTI) ratio is greater than or equal to 40 percent (DSTI >= 40).
  - Cost-of-living adjusted measure:
    - HH defined as vulnerable when debt service payments and essential expenses on food, utilities, and rents exceed 70 percent of its income (>= 70 percent).
  - Note: role of assets taken into account only through interest payments and rents they generate; liquidation not contemplated.
- Simulation approach:
  - Simulate evolution of HH income statement using EFF2020 under baseline and adverse scenarios from bank stress tests.
  - For 2020–2022, simulation relies on actual outturns; for 2023 onwards generate results for baseline and adverse scenarios.
  - Appendix VIII details application of macro shocks to individual HH income statements.

### Household results and projections
- Baseline trajectory:
  - HH vulnerability expected to gradually improve under the baseline scenario from peak levels in 2022.
  - Drivers: income catching up with inflation in 2023, strong labor market, normalization of energy and food prices, HH deleveraging.
- Comparative position:
  - Spanish households remain relatively more vulnerable compared to European peers in the baseline scenario (Figure 15).
  - Simulation for European HHs using 2021 ECB HFCS suggests Spain’s share of HHs at risk by 2024 is on the higher end compared to other EU countries.
  - Although Spanish HHs accumulated higher financial assets relative to European peers overall, low-income HHs accumulated less, increasing vulnerability.
- Adverse scenario impacts (selected exact figures):
  - Average DSTI ratio for borrowers:
    - Baseline: 14.9 in 2024 and 14.3 in 2026.
    - Adverse: 15.6 in 2024 and 16.4 in 2026.
  - Share of HHs with DSTI >= 40:
    - Expected to increase marginally compared to baseline by an average of 0.8 percentage points in 2024 and 2026.
  - Cost-of-living adjusted vulnerability:
    - Share of vulnerable HHs would increase by 8.5 percentage points in 2024 and 13.7 percentage points in 2026 under the adverse scenario, reflecting elevated food and energy prices through 2026 in this scenario.
  - Distributional concentration:
    - Pockets of vulnerability concentrate among low-income HHs.
    - Only the lowest-income quintile (HHs earning less than 1150 euro in 2019) expected to see sizable increases in borrowers’ DSTI ratio and share of vulnerable HHs.
    - Lowest-quintile holds only a small share of total household debt, implying limited increases in aggregate HH debt risk.
  - Cost-of-living adjusted population shares:
    - Baseline already-higher-than-peers level: 26 percent.
    - Adverse scenario: increases to 34 percent in 2024.
    - Share of HH debt-at-risk would increase by 8.4 percentage points, with contributions from all income quintiles.

### NFC (Non-Financial Corporations) vulnerabilities: overview and metrics
- Data and sample coverage:
  - Source: Central Balance Sheet Data Office Microdata.
  - Annual financial and income statements for all corporations for 1995 to 2022.
  - Coverage: around 19 industries and around 950,000 firms a year, of which around 50,000 firms are large corporations and the rest are micro, small and medium enterprises (SMEs).
  - Entities in this sample represent 51 percent of the gross loan portfolio of banks, with large corporations accounting for two-thirds and SMEs contributing one-third.
- Sectoral concentration:
  - Large corporations predominantly clustered in professional activities, financial services, and manufacturing; minimal presence in accommodation and agriculture.
  - SMEs concentrated in retail, real estate, manufacturing, and professional activities; SMEs do not compete with large corporations in many sectors such as finance and infrastructure (electricity, gas, water).
- Corporate performance trends:
  - Profitability:
    - Consistent upward trend in NFC profitability over past 10 years, somewhat diminished by COVID-19.
    - Pre-COVID: large corporations outperformed smaller ones; post-COVID smaller firms generating more profits, surpassing pre-COVID earnings.
  - Liquidity and repayment capacity:
    - Liquidity indicators improved post European debt crisis.
    - SMEs typically maintain higher levels of liquid assets compared to large corporations.
    - Cash-to-asset ratios as of end-2022 remain above 20 percent for large firms and SMEs in most sectors.
    - Interest Coverage Ratio (ICR) increased from 2.5 in 2012 to close to 6 for large firms and 7.5 for SMEs.
  - Leverage:
    - Debt-to-asset and debt-to-equity ratios have been continuously declining since the European debt crisis.
    - Debt ratio defined as (current liabilities + non-current liabilities)/total liabilities.
    - Debt ratio at risk exhibits similar declining trend for SMEs and large firms, with greater cross-sector diversity; extractive industries, utility, and wholesale and retail sectors more susceptible.
- Liquidity and solvency metrics highlighted:
  - Cash Availability Ratio defined as (Cash and Cash Equivalents) / (Short-Term Liabilities or Current Liabilities).
  - Debt ratio defined as the sum of current and non-current liabilities to total assets.

### NFC stress-testing methodology and implications
- Solvency stress test aims:
  - Compare scenarios related to NFCs' ability to manage debt obligations and borrowing needs using bank solvency stress test scenarios and data till end-2022.
- Key thresholds and indicators:
  - Interest Coverage Ratio (ICR) used to assess debt servicing capacity:
    - ICR < 1: firm not generating sufficient revenue to service debt; likely need corrective action or face risk of nonperformance.
    - ICR between 1 and 2: firms could potentially face difficulties in servicing debt if financial conditions worsen.
  - Liquidity problems proxied by a firm’s cash ratio turning negative, indicating need to raise debt to match incoming and outgoing cash flows.
- Exercise outcomes focus on:
  - Number of firms facing challenges servicing debt under baseline vs adverse scenarios.
  - Number and share of NFCs with ICR < 1, and those with 1 <= ICR < 2.
  - Number of firms facing liquidity shortfalls (negative cash ratio).

*Source: 1espea2024004-print-pdf - 22.      The key risks to financial stability in the current situation are geopolitical risks, a*

### 39.      The analysis reveals a moderate increase in NFC debt-at-risk

### The analysis reveals a moderate increase in NFC debt-at-risk

### Nonfinancial Corporates: Key findings on debt-at-risk and liquidity
- The share of firms with ICR less than one is estimated to rise by around 6pp under the adverse relative to the baseline scenario within a year and remain 5pp above the baseline estimate at end-2025.
- The share of debt of NFCs with ICR less than one increases by 0.9pp with respect to the baseline scenario, reaching around 29 percent of the total debt of these firms by end-2024.
- Econometric estimates indicate that the primary driver of the increase in NFCs’ debt distress in the adverse scenario is the fall in economic growth.
- The difference in NFC debt-at-risk across the two scenarios falls gradually during 2024–2026 as the shock abates.
- NFCs’ liquidity problems under the adverse scenario peak by 2025.
- Liquidity metric used: (Cash and Cash Equivalents) / (Short-Term Liabilities or Current Liabilities).
- Insolvency metric used: Share of NFCs with ICR less than unity.
- The stress test exercise is based on the model described in Tressel and Ding (2021).

### Banks stress tests: scope and approach
- The FSAP stress tests examine resilience of the banking system to solvency and liquidity risks through Top-Down (TD) tests based on supervisory and other confidential and market data.
- The solvency stress test measures effects of macroeconomic shocks on profitability and capitalization via satellite models and IMF staff methodologies.
- The TD liquidity tests assess capacity to withstand large withdrawals, estimating LCR and NSFR under alternative liquidity stress scenarios and a cash flow-based test over different stress horizons.
- The stress tests cover the 10 banking SIs. The three largest banks hold over 80 percent of banks assets, while the 10 SIs reach close to 95 percent of banking system asset.

