## EXECUTIVE SUMMARY

## Source details

**Canonical URL:** [EXECUTIVE SUMMARY](https://www.imf.org/-/media/files/publications/cr/2022/english/1mexea2022007.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/cr/2022/english/1mexea2022007.pdf.md)
- [Structured JSON version](/-/media/files/publications/cr/2022/english/1mexea2022007.pdf.json)

---

### Macrofinancial context
- Growth: 2020 contracted by 8.1 percent; 2021 growth rebounded to 4.8 percent; growth is currently forecast at 2.4 percent in 2022.
- Labor market: unemployment rate was at 3.3 percent in May 2022.
- Inflation and monetary policy:
  - Inflation reached 8.7 percent in August (the highest reading since 2001).
  - Banco de México raised the policy rate by a cumulative 525 bps since June 2021, including a 75 bps increase to 9.25 percent in September 2022.
  - The Governing Board signaled intention to continue raising the reference rate if conditions so require.

### Financial sector structure and recent performance
- System size and composition:
  - Total financial system assets about 100 percent of GDP, smaller than EM peers.
  - Banking sector (commercial and development banks) accounts for more than half of the financial system.
  - Pension funds: about 20 percent of financial sector assets; insurers: about 8 percent.
  - Stock market capitalization: 31 percent of GDP.
- Banking sector concentration and major institutions:
  - Mexico has 50 commercial banks with total assets at approximately 11 trillion pesos (43 percent of the financial sector’s assets) in 2021.
  - Six D-SIBs comprise almost ¾ of total banking sector assets.
  - Table 2 snapshots (selected entries preserved):
    - Commercial banks: 2016 — 47 banks, 8,668 billion pesos; 2021 — 50 banks, 11,078 billion pesos.
    - Foreign subsidiaries: 2016 — 15 banks, 5,865 billion pesos; 2021 — 20 banks, 7,435 billion pesos.
    - Pension funds (Siefores): 2016 — 2,754 billion pesos; 2021 — 5,236 billion pesos.
    - Total assets: 2016 — 18,034 billion pesos; 2021 — 25,271 billion pesos.
- Development banks:
  - Six development banks represent 9 percent of financial sector assets.
  - Sovereign backstops their capital and fully guarantees their liabilities.
- Investment funds and pension funds:
  - Investment funds’ assets: 2.8 trillion pesos (10 percent of GDP).
  - Pension funds increased from 2.7 trillion pesos (15 percent of GDP) in 2016 to 5.2 trillion pesos (20 percent of GDP) in 2021.
- Banking sector performance indicators:
  - Credit-to-GDP ratio about 40 percent; credit-to-GDP gap is negative.
  - Net interest margin (net of loan-loss provisions): around 5 percent of interest-bearing assets in 2021.
  - Profitability 2021: ROA around 2 percent; ROE around 18 percent.
  - Aggregate capital adequacy ratio increased to 19.5 percent at end-2021.
  - Aggregate LCR and NSFR at end-2021: above 200 and 120 percent respectively.
- Financial soundness indicators (selected time series preserved exactly):
  - Regulatory capital to risk-weighted assets: 2016 14.9; 2017 15.6; 2018 15.9; 2019 16.0; 2020 17.7; 2021 19.5; 2022 Q2 18.7.
  - Nonperforming loans to total gross loans: 2016 2.1; 2017 2.1; 2018 2.1; 2019 2.1; 2020 2.4; 2021 2.0; 2022 Q2 2.3.
  - Return on assets: 2016 1.7; 2017 2.0; 2018 2.2; 2019 2.2; 2020 1.2; 2021 2.1; 2022 Q2 1.9.
  - Liquid assets to total assets: 2016 31.4; 2017 32.0; 2018 31.6; 2019 31.1; 2020 35.7; 2021 36.3; 2022 Q2 38.5.
  - Customer deposits to total loans, excluding interbank loans: 2016 88.9; 2017 91.4; 2018 89.3; 2019 90.7; 2020 100.2; 2021 105.2; 2022 Q2 99.5.

### Key macrofinancial risk drivers and adverse scenario narrative
- Principal risk: sustained tightening of global liquidity conditions since the Global Financial Crisis combined with risks of lower growth and higher inflation.
- Potential transmission channels:
  - Disorderly tightening could weaken activity, erode banks’ net interest margins via higher corporate defaults, induce exchange rate volatility, and trigger system-wide liquidity stress.
  - Reemergence of pandemic waves could worsen inflation via supply-chain channels and underpin the adverse stress scenario.
- Adverse scenario (2022–2024, starting end-2021; baseline aligned with IMF projections as of June 15, 2022):
  - Real GDP level falls about 11 percent below baseline by 2023 (equivalent to 2¼ standard deviation cumulative 2-year growth rate shock).
  - Selected indicator paths (2021–2024 preserved exactly):
    - Real GDP (2021=100) — Baseline: 100.0, 102.4, 104.6, 106.0; Adverse: 100.0, 97.5, 93.9, 98.3.
    - Real GDP Growth Rate (percent, y-o-y) — Baseline: 4.8, 2.4, 2.2, 1.4; Adverse: 4.8, -2.5, -3.7, 4.6.
    - CPI Inflation Rate (percent, y-o-y) — Baseline: 5.7, 7.2, 4.4, 3.3; Adverse: 5.7, 9.6, 8.7, 5.8.
    - Exchange Rate (MXN per USD, end of period) — Baseline: 20.6, 21.4, 21.6, 21.9; Adverse: 20.6, 24.1, 26.1, 25.8.
    - Policy Rate (percent, year averages except 2021) — Baseline: 5.5, 7.9, 9.1, 8.1; Adverse: 5.5, 9.4, 10.1, 6.9.
    - 10-year Sovereign Bond Yield (percent) — Baseline: 8.0, 9.4, 9.4, 9.4; Adverse: 8.0, 10.9, 11.2, 10.2.
    - Equity prices (2021=100) — Baseline: 100.0, 106.7, 113.3, 119.7; Adverse: 100.0, 90.3, 78.1, 76.0.
    - Housing prices (2021=100) — Baseline: 100.0, 106.7, 113.3, 119.7; Adverse: 100.0, 82.6, 82.6, 105.1.
    - Price of Commodities (energy/oil, 2016 = 100) — Baseline: 184.4, 346.5, 262.4, 224.1; Adverse: 184.4, 443.9, 258.8, 191.9.
  - Long term yields assumed sustained at 11 percent on average during 2022-23 in the adverse narrative.

### Solvency stress-test findings (IMF top-down)
- Coverage and approach:
  - Top-down solvency stress test on ten largest commercial banks capturing >80 percent of commercial banking system assets; sample included all six D-SIBs.
  - Static balance sheet assumption; three-year projection horizon; cut-off date end-2021.
  - Hurdle rates: CET1 4.5 percent, T1 6 percent, total capital 8 percent; baseline included additional 2.5 percent CCB and bank-specific D-SIB surcharge included.
  - Dividend payout assumptions: 40 percent for year 1 and 50 percent for years 2 and 3; zero payout for loss-making years.
- Main results and attribution:
  - System-wide resilience: most banks remain above minimum requirements in the adverse scenario; aggregate capital shortfalls relatively small (less than 0.4 percent of GDP).
  - Aggregate CAR dynamics (exact figures):
    - Aggregate CAR declines by 4.7 percentage points to 14.5 percent by 2022 under adverse scenario.
    - In the baseline, aggregate CAR increases by 2.8 percentage points to 22 percent by 2024.
    - In the adverse scenario, aggregate CAR would decline by 4.1 percentage points to 15.1 percent in 2023.
  - Drivers of capital depletion:
    - Material market losses from rise in interest rates and credit spreads.
    - Increase in loan loss provisions due to credit deterioration.
    - Cumulative credit losses up to year 2: 6.1 percentage points of capital under adverse scenario versus 2.8 percentage point decline in baseline.
    - Net interest income in adverse scenario lower on average by 1.3 percentage points of capital for first two years; NII reduced by approximately 14.2 percent during the first two years relative to cut-off year.
    - Cumulative NTI and OCI losses: 2.2 percentage points of capital under baseline and 3.6 percentage points under adverse.
  - Heterogeneity and mitigants:
    - Substantial heterogeneity across banks; consumer-focused banks and those with sovereign concentration or weaker starting capital more vulnerable.
    - Negative capital impact partially mitigated by reduced dividend distributions and tax differences.