### Solvency stress test methodology (high-level)
- Cut-off date for the stress test: September 30, 2023.
- Main supervisory data: FINREP and COREP.
- Credit risk:
  - Provisions are calculated as expected credit losses (ECL) for all asset classes/economic sectors with exposure at default (EaD), including off-balance sheet.
  - Key risk parameters: probability of default (PD), Loss Given Default (LGD), EaD, RWA broken down by exposure class and domicile for a total of 29 portfolios.
  - Starting points for risk parameters sourced from COREP C09.01 (SA) and C09.02 (IRB).
  - PD projections obtained via panel Bayesian Model Averaging (BMA) for domestic portfolios; foreign exposures use Moody’s one-year Expected Default Frequency (EDF) average estimates.
  - Transition matrices (TM) for scenario projections estimated through beta linking; domestic TMs provided by BdE based on national credit registry data; foreign TMs computed from vintages of supervisory data on exposure stage migration.
  - LGDs: collateralized lending calibrated through structural modelling using collateral values and house price paths; unsecured lending in advanced economies calibrated through Frye-Jacobs method; other exposures have constant LGDs.
  - Credit risk RWA updated according to portfolio regulatory treatment; SA densities at cut-off assumed constant; IRB exposures use Basel formulas and point-in-time to through-the-cycle PD conversion.
- Interest rate risk:
  - Structural model projects interest rates on new business; 50 accounts aggregated across interest-bearing assets and liabilities.
  - Bank-specific projections based on repricing and maturity profile derived from IRRBB and maturity ladder reporting (constant during stress horizon).
  - Interest on nonperforming exposures excluded; no relative change in asset or liability composition assumed.
  - Pass-through rates (betas) for domestic portfolios estimated via Panel Error Correction Models (PECM) using BdE bank-specific panel data.
  - Stylized pass-through used where data insufficient: 50 percent for deposits and 100 for loans.
  - Table of domestic betas (banking system averages):
    - Deposit sight Beta 0.06
    - Deposit term Beta 0.27
    - Mortgage loans Beta 0.78
    - Consumption loans Beta 0.6
    - Household sight deposits Beta 0.49
    - Household term deposits Beta 0.48
    - NFC loans (all) Beta 0.73
  - Table of foreign betas (selected):
    - BR: Loans Beta 1, Deposits Beta 0.67, NFC Beta 1
    - MX: Loans Beta 0.95, Deposits Beta 0.49, NFC Beta 0.49
    - TK: Loans Beta 1, Deposits Beta 0.50, NFC Beta 0.50
    - UK: Loans Beta 0.50, Deposits Beta 0.34, NFC Beta 0.59
    - US: Loans Beta 0.50, Deposits Beta 0.32, NFC Beta 0.59
    - EA: Loans Beta 0.71, Deposits Beta 0.28, NFC Beta 0.38
    - LatAm: Loans Beta 0.98, Deposits Beta 0.58, NFC Beta 0.74
- Market risk:
  - Valuation losses for debt securities assessed through a modified duration approach for FVPL and FVOCI fixed income securities due to interest rate and sovereign spread shocks.
  - FVOCI losses contribute to accumulated OCI; amortized cost portfolio not included in this module (sensitivity analysis performed separately).
  - Supervisory reporting on sovereign exposures (C33.00.a) used as main source for positions and duration across seven classes and seven maturity buckets.
  - Impact of hedging disregarded; derivatives portfolio stress limited by data availability.
- Non-interest income and expenses:
  - Net fee and commissions income (NFCI) projected under stress by adjusting latest income minus one standard deviation of historical variability for adverse scenario; baseline grows at balance sheet rate.
- Other assumptions:
  - Domestic balance-sheet growth follows credit growth rate; adverse scenario entails deleveraging with no feedback loop from banks’ performance to subsequent credit growth.
  - Foreign exposures semi-static: growth equal to scenario GDP growth when positive, null otherwise.
  - Other operational income and expenses constant in adverse scenario; grow at balance sheet rate in baseline.
  - Extraordinary items and minority interest set to zero.
  - Tax rate: bank-specific median effective tax rate across past five years for positive net income, zero otherwise; floor of 29 percent applied where historical tax rate significantly below peers; bank levy of 4.8 percent on NII and NFCI applied in both scenarios.
  - Dividend payouts: paid out of current year’s profit if positive, set at zero otherwise; dividend payout ratio = bank-specific median over past five years with a floor at 45 percent; no new share issuance or repurchases assumed.
  - Hurdle rates: CET1 hurdle rate around 8.4 (includes 4.5 percent Pillar 1 minimum, bank-specific Pillar 2, and 2.5 percent CCoB; CCyB kept at zero). Leverage ratio hurdle set at 3 percent.

### Solvency stress test results and key statistics
- Baseline scenario:
  - Aggregate CET1 capital ratio rises from a starting point of 12.6 percent in Q3-2023 to 15.4 percent at end of 2026.
  - Drivers: interest income from loan portfolios, relatively low cost of retail funding, stable loan loss provisioning.
  - Caveat: results could overestimate capitalization if current pace of share buy-backs and migration of sight to term deposits continue.
- Adverse scenario:
  - Aggregate CET1 ratio of the 10 SIs declines up to 3 percentage points, to 9.6 percent at end-2025, and recovers to just over 10 percent by end-2026.
  - SIs record losses in all three years, with a peak in the second year.
  - CET1 capital of two SIs fall below their Maximum Distributable Amount (MDA) trigger; one breaches the leverage ratio requirement.
  - Outcome reflects substantial deleveraging undertaken by banks in the scenario; deleveraging mitigates solvency deterioration but reduces banks’ capacity to meet credit demand.
  - Three-year cumulative credit impairments are 135 percent of starting CET1 capital by end-2026 in the adverse scenario.
  - Under the baseline scenario, cumulative three-year impairments are 54 percent of starting capital.
  - Majority of credit impairments are recorded in foreign portfolios; global synchronized shocks reduce geographical diversification benefits.

*Source: IMF staff calculations and supervisory data as presented in the chapter.*

### 64.      The increase in the interest rates in the adverse scenario allows banks to initially

### 1espea2024004-print-pdf - 64.      The increase in the interest rates in the adverse scenario allows banks to initially

### Interest rate pass-through, net interest income (NII), and market risk
- NII dynamics in the adverse scenario:
  - NII grows in the first year due to the interest rate pass-through differential between assets and liabilities, especially on domestic exposures, as price effects dominate volume effects.
  - The attenuative effect of NII decreases during the last two years as volume effects start to dominate because of protracted negative credit growth in Spain and increasing nonperforming exposures in both domestic and foreign loan books.
- Market risk:
  - Market risk losses are material only in the first year of the adverse scenario and contribute positively to earnings in the outer years as rates start to normalize.
  - Market risk contributes negatively to profitability and capital in the first year but is not the main driver of solvency analysis results.
  - Explanations for limited impact:
    - Spanish banks, on aggregate, hold 60 percent of their bond portfolios at amortized cost, with a peak at 96 percent, so changes in market valuation do not get reflected in banks’ earnings.
    - Negative impact in the first year is more than compensated by gains in subsequent years when interest rates fall.
  - Stressed net fee and commission income is lower than in baseline but remains positive.