### Key solvency sensitivities and stress tests
- Contingent credit lines:
  - Contingent credit lines at end-2021: 1.6 trillion pesos — almost 15 percent of total banking sector’s assets or 31 percent of total loan book.
  - Majority revocable; central sensitivity assumed 25 and 30 percent of outstanding lines would be triggered through three years of the adverse scenario.
  - Triggering reduces landing capital ratio by 2024 by almost 2 percentage points relative to the normal adverse scenario.
- Additional +150 bps interest-rate shock on top of adverse:
  - Capitalization impact: decline by 2.4 percentage points by 2022 or 3.1 percentage points by end of horizon.
  - Losses mainly from FV and OCI portfolios; credit losses increase but mitigated by increased NII.
- Digital payments / private-money scenarios:
  - Most adverse private-money scenario: system total capital depletion of 34 basis points by 2024 relative to baseline solvency stress test.
  - Some banks could see capital ratios decline up to 75 basis points by 2024.
  - CBDC scenario with cap of UDI 3000 (approximately 21,000 Mexican pesos): capital adequacy ratio drop of 25 basis points.
- Market risk materiality:
  - Revaluation of debt securities drives market risk impact: "3.6 percent of RWAs during year 1 of the baseline and almost 6 percent under the adverse."
  - Impact almost entirely driven by bond securities portfolios; FVTP portfolio impact almost double.

### Deferred loans and loan performance (selected figures)
- Deferred loans accounted for 13 percent of total outstanding loans at end-2021 and have been phased-out.
- Banks restructured about 17 percent of loans that benefitted from the deferral program; these restructured loans are currently mostly performing.
- Table 5 (end-2021, amounts in millions of Mexican pesos, selected entries preserved):
  - Total bank loans 1/: 5,459,257
  - Bank loans benefitted from loan deferral program: 1,067,334
  - Amount of loans under deferred loan category: 708,660
    - Performing: 553,424
    - Restructured: 121,024
    - Nonperforming: 34,212
- Performance shares (percent, end-2021):
  - Performing share of loans under deferred loan category: Total 78.1; Commercial 73.1; Consumer 72.2; Mortgage 88.0.
  - Nonperforming share: Total 4.8; Commercial 4.1; Consumer 5.6; Mortgage 4.8.

### Liquidity and system-wide liquidity analysis
- Novel system-wide liquidity framework:
  - Traces liquidity linkages among agents, assesses transmission/amplification channels, evaluates liquidity capacity, and conducts policy counterfactuals.
- Role of commercial banks:
  - Commercial banks backstop liquidity needs of other agents, acting as shock absorbers through repo transactions.
  - Under most severe narratives and assuming no binding regulatory constraints, commercial banks show only marginal liquidity shortfalls.
- Vulnerabilities outside commercial banks:
  - Development banks appear more vulnerable; binding liquidity constraints lead to larger system shortfalls.
  - Investment funds are mainly invested in liquid instruments (government securities and repo) and are potential sources of short-term funding but have limited access to repo markets currently.
- Key system-wide metrics and cash-flow analysis results:
  - Cash-flow three-month severe scenario: 21 banks (representing almost 70 percent of total banking sector assets) would encounter liquidity shortfalls; combined bank-level shortfall 0.3 trillion pesos (4 percent of total assets); aggregate net shortfall after netting surpluses and shortfalls around 0.1 trillion pesos (1 percent of total assets).
  - LCR-based stress test (30 days) scenarios: Aggregate LCR = 225 percent (regulatory scenario); Retail Shock aggregate LCR 157 percent with 6 banks LCR<100; Wholesale Shock aggregate LCR 128 percent with 13 banks LCR<100.
  - Aggregate LCR in national currency: 182 percent; in U.S. dollars: 348 percent.
  - NSFR end-2021: 145 percent.
- Contingent credit lines and deposit behavior risks:
  - Undrawn credit lines (May 2022): Revocable 2.5 trillion pesos; Irrevocable 0.25 trillion pesos; Liquidity lines 0.15 trillion pesos.
  - Retail deposits from high net-worth individuals may behave like wholesale deposits, warranting reassessment of deposit classification in liquidity frameworks.
- System-wide liquidity analysis results and policy insights:
  - Commercial banks act as final shock absorbers via repo; development banks’ vulnerability amplified when commercial banks face binding constraints or act conservatively.
  - Promoting investment funds’ participation in the repo market could reinforce system-wide resiliency.
  - Higher correlation of liquidity shocks amplifies downside risks and fatten left tails of liquidity distributions for commercial banks.