### Sensitivity analyses: banks’ pass-through rates
- Alternative pass-through calibrations considered:
  - Doubling the pass-through on loans and deposits separately (“2x-Deposit” and “2x-Assets”).
  - BdE estimates of pass-through on loans and deposits as of June 2023 (“BdE-Assets”, “BdE-deposits”).
  - Long-run estimates of pass-through using an ARDL model.
  - Stylized calibration with 50 percent pass-through on sight deposits and 100 percent on term deposits (“50-100”).
  - All calibrations assume a cap of 100 percent for the pass-through and pertain to the adverse scenario.
- Key results and comparisons:
  - “2x-Assets” and “50-100” show extreme outcomes:
    - Doubling pass-through on assets (implying 100 percent rate on the domestic loan portfolio) boosts NII significantly.
    - Increasing cost of deposit funding to unprecedented levels brings about three more capital breaches and further erodes capital buffers of banks already in breach in the adverse scenario.
  - Intermediate calibrations:
    - Lowering pass-through to assets or increasing them for deposits causes one CET1 breach for a bank already close to the threshold under the adverse scenario.
    - A milder pass-through on deposits does not materially affect system-level capital ratios.
  - At the banking system level, capital ratios do not vary materially in all intermediate cases.

### Unrealized losses on amortized cost securities
- Portfolio composition and encumbrance:
  - Securities valued at amortized cost represent close to 60 percent of the bond portfolio.
  - Of these: 60 percent issued by the Spanish government, 20 percent by other euro area sovereign (rest of EA), and 20 percent by a mix of advanced and emerging economies (RoW).
  - Level of encumbrance ranges between 29 and 82 percent, with higher encumbrance for shorter maturities and domestic bonds on average.
  - Loans portfolio is largely at variable rate: 63 percent for household loans and 57 percent for NFC loans.
  - Majority of amortized cost securities are classified as HQLA and held for liquidity purposes.
- Data limitations and methodology notes:
  - Prudential data (C33.00.a) include loans to local and central government and do not distinguish bonds and loans with sufficient granularity; additional BdE aggregate security-level data were used.
  - FSAP estimates loans exposures to represent between 10 to 67 percent of the carrying amount for a given maturity bucket.
  - Given data limitations and lack of information on hedges, loss estimates are to be considered upper bounds.
- Estimated unrealized losses:
  - Unrealized losses amounted to EUR 26 billion as of June 2023.
  - Unrealized losses would reach EUR 78 billion in the adverse scenario (triple the initial amount).
  - Median unrealized loss at individual bank level for exposures to Spanish sovereign: EUR 2 billion initially, reaching EUR 4 billion in the adverse scenario.
  - Individual banks’ unrealized losses are heterogeneous because of size, geographical and maturity profile of banks’ portfolios.

### Liquidity stress tests — overview and methodology
- Tests performed:
  - LCR test (one-month horizon) and NSFR (one-year horizon).
  - Granular cash-flow based (CF) analysis for one-week and three-month horizons, including a reverse-stress test based on a “Distance to Liquidity Stress Indicator (DLSI)”.
  - All bank-level tests use August 2023 regulatory data from COREP and FINREP.
- Scenario grid and calibration:
  - LCR analysis based on four scenarios; CF analysis uses a grid with 25 levels of scenario severities.
  - Severity ranges from European transposition of the Basel III scenario to an “aggressive” adverse stress scenario simultaneously stressing market value of liquid assets, inflows, and outflows.
  - Two intermediate scenarios consider only “market stress” and “funding stress”.
  - Sensitivity analysis includes increasing runoff rates on outflows for LCR and reverse stress tests for each bank.
  - Adverse scenario calibration uses historical episodes (Banco Popular June 2017, Caixabank 2017), March-2023 banking turmoil, and concurrent FSAPs.

### Funding structure, counterbalancing capacity, and encumbrance
- Funding mix and stability:
  - Spanish SIs rely primarily on deposits for funding: 77 percent of bank liabilities in June 2023.
  - Stable retail funding accounts for 48 percent of the Spanish banks’ deposit mix subject to low run-off rates.
  - Remaining 52 percent is a combination of non-stable retail, operational and non-operational wholesale deposits.
- Counter-balancing capacity (CBC):
  - CBC across the 10 SIs has been stable since monetary policy normalization started.
  - Majority of CBC comprises central bank reserves (31 percent) and level 1 assets (32 percent).
  - Spanish banks repaid over 90 percent of their TLTRO in 2022-23, largely by reducing central bank reserves and partially by issuing secured and unsecured debt.
  - Asset encumbrance: 17 percent in 2023-Q2 compared to 23 percent at end of 2021.

### LCR and NSFR results
- LCR scenario setup:
  - Adverse scenarios capture market-driven stress and funding pressures via higher haircuts on HQLA and higher run-off rates on outflows.
  - A set of intermediate “Outflow_25”, “Outflow_50” and “Outflow_75” scenarios are weighted sums of Basel III and Outflow scenarios.
- LCR outcomes:
  - Aggregate LCR for the 10 SIs falls below 100 percent under the “outflow” and “aggressive” scenarios.
  - Four out of 10 SIs have LCR above the regulatory hurdle rate of 100 percent for the “outflow” scenario.
  - Three SIs have LCR above 100 percent for the “aggressive” scenario.
  - Scenario bank counts table (number of banks falling below regulatory hurdle):
    - Basel III: 0
    - Market: 0
    - 25% Outflow: 0
    - 50% Outflow: 1
    - 75% Outflow: 2
    - 100% Outflow: 6
    - Aggressive: 7
  - If 75 percent of the run-off rates considered under the “outflow” scenario were to materialize, SIs in aggregate would still exceed the hurdle rate.
- NSFR outcomes:
  - Aggregate NSFR for the 10 significant institutions has been well above the hurdle rate for the last two years.
  - The most recent NSFR for June 2023 was 129 percent.

### Cash-flow (CF) and reverse stress test analysis
- CF test design and indicators:
  - Cash flow stress test uses two key indicators: bank’s cumulated net funding gap and counterbalancing capacity.
  - Tests use COREP C66 contractual cash flows by maturity buckets for one-week and three-month horizons.
  - Net-funding gap = sum of (inflows − outflows) across buckets within a horizon.
  - Counterbalancing capacity = cumulated value of liquid assets that banks can monetize under stress at reasonable prices across time buckets.
  - A liquidity shortfall arises when counterbalancing capacity is exhausted to fulfill the net-funding gap.
- Purpose:
  - CF analysis provides a granular balance-sheet approach to assess liquidity position vis-à-vis LCR and helps identify tipping points for policy discussions.

*Source: IMF staff calculations and supervisory sources as presented in the provided content.*

### 81.      The CF analysis considers a range of 25 scenarios affecting asset values, inflows, and

### 1espea2024004-print-pdf - 81.

### Cash Flow (CF) analysis and stress scenarios
- The CF analysis considers a range of 25 scenarios affecting asset values, inflows, and outflows uniformly across different maturity buckets.  
- Scenarios are based on a linear grid of scenario severities across all factors (haircuts, run-off rates and roll-off rates) spanning the “Basel III” and “aggressive” scenario analogues of the LCR applied to bank level liquidity flows over one-week and three-month horizons.  
- Analysis is based on the August-2023 data vintage and done for the entire set of significant currencies (only EUR exposure results are presented). Results are robust to different choice of data vintages.  
- Note: the CF analysis note also refers to "more than 20" stress scenarios affecting runoff rates and rollover rates, with scenarios scaled using a linear grid between Basel LCR assumptions and the “aggressive” scenario.

Findings
- On aggregate, SIs would maintain a marginal liquidity surplus under the aggressive scenario over a one-week horizon while having a small shortfall over the three-month period.  
- Heterogeneity: three banks experience a marginal liquidity shortfall in the one-week cash flow at the “aggressive” scenario severity; this increases to four banks for the three-month horizon.  
- Considering only EUR exposures, banks appear more resilient; the largest banks maintain higher liquidity buffers over both one-week and three-month horizons (EUR-only results presented for three international SIs).  
- Key drivers of CBC depletion are wholesale deposit outflows and credit lines (waterfall-decomposition of the three-month CF analysis).

### Distance to Liquidity Stress Indicator (DLSI / d)
- Definition: d measures the gap in the ratio of “cumulative outflows to total assets” between the “aggressive scenario” and the “first shortfall scenario” (the first scenario severity for which the bank liquidity surplus turns to shortfall) for each of the 10 SIs.  
- Interpretation: d measures the amount of additional stress (in percent of cumulative outflows to total assets) banks can withstand beyond the aggressive scenario.  
- Empirical results:
  - Only two banks have a negative DLSI for the one-week horizon.  
  - One bank has a negative DLSI for the three-month horizon.  
- Explanation: negative DLSI occurrences are partly explained by inclusion of contingent outflows due to committed credit lines in the aggregation of outflows for the CF exercise.

### Liquidity–Solvency interaction and forced liquidation scenario
- If banks breach LCR and experience further funding outflows, they may be unable to access the ECB standing liquidity facility; if BdE emergency liquidity assistance (ELA) is unavailable, banks may need to sell securities (e.g., government bonds held at amortized cost) in the open market.
- Holdings considered for sale-loss estimates:
  - Unencumbered securities issued by EA sovereigns (including Spain) amount to EUR 105 billion in total, or 39 percent of the whole bond portfolio for the subset of banks in scope.  
  - Securities issued outside the EA were excluded.  
  - Any provision related to sovereign exposures is deducted when estimating losses (effect negligible).  
  - Data limitations imply estimates should be considered upper bounds.

Estimated impact on capital
- Potential sale losses from liquidation of securities in the open market induce a median decline in CET1 ratio between 1.3 and 3.5 percentage points (pp).  
  - Lower bound: considers only unrealized losses up to June 2023 (the “initial” scenario).  
  - Upper bound: considers peak losses from the adverse scenario (occur at the end of the first year).  
  - In the adverse scenario, the first quartile corresponds to a decline of 4.1 pp, the third quartile corresponds to a decline of 1.7 pp in CET1 ratio.
- Partial liquidation outcomes:
  - If banks liquidate half of their portfolio (pecking order sells shorter maturity bonds first), losses entail a median CET1 ratio decline of 0.2 pp in the initial scenario and 0.5 pp in the adverse scenario.
- Factors that make forced bond-sale an extreme event:
  - A breach of LCR does not automatically preclude Eurosystem standing facility access; a grace period may be granted.  
  - EA sovereign bonds can be used as collateral in cleared repo, albeit at higher haircuts; ample HQLA inventory implies banks could still raise significant liquidity.  
  - The portion of the portfolio already marked-to-market would likely be liquidated first; for the sample considered this would generate EUR 38 billion in total under adverse scenario haircuts (accounting only for the unencumbered part).  
  - Banks can generate liquidity via issuing covered bonds backed by mortgage loan portfolios—particularly viable for banks with regular market issuance.  
  - Assuming simultaneous denial of standing facility and ELA access is likely unreasonable given the degree of economic adversity.

### Cross-border interconnectedness and contagion analysis
Model and assumptions
- Network model: Espinosa-Vega and Solé (2011) applied to matrix of bilateral interbank gross credit exposures.  
- Sample: 11 banking centers (predominantly Europe) with largest direct banking exposures with Spain. BIS consolidated banking statistics up to 2023-Q2 used.  
- Shock parameters and recovery assumptions:
  - 50 percent loss on unsecured asset claims following counterparty default.  
  - 30 percent loss for portion of secured lending with collateral.  
  - Due to data scarcity, half of bilateral exposures assumed collateralized.  
  - 35 percent reduction in interbank funding (reflecting a 65 percent roll-over rate).  
  - 30 percent discount on assets forced into liquidation due to funding shortfalls.  
  - Collateralized portion assumed to be 50 percent, of which 70 percent is recoverable in default.  
- Scenarios explored: (i) credit shock alone; (ii) both credit and funding shocks.

Findings
- Spanish banks’ vulnerability to contagion from distress in foreign banking systems via direct exposures is modest because most cross-border claims are to non-banking sectors.  
- Vulnerability index: Spanish banks could face, on average across all in-sample counterpart countries, a loss of approximately 1.2 percent of their current total capital if banks were forced to forfeit their cross-border banking claims under these recovery assumptions.  
- Losses due to banking distress in some individual European countries could reach up to 4 percent of bank capital in Spain.  
- Even in such cases, the post-shock capital adequacy ratio remains above 14 percent.

### Domestic interbank linkages and contagion
- Network: SIs, 6 LSIs and two Spanish subsidiaries of foreign banks; model of Espinosa-Vega and Solé (2011).  
- Data: domestic interbank exposure network at the group level from BdE’s credit registry; interbank exposures include Temporary Asset Acquisition, Fixed-Income Securities, Financial Credit, Securities Loans, Non-Recourse Factoring with Investment, Leasing Operations, Commercial Credit, and Overdue and Uncollected Products from Doubtful Assets.  
- Collateralization assumptions: alternative assumptions of 30 percent and 50 percent used due to lack of precise data.

Findings
- Domestic interbank contagion risk is concentrated within the sub-system of SIs: losses within this sub-system account for over 72 percent of total losses.  
- Foreign subsidiaries are more vulnerable through exposures to LSIs compared to SIs.  
- The credit channel is more impactful than the funding channel: the credit channel accounts for 3/4 of total losses, with the remainder due to funding shock losses.

Table summary (contagion losses by trigger / recipient segments)
- Recipient of the shock (in percent of total losses):
  - Source SIs → SIs: 72.3; SIs → LSIs: 4.6; SIs → Foreign Subs.: 3.5  
  - Source LSIs → SIs: 5.4; LSIs → LSIs: 0.7; LSIs → Foreign Subs.: 8  
  - Source Foreign Subs. → SIs: 1.3; Foreign Subs. → LSIs: 4.2; Foreign Subs. → Foreign Subs.: 0

### Financial sector sovereign exposures
Key statistics and trends
- Government debt securities comprised approximately 80 percent of Spanish banks' debt securities holdings.  
- As of 2023-H1, Spanish banks' debt securities holdings were valued at €590 billion, making up around 14.8 percent of their total assets, a sharp annual increase of 8.6 percent.  
- Since June 2019, cumulative growth in debt holdings has exceeded 15 percent, outpacing growth in total assets by 2.3 percentage points during this period.  
- Share of local government debt holdings remains stable at around 30 percent of total exposure to the public sector.  
- Among advanced European countries, Spanish banks are among the most exposed to sovereign risk in Europe relative to both assets and capital; only Italian banks are more exposed among larger EA economies.  
- Asset allocation shifts: banks and open-ended investment funds have increased their allocation to sovereign debt in recent years, while pension funds and insurance companies have decreased theirs. Consequently, the share of banks and funds in domestic financial sector sovereign exposures has risen and that of pension and insurance has fallen.  
- Holdings patterns: banks and insurance companies show a clear home bias in sovereign debt holdings compared with investment funds. Investment funds have larger exposure to medium-term bonds; insurance companies tend to hold longer duration assets through long-term bonds.

*Source: IMF Staff.*

### 96.      While the public sector is the primary

### 1espea2024004-print-pdf - 96.      While the public sector is the primary

### Exposure to sovereign debt
- As of June 2023 (excluding BdE’s holdings), Spanish resident banks hold the largest share, of 14 percent, of outstanding Spanish sovereign securities.
- Domestic NBFI sector holds 11 percent of outstanding Spanish sovereign securities.
- Foreign entities account for 42 percent of the outstanding volume of Spanish sovereign debt.
- Figure labels (Jun 2023): 42%, 1%, 3%, 7%, 14%, 33% (as presented in the source).