### Corporate risk analysis and machine learning application
- Corporate defaults rise under the adverse scenario; model outcomes point to stronger corporate default paths in the adverse stress scenario.
- Machine learning tools used (penalized regressions, Random Forests) outperformed linear regressions in capturing non-linearities and interaction effects.
- Model performance (Mexico sample cross-fold validation averages preserved):
  - Average R2s varied from about 0.35 for linear regression to about 0.65 for Random Forest.
  - Random Forest outperformed other methods across cross-validation groupings.
- Scenario simulation results:
  - Modeled EDFs under solvency stress scenario: risks rose even in the baseline due to rising interest rates; in the stress scenario EDFs rose roughly tripling baseline levels in 2023 and 2024 (modeled rise worse than the modeled rise of slightly less than double estimated elsewhere).

### Contagion, interconnectedness and cross-border linkages
- Cross-border concentrations (2021Q4):
  - Total foreign claims of Mexican banks: US$37 billion; United States comprises around 80 percent of total foreign claims.
  - Claims of foreign banks in Mexico: US$130 billion; United States and Spain cover 55 percent of total claims on Mexico.
- Network simulation results:
  - Inward and outward spillovers to/from Mexico are relatively modest.
  - Hypothetical distress in Mexico would impact banks in Spain the most through both channels (loss of approximately 5 percent of bank capital).
  - Hypothetical distress in the United States would have most significant impact on banks in Mexico (loss of 30 percent of bank capital) under specified assumptions.
- Domestic network (bilateral exposures dataset of 132 entities as of December 9, 2021):
  - Development banks are the largest net borrowers; commercial banks key providers of funds.
  - Failure of a development bank causes average loss of 1.8 percent of counterparties’ capital; default of a commercial bank causes average loss of 0.2 percent.

### Policy recommendations (summarized, priorities and timing preserved)
- 1. Address bank idiosyncratic risk profiles by applying appropriately calibrated capital add-ons through Pillar II requirements; enhance prudential oversight over concentration risks and business models. Responsible: CNBV. Timing: NT. Priority: H.
- 2. Monitor the dynamics of contingent credit lines closely and assess relevant risks. Responsible: Banxico, CNBV. Timing: NT. Priority: H.
- 3. Incorporate the liquidity analysis in the Supervisory Review Process (SRP) to inform Pillar 2 capital requirements for banks. Responsible: Banxico, CNBV. Timing: MT. Priority: M.
- 4. Continue work on bridging data gaps (focusing on IFRS 9 and full IRB model application) and further enhance data reporting and modelling capacity. Responsible: CNBV, Banxico. Timing: MT. Priority: M.
- 5. Improve robustness checks in the liquidity stress test framework to address that part of retail deposits (of high net-worth individuals) that could behave as wholesale deposits. Responsible: Banxico. Timing: NT. Priority: M.
- 6. Utilize granular collected data to set up a regular maturity ladder template and use cash flow to further complement the current stress test framework. Responsible: Banxico. Timing: NT. Priority: M.
- 7. Consider incorporating and adjusting the system-wide liquidity analysis, ideally at the entity level, to monitor relative contributions of different agents to liquidity stress and identify/assess policies to strengthen resiliency. Responsible: Banxico. Timing: MT. Priority: M.
- Timing legend preserved: C: continuous; I: immediate (<1 year); NT: short term (1–2 years); MT: medium term (3–5 years).
- Priority legend preserved: H: high; M: medium; L: low.