### Distribution and maturity preferences across financial sectors
- Pension funds and insurance firms have reduced their exposure to sovereign debt but remain the most exposed in their portfolios.
- The share of banks and open-ended funds in total sovereign debt holding has increased in recent years.
- Home biases:
  - Banks and insurances have home biases in sovereign debt holdings.
  - Funds prefer medium-term maturity bonds.
  - Insurance firms hold long-term bonds.

### Sovereign-bank distress linkages (mechanism)
- Sovereign-bank distress channel arises from direct bank exposure to government debt used as collateral.
- A rise in sovereign distress can reduce market value of government debt and collateral, tightening bank capital constraints and lending standards.
- The relatively high exposure of Spanish banks to the public sector could reinforce the link between banking stability and public finances during downturns or large exogenous shocks.

### FSAP identification and methodology summary
- Sovereign distress defined using thresholds of:
  - (1) the one-year sovereign CDS for Spain; and
  - (2) the ECB’s composite indicator of sovereign stress (CISS).
- These thresholds primarily encompass the years 2009-2013 (European sovereign debt crisis).
- Sovereign exposure measured as the share of government debt securities in total assets in the previous year.
- Analysis includes bank and time fixed effects; standard errors clustered at the bank level.

### FSAP findings (quantitative effects)
- Banks with greater exposure to sovereign debt are associated with:
  - Greater reduction in lending following sovereign distress.
  - Higher likelihood of reduced net interest income (NII) and profitability (ROE).
- Quantified effect on ROE:
  - Banks with initial government-debt-holdings-to-total-assets 10 percentage points higher experienced a decline of 0.7 percentage points in their ROE following an increase in sovereign distress, defined as the CISS exceeding the third quartile of its historical distribution (CISS ≥ 0.48).
- Effect is more pronounced for banks whose capitalization is below the median of the sampled banks (“less capitalized banks”).

### Stress scenarios — selected adverse macro scenario values for Spain (Percent change unless otherwise indicated)
- 2024:
  - Real GDP growth: -4.09
  - Core Inflation: 6.77
  - Unemployment: 14.03
  - Short term sovereign yield: 5.89
  - Long term sovereign yield: 7.73
  - Sovereign spread over Germany: 2.06
  - House Price: -17.16
  - Credit Growth: -2.06
- 2025:
  - Real GDP growth: -3.10
  - Core Inflation: 2.46
  - Unemployment: 18.61
  - Short term sovereign yield: 4.80
  - Long term sovereign yield: 6.35
  - Sovereign spread over Germany: 1.71
  - House Price: -3.29
  - Credit Growth: -0.22
- 2026:
  - Real GDP growth: 1.00
  - Core Inflation: 0.86
  - Unemployment: 19.84
  - Short term sovereign yield: 3.71
  - Long term sovereign yield: 3.85
  - Sovereign spread over Germany: 1.20
  - House Price: 9.70
  - Credit Growth: 1.32

### Stress Test Matrix (STeM) — key parameters and scope
- Institutional perimeter:
  - Ten SI banks included, of which one G-SIB.
  - Market share: Almost 95 percent of the banking sector assets.
  - Data vintages: 2023 Q3 (starting point for P&L (annualized), balance sheet and capital); time series 2015 Q1–2023 Q3 (net fee and commission income).
- Channels and models:
  - FSAP team satellite models and methodologies; balance-sheet regulatory approach.
  - Provisioning for IRB and SA modeled using IFRS9 transition matrix approach.
  - Traded risk: revaluation of FVPL and FVOCI securities assessed using a modified duration approach.
  - Structural model of bank NII based on repricing ladder and estimated betas.
  - Satellite models for credit losses (PD and LGD by portfolio), funding costs, lending rates, net fee and commission income and risk weights.
- Tail shocks and horizon:
  - Stress test horizon: 2023 Q3–2026 Q4 (three years, one quarter).
  - Two scenarios: baseline (October 2023 WEO) and adverse (captures key RAM risks; uses GFM structural macro-econometric model).
- Risks covered:
  - Credit (loans and debt securities), market (repricing and credit spread risk), and interest rate risk on the banking book (IRRBB).
- Behavioral adjustments and assumptions:
  - Balance sheet growth aligned with nominal GDP paths, floored at 0, except domestic loans (credit growth negative for 2024 and 2025).
  - For IRB portfolios, through-the-cycle PDs, downturn LGDs and EAD for each asset class/industry used to project RWAs.
  - Interest income from nonperforming loans is not accrued.
  - Dividends are paid by banks that remain adequately capitalized throughout the stress.
- Regulatory parameters:
  - Basel III regulatory minima on CET1 (4.5 percent) and macroprudential buffers for O-SII included.
  - Leverage ratio assessed against 3 percent Basel III minimum.
- Reporting outputs:
  - System-wide capital shortfall.
  - Number of banks and percentage of banking assets below regulatory minima.
  - Impact drivers, including profit components.

### Selected Risks from the Risk Assessment Matrix (RAM)
- Intensification of regional conflict(s) and geo-economic fragmentation
  - Overall Level of Concern: High
  - Relative Likelihood: Medium
  - Expected impacts include trade disruptions, shortages, higher inflation, and migration pressures.
- Abrupt global slowdown or recession leading to systemic financial instability
  - Overall Level of Concern: Medium
  - Relative Likelihood: High
  - Expected impacts include tighter financial conditions, cross-border spillovers via Spanish banks, bond repricing, housing and CRE corrections, erosion of banks’ capital buffers, and amplified adverse effects if social/political tensions rise.
- Commodity price volatility
  - Overall Level of Concern: High
  - Relative Likelihood: Medium
  - Expected impacts include higher energy-driven inflation and reduced export competitiveness.
- Monetary policy miscalibration
  - Overall Level of Concern: Medium
  - Relative Likelihood: High
  - Expected impacts include de-anchored inflation expectations and higher inflation risk premia.
- Weak implementation of fiscal commitments, delays in EU funded projects or reassessment of sovereign risk
  - Overall Level of Concern: Medium
  - Relative Likelihood: High
  - Expected impacts include higher sovereign risk premia, pressure on public debt dynamics, reduced value of fixed-income assets, and pressure on financial sector balance sheets.
- Extreme climate events
  - Overall Level of Concern: Medium
  - Relative Likelihood: Medium
  - Expected impacts include loss of lives, infrastructure damage, disrupted operations, impaired borrowers’ ability to repay, and asset value declines.
- Cyberthreats
  - Overall Level of Concern: Medium
  - Relative Likelihood: High
  - Expected impacts include financial and economic instability from cyber-attacks.

*Sources: Bank of Spain, CNMV, ECB SHSS, IMF staff calculations.*

### 8. Channels of

### 8. Channels of Risk Propagation

### Liquidity Stress Test — Overview and Design
- Risks considered: Funding liquidity; Market liquidity.
- Buffers: The counterbalancing capacity, including liquidity obtained from markets and/or the central bank’s facilities. Expected cash inflows are also included in the cash-flow based and LCR-based analysis.
- Top-down by IMF: Banking Sector: Liquidity Stress Test (Table 1: Stress Test Matrix (STeM) concluded).

### Tail Shocks — Size and Calibration
- Run-off calibration:
  - Run-off rates are calibrated to reflect scenarios of system-wide deposit runs and dry-up of unsecured wholesale and retail funding, with additional run-off for non-resident deposits on top of the retail and wholesale run-off.
  - Calibration informed by historical events, recent international experience in liquidity crisis and IMF expert judgment.
- HQLA haircuts:
  - Calibrated against ECB haircuts, past EA FSAPs, and market shock for investment securities and money market instruments used in the solvency stress test.

### Regulatory and Market-Based Standards
- Regulatory Standards:
  - Consistent with Basel III regulatory framework (LCR).
  - Liquidity shortfall assessed by bank.

### Reporting Format for Results
- Output Presentation:
  - Liquidity ratio or shortfall by groups of banks and systemwide.
  - Number of banks that still can meet or fail their obligations.
  - Distribution of the distance to liquidity stress indicator for banks and systemwide.

### Interconnectedness and Contagion Analysis
- Institutional Perimeter: Banks and NBFIs, data permitting.
- Methodology: Network-based contagion analysis based on three potentially mutually reinforcing contagion channels:
  - Price channel: selling of assets by institutions in distress affects other institutions’ balance sheet through the price channel. The change of asset price will be a function of market depth, which depends on authorities’ data provision.
  - Credit contagion: default of institutions can cause other institutions to default in turn.
  - Funding contagion: refusal of institutions to rollover funding can cause other institutions to sell assets or stop rolling over funding in turn.
- Note: This analysis is not part of the stress testing work but is included in the systemic risk analysis.

*Source: Spain — International Monetary Fund (excerpts from Chapter 8 and related appendices).*

---

### Appendix IV. Credit Risk — Probability of Default (PD) and Estimation

### Domestic PDs
- Domestic PD paths were estimated via panel BMA, obtaining bank-specific paths.
- Table 1. Domestic PDs: BMA Output (variables and reported coefficients/PIPs shown in source table).  
  - Variables listed include: Real GDP_- YoY; CPI – YoY; UR; UR - YoY; FX HPI - YoY; Short term rate; Term Spr.; Sov. Spr.; 10Y NFC LRM; PIP; MORTGAGE LRM; PIP; CONSUMPTION LRM; PIP.
- Notes: 1. Long-Run Multiplier, normalized by standard deviation; 2 Posterior Inclusion Probability; 3. Unemployment Rate; 4. House Price Index; 5 Spread. Results refer to BMA with sign constraints.

### Foreign PDs
- Foreign PD paths were estimated on aggregate level, then adapted to bank-specific PDs by anchoring to bank-specific starting point in distance-to-default space using:
  - PD_{i,t}^k = Φ(Φ^{-1}(PD_{i,0}^k) + Φ^{-1}(PD_t^k)) − Φ^{-1}(PD_0^k) where Φ standard normal is the cumulative distribution function.
- Table 2. Foreign PDs: BMA Output (country-specific LRM and PIP entries for BR, MX, TR, UK, US presented in source table).
- Notes: 1. Long-Run Multiplier, normalized by standard deviation; 2 Posterior Inclusion Probability; 3. Unemployment Rate. Results refer to BMA with sign constraints.

---

### Appendix IV. Credit Risk — Loss Given Default (LGD)

### Domestic Mortgages
- LGD computed with structural model based on Loan-to-value updated with the scenario:
  - LGD_t = (1 − TM_{32}) ⋅ max(LTV_t − SR_{LTV_t}, 0)
  - LTV_t = LTV_0 ⋅ HP_0 / HP_t
- Where TM_{32} indicates the transition probability from Stage 2 to Stage 3, LTV is the loan-to-value, HP is the house price and SR is sale revenue.

### Other Portfolios (Advanced Economies)
- LGDs linked to PDs through a Vašíček equation:
  - LGD_{b,t0+h} = Φ(Φ^{-1}(PD_{b,t0+h}) − k) PD_{b,t0+h}
  - k = [Φ^{-1}(PD_{t0}) Φ^{-1} − (PD_{t0} ⋅ LGD_{t0} ⋅ f)] / √(1 − ρ)
  - Factor f set (implied) at outset so equations imply an LGD matching observed bank-portfolio specific LGDs.
- For exposures in emerging market LGD are kept constant.

---

### Appendix IV. Credit Risk — IFRS9 and Stock Projections

### Stocks and Flows
- Performing and nonperforming exposure stocks projected based on transition matrix-implied flows, allowing for write-offs and asset sales.
- Stage flows equations:
  - S2_t = S2_{t−1} + TM_{t,12} S1_{t−1} + TM_{t,32} S3_{t−1} − TM_{t,21} S2_{t−1} − TM_{t,23} S2_{t−1} − M_{t,2} S2_{t−1}
  - S3_t = S3_{t−1} + TM_{t,13} S1_{t−1} + TM_{t,23} S2_{t−1} − TM_{t,31} S3_{t−1} − TM_{t,32} S3_{t−1} − WRO_t S3_{t−1}
- Total stock:
  - S_t = S1_t + S2_t + S3_t = (1 + g_t) S_{t−1}
  - S1_t = max(0, S_t − S2_t − S3_t)
- Write-off and asset sale rates (WRO_t) assumed constant at end-sample observed values at observed bank-portfolio specific levels.
- Reference: Details in Frye and Jacobs (2012).

### Transition Matrix Updating (four steps)
a. TM_{t,13} and TM_{t,23} derived from PD_t path:
  - TM_{t,23} = Φ(Φ^{-1}(TM_{0,23}) + (Φ^{-1}(PD_t) − Φ^{-1}(PD_0)))
  - TM_{t,13} = min(1, max(0, PD_t ⋅ (S_{t−1}^1 + S_{t−1}^2) − TM_{t,23} ⋅ S_{t−1}^2) / S_{t−1}^1)
b. Project other TM_{t,ij}, i≠j based on sensitivity to shock in distance-to-default space:
  - TM_{t,ij} = Φ(Φ^{-1}(TM_{0,ij}) + DD_{ij}(Φ^{-1}(TM_{t,i3}) − Φ^{-1}(TM_{0,i3}))), i≠3
  - TM_{t,3j} = Φ(Φ^{-1}(TM_{0,3j}) + DD_{3j}(Φ^{-1}(TM_{t,23}) − Φ^{-1}(TM_{0,23})))
c. Derive diagonal terms TM_{t,ii} so rows sum to 1.

### Expected Credit Losses (ECL) by Stage
- PROV_t^{S1} = TM_{t,13} ⋅ LGD_t ⋅ S1_{t−1}
- PROV_t^{S2} = ∑_{s=t+1}^{L_s} TM_{s,23,*} ⋅ LGD_s ⋅ S2_{s−1} / (1 + r)^s
- PROV_t^{S3} = LGD_t ⋅ S3_{t−1}
- PROV_t^k = PROV_t^{S1} + PROV_t^{S2} + PROV_t^{S3}

---

### Appendix IV. Risk-Weighted Assets (RWA)

### STA — Performing
- Densities at the cut-off point assumed constant over the scenario horizon.
- RWA densities per portfolio/segment ‘e’:
  - ρ_e^{PE}[0] = RWA(STA)_e^{PE}[0] / (EAD(STA)_e^{PE}[0] − PROV(STA)_e^{PE}[0])
- RWA over time:
  - RWA(STA)_e^{PE}[t] = ρ_e^{PE}[0] * (EAD(STA)_e^{PE}[t] − PROV(STA)_e^{PE}[t])

### STA — Nonperforming
- Densities at the cut-off point assumed constant over the scenario horizon.
- RWA densities per portfolio/segment ‘e’:
  - ρ_e^{NPE}[0] = RWA(STA)_e^{NPE}[0] / (EAD(STA)_e^{NPE}[0] − PROV(STA)_e^{NPE}[0])
- RWA over time:
  - RWA(STA)_e^{NPE}[t] = ρ_e^{NPE}[0] * (EAD(STA)_e^{NPE}[t] − PROV(STA)_e^{NPE}[t])