*Technical Note prepared by Dimitrios Laliotis (lead), Priscilla Toffano, Xiaodan Ding, Kevin Wiseman, Padamja Khandelwal, Lu Zhang, and Sujan Lamichhane under the guidance of Vikram Haksar and Heedon Kang (IMF).*

### EXECUTIVE SUMMARY __________________________________________________________________________ 7

### EXECUTIVE SUMMARY

### Macrofinancial context
- The Mexican economy experienced its deepest recession in decades: growth in 2020 contracted by 8.1 percent.
- Rebound in 2021: growth rebounded to 4.8 percent.
- 2022 forecast: growth is currently forecast at 2.4 percent in 2022.
- Labor market: unemployment rate was at 3.3 percent in May 2022.
- Inflation dynamics:
  - Inflation reached 8.7 percent in August (the highest reading since 2001).
  - Banco de México raised the policy rate by a cumulative 525 bps since June 2021, including a 75 bps increase to 9.25 percent in September 2022.
  - The Governing Board signaled intention to continue raising the reference rate if conditions so require.

### Financial sector structure and recent performance
- The financial system is smaller than in peer countries and dominated by commercial banks with large capital and liquidity buffers.
- Banking sector characteristics:
  - High profitability in the banking sector.
  - Credit growth has been low due to both supply and demand factors; banks target mainly prime segments.
- COVID-19 impact: The pandemic had a limited impact on the financial system, reflecting resumption in mobility and support from global and domestic policies.

### Key macrofinancial risk drivers and stress scenario narrative
- Principal risk: the first sustained and ongoing tightening of global liquidity conditions since the Global Financial Crisis, combined with risks of lower growth and higher inflation.
- Potential transmission channels and adverse outcomes:
  - Disorderly tightening in global and domestic financial conditions could further weaken activity.
  - Erosion of banks’ net interest margins via higher corporate defaults.
  - Exchange rate volatility.
  - System-wide liquidity stress in Mexico.
- Pandemic risk: possible reemergence of pandemic waves could worsen inflation via supply-chain channels and underpin the adverse stress scenario.

### Solvency stress-test findings
- Overall resilience:
  - Banks are resilient to severe macrofinancial shocks; capital and liquidity ratios for most banks in the sample would still be above minimum requirements in an adverse scenario, with limited shortfalls for some banks.
  - Solid internal capital generation capacity due to robust interest margins and ample capital buffers helps the system withstand severe shocks.
- Areas for close monitoring:
  - Large contingent credit lines extended by banks to corporates are a key weakness in both solvency and liquidity stress tests; the majority are revocable credit lines, which attenuates the risk.
  - Exposures to contingent credit lines are unevenly distributed among banks and could be triggered quickly and simultaneously during crises.
  - Risks from large exposures, concentration risks, and bank-specific business-model linked risks merit supervisory attention.
  - Retail deposits from high net-worth individuals seeking higher yields could behave like wholesale deposits and be prone to outflows.

### Liquidity and system-wide liquidity analysis
- Novel framework:
  - A system-wide liquidity framework was developed to trace liquidity linkages among various agents, understand transmission and amplification channels, evaluate liquidity capacity, and conduct policy counterfactual experiments.
- Role of commercial banks:
  - Commercial banks ensure the liquidity of the financial system by backstopping liquidity needs of other agents, acting as shock absorbers through repo transactions.
  - Commercial banks show only marginal liquidity shortfalls even under the most severe narratives.
- Vulnerabilities outside commercial banks:
  - Development banks appear more vulnerable during stress, with binding liquidity constraints (e.g., mandatory LCRs for commercial banks or minimum liquidity buffers for investment funds) leading to larger system shortfalls.
  - Under binding constraints, agents with liquidity surplus may be less willing to roll over funding, amplifying stress on development banks’ liability side.
- Policy insight: promoting participation of investment funds in the repo