### IRB — Performing and Nonperforming
- Performing (Use of Basel formulas). PD TTC and LGD DT are projected as follows:
  a. PD TTC:
    i. Option A: If historical PD PiT available for a period equal to TTC Window − 1, then PD TTC is calculated based on moving average of PD PiT with an adjustment at cut-off point:
      - PD_TTC[t] = (∑_{k=0}^{TTC Window} PD_{PiT}[t−k]) / (TTC Window) + (PD_{TTC}[T0] − ∑_{k=0}^{TTC Window} PD_{PiT}[T0−k] / (TTC Window))
    ii. Option B: If no historical PD PiT series but PD PiT and PD TTC known at cut-off, a moving-average-like formula used:
      - PD_TTC[t] = ((TTC Window − (t+1)) * PD_{TTC}[T0] * TTC Window − PD_{PiT}[T0]) / (TTC Window −1) + ∑_{k=0}^{t} PD_{PiT}[k] / (TTC Window)
    - Main caveat: implicit assumption that TTC is a simple cycle average which ignores conservatism when banks do not update downwards PD TTCs during boom cycles.
  b. LGD DT:
    - LGD_T^e[t] = max(LGD_{PiT}^e[t], LGD_{DT}^e[T0])

- IRB — Nonperforming:
  - RWA(IRB)_e^{NPE}[t] = EAD(IRB)_e^{NPE}[t] * (LGD_T^e[t] − ELBE^e[T0]) * 12.5
  - RWA(IRB)_e^{NPE}[t] = EAD(IRB)_e^{NPE}[t] * (ELBE^e[T0] − LGD_T^e[t]) * 12.5

---

### Appendix V. Interest Income — Notation, Assumptions, and Calculation

### Notation
- E_t: total loans outstanding at time t.
- E_t[T,T+1]: loans in bucket [T,T+1] (loans at end of period t with between T and T+1 years left until repricing).
- θ_t[T,T+1] = E_t[T,T+1] / E_t (fraction of loans in each bucket at time t).

### Assumptions
i. Total stock of loans grows at exogenous rate g_t.
ii. E_{t+1} = (1 + g_{t+1}) E_t.
iii. r_t^L is the interest on new loans. Loans pay a fixed nominal rate, principal not inflation-adjusted.
iv. No refinancing of loans assumed.
v. θ_t[T,T+1] kept constant at cut-off date.
vi. r_0[k−1,k] = r_0 ∀ k.

### Interest Income Calculation
- Total interest income in period t:
  - IĨ_t = ∑_{k=3}^{∞} r_{t−1}[k,k+1] E_{t−1}[k,k+1] + (1/2) r_{t−1}[0,1] E_{t−1}[0,1] + (1/2) r_t^{E,nb} (∑_{k=0}^{5} I_t[k,k+1])
  - The three terms account for:
    i. Interest income from exposures that at end of year-(t−1) had at least 1 year left until repricing.
    ii. Income from exposures that at end of year-(t−1) had less than 1 year left until repricing.
    iii. Newly issued/repriced loans.
- Additional equations:
  - I_t[k−1,k] = E_{t−1}[k,k+1] (1 + r_{t−1}[k,k+1]) − E_t[k−1,k]
  - r_t[k−1,k] = [E_{t−1}[k,k+1] (1 + r_{t−1}[k,k+1]) / E_t[k−1,k]] ⋅ r_{t−1}[k,k+1] + [I_t[k−1,k] / E_t[k−1,k]] ⋅ r_t[E,nb]
  - E_t = (1 + g_t^E) E_{t−1}
  - E_t[k,k+1] = θ_0[k,k+1] ⋅ E_t
- Adjustment for nonperforming exposures:
  - II_t = av{(1 − NPEr_t) E_t ; (1 − NPEr_{t−1}) E_{t−1}} / av{E_t; E_{t−1}} ⋅ IĨ_t

### Alternative Calibration Table (deposit and loans beta)
- Table 1. Spain: Alternative calibration considered for deposit and loans beta — sector-item calibrations with numeric entries (source table includes values: e.g., PECM, 2x Assets, 2x Deposits, BdE - Assets, BdE - Deposits, Long run (ARDL), and values for Sight, Term, Mortgages, Consumption, NFC deposits and loans).  
- Note: Shaded cells indicate a deviation from central calibration (PECM).

---

### Appendix VI. Market Risk — Debt Securities Valuation and Shock Mechanics
- Fair value changes from risk-free rates (r) and sovereign spread (s).
- Loss: modified duration of position/portfolio multiplied by changes in benchmark risk free rate/credit spread and portfolio valuation.
- Securities partitioned by duration/maturity; relevant interest rate considered for shock.
- Formulae:
  - MD_t ~ MD_{t−1} / (1 + Δs_t + Δr_t)
  - FV_t = FV_{t−1} (1 − MD_t ⋅ Δs_t − MD_t ⋅ Δr_t)

---

### Appendix VII. Liquidity Stress Test — Scenario Specification and Weights (percent)

### Scenario Weights (percent) — Outflows (selected entries preserved exactly)
- Retail Deposits:
  - Stable Demand Deposits: 5 5 10 10
  - Less Stable Demand Deposits: 10 10 26.3 26.3
  - Term Deposits, residual maturity > 30 days: 0 0 19.2 19.2
- Unsecured Wholesale Funding:
  - Demand and Term deposit, small business, residual maturity <30 days:
    - Stable Deposits: 5 5 10 10
    - Less Stable Deposits: 10 10 33 33
  - Operational deposits due by clearing, custody and cash mgt activities: 25 25 50 50
    - Portion covered by deposit insurance: 5 5 10 10
  - Cooperative banks in an institutional network: 25 25 35 35
  - NFCs, sovereigns, central banks, multilateral development banks, PSEs:
    - Fully covered by deposit insurance: 20 20 35 35
    - Not fully covered by deposit insurance: 40 40 60 60
  - Other legal entity customers: 100 100 100 100
- Secured Funding:
  - Secured funding with a central bank, or backed by Level 1 assets: 0 0 5 5
  - Secured funding backed by Level 2A assets: 15 15 30 30
  - Secured funding backed by non-Level1 or non-Level2A asset, with dom. sovereign, multilateral dev banks, or domestic PSEs as a counterparty: 25 25 40 40
  - Funding backed by RMBS eligible for Level 2B: 25 25 40 40
  - Funding backed by other Level 2B assets: 50 50 60 60
  - Other secured funding transactions: 100 100 100 100
- Additional Requirements (selected):
  - Valuation changes non-Level 1 posted collateral securing derivatives: 20 20 35 35
  - Excess collateral held by bank related to derivate transactions that could be called anytime: 100 100 100 100
  - Liquidity needs related to collateral contractually due on derivatives transactions: 100 100 100 100
  - ABCP, SIVs, conduits, SPVs, or similar — Liabilities from maturing: 100 100 100 100; Asset backed securities: 100 100 100 100
- Undrawn but committed credit and liquidity facilities (selected):
  - Retail and small business: 5 5 10 10
  - NFCs, sovereigns, CBs, multilateral dev banks, PSEs — Credit facilities: 10 10 30 30; Liquidity facilities: 30 30 50 50
  - Supervised banks: 40 40 50 50
  - Other financial institutions — Credit facilities: 40 40 50 50; Liquidity facilities: 100 100 100 100
  - Other legal entity customers, credit, and liquidity facilities: 100 100 100 100
- Other contingent funding liabilities:
  - Trade finance: 5 5 10 10
  - Customer short positions covered by customers' collateral: 50 50 75 75
  - Other products and services: 10 10 10 10
  - Additional contractual outflows: 100 100 100 100
  - Net derivative cash outflows (including planned): 100 100 100 100
  - Any other contractual cash outflows (not listed above): 100 100 100 100

### Inflows (selected entries)
- Level 1 assets: 0 0 0 0
- Level 1 assets (extremely liquid): 7 15 7 15
- Level 2a assets: 15 15 15 15
- Level 2b assets:
  - Eligible Residential Mortgage-Backed Securities (RMBS): 25 50 25 50
  - Other: 50 70 50 70
- Margin lending backed by all other collateral: 50 100 50 100
- All other assets: 100 80 100 80
- Loans with an undefined contractual end date: 20 10 20 10
- Net derivative cash inflows: 100 100 100 100
- Other (contractual) cash inflows: 100 80 100 80
- Other inflows, by counterparty:
  - Retail counterparties: 50 35 50 35
  - Nonfinancial wholesale counterparties, transactions not listed above: 50 35 50 35
  - Other inflows from NFCs, non-principal repayment: 100 80 100 80
  - Financial institutions and central banks, transactions not listed above: 100 80 100 80

### Assets — Haircuts / Valuation Rates (selected entries)
- Level 1 assets:
  - Cash: 100 100 100 100
  - Qualifying marketable securities (sovereign, CB, PSEs, MBDs): 100 95 100 95
  - Qualifying CB reserves: 100 100 100 100
  - Domestic Sovereign or CB debt for non-zero risk weighted entities: 100 95 100 95
  - Level 1 extremely high-quality covered bonds: 93 85 93 85
- Level 2A assets:
  - Qualifying marketable securities from sovereigns, central banks, PSEs, and multilateral development banks (with 20% risk weighting): 85 85 85 85
  - Qualifying corporate debt securities rated AA- or higher: 85 85 85 85
  - Qualifying covered bonds rated AA- or better: 85 85 85 85
- Level 2B assets:
  - Qualifying Mortgage-Backed Securities: 75 50 75 50
  - Qualifying corporate debt securities rated between A+ and BBB-: 50 30 50 30
  - Qualifying common equity shares: 50 30 50 30

*Source: Spain — International Monetary Fund (Appendix VII: Liquidity Stress Test Scenario Specification).*

### Appendix VIII. Household Vulnerability Analysis Methodology

### Appendix VIII. Household Vulnerability Analysis Methodology

### Data and general approach
- Micro data source: BdE’s 2020 Survey of Household Finances.
- Purpose: conduct simulations of macroeconomic shocks to household balance sheets and compute household vulnerability measures based on DSTI and essential consumption on food, utilities, and rents.
- Heterogeneity: households divided into five income quintiles; analysis studies differences across income groups.
- Time horizon and shock application:
  - Given survey date 2020, cumulative changes in macroeconomic conditions are applied from 2020 to 2024 and from 2020 to 2026 to assess household balance-sheet health in those two years.
  - The application of macroeconomic variable changes to individual household balance sheets follows Valderrama et al. (2023).

### Definition of household vulnerability
- Two vulnerability definitions used:
  - DSTI greater than or equal to 40 (DSTI >= 40).
  - Debt-service-and-essential-consumption-to-income-ratio (DSECTI) >= 70, defined as debt service plus consumption of food, utilities, and rents exceeding 70 percent of household income.

### Application of macroeconomic shocks (assumptions and transmission)
- Income:
  - 2020 to 2022: extrapolate each household’s income growth using cumulative growth in disposable income per capita.
  - 2023 onwards: use projected wage growth to proxy household income growth.
  - The same growth rate is applied to each household.
- Debt:
  - Simplification: apply the sectoral-wide save ratio of debt growth to all households, recognizing real-world heterogeneity (repayments, new borrowing, early repayment of variable-rate loans).
- Interest payments:
  - Up to 2022: assume adjustable-rate loans had not been repriced.
  - For 2024 onwards: interest rate changes are assumed fully passed through to variable-rate loans; fixed-rate loans are not affected.
- Consumption:
  - Assume no change to the structure of the real consumption basket.
- Prices:
  - Food and energy prices follow global wholesale prices sourced from IMF’s WEO.
  - Rents, non-essential goods, and services are adjusted by core inflation.
- Unemployment rate:
  - Changes in unemployment are applied evenly across the labor force.
  - When a worker becomes unemployed: wage income becomes the average social benefits available to unemployed (unemployment benefit, unemployment assistance, private insurance, and ERTE).
  - When an unemployed worker becomes employed: loss of unemployment-related social benefits and start earning average wage income.

### Summary table for key assumptions (changes relative to 2020 levels)
- Table 1. Spain: Key Macroeconomic Variables Assumptions Summary Table (values are percent differences relative to 2020 levels; unemployment and interest-rate changes are percentage points relative to 2020)
  - 2022
    - Income +8.1
    - Unemployment rate -2.7
    - Mortgage interest rate 0
    - Other household loan interest rate 0
    - Food price +36.7
    - Energy price +226.6
    - CPI +11.4
    - Household debt +2.6
  - 2024, baseline
    - Income +16.4
    - Unemployment rate -3.9
    - Mortgage interest rate +4.3
    - Other household loan interest rate +2.6
    - Food price +24.7
    - Energy price +91.7
    - CPI +18.7
    - Household debt +0.3
  - 2024, adverse
    - Income +16.4
    - Unemployment rate +2.7
    - Mortgage interest rate +6.9
    - Other household loan interest rate +4.5
    - Food price +37.9
    - Energy price +131.5
    - CPI +21.7
    - Household debt -1.7
  - 2026, baseline
    - Income +22.8
    - Unemployment rate -4.5
    - Mortgage interest rate +4.1
    - Other household loan interest rate +2.5
    - Food price +24.9
    - Energy price +74.1
    - CPI +22.5
    - Household debt +0.9
  - 2026, adverse
    - Income +20.5
    - Unemployment rate +5.1
    - Mortgage interest rate +5
    - Other household loan interest rate +2.8
    - Food price +47.1
    - Energy price +136.8
    - CPI +25.8
    - Household debt -6.4

### Computing household vulnerability (formulation)
- DSTI and DSECTI computed per household under scenarios.
- Equations as presented in source (symbols preserved):
  - (first displayed multi-term expression for h and DSTI)
    횫풊 푻−풕,풋 as the interest rate shock from time  
    푫푺푻푰 푻,풋 풉 = ∑(푷 풕,풌 풉 +푶 풕,풌 풉 ×풊 풕,풌 풉)+ ∑(푶 풕,풔 풉 ×횫풊 푻−풕,풋 풔 |풔=풗풂풓풊풂풃풍풆) 푴 풔=ퟏ 푵 풌=ퟏ 푰 풕 풉 ×(ퟏ+횫풊풏풄 푻−풕,풋 )×품푫 푻−풕,풋
  - DSECTI definition (as rendered in source):
    푫푺푬푪푻푰 푻,풋 풉 = 푫푺푻푰 푻,풋 풉 + 풇풐풐풅 풕 풉 × (ퟏ+횫푪푷푰 푻−풕,풋 풇풐풐풅) + 풖풕풊풍풊풕풊풆풔 풕 풉 (ퟏ+횫푪푷푰 푻−풕,풋 풆풏풆풓품풚) + 풓풆풏풕 풕 풉 ×(ퟏ+횫푪푷푰 푻−풕,풋 ) 푰 풕 풉 ×(ퟏ+횫풊풏풄 푻−풕,풋 )
- Implementation notes:
  - Interest-rate passthrough assumptions and household debt growth assumptions feed into DSTI calculations.
  - Price inflation adjustments (food, energy, CPI) feed into essential-consumption components of DSECTI.
  - Unemployment transitions convert wage income to average unemployment-related social benefits when relevant.

*Source: Appendix VIII. Household Vulnerability Analysis Methodology, 1espea2024004-print-pdf*

---


_Source: https://www.imf.org/-/media/files/publications/cr/2024/english/1espea2024004-print-pdf.pdf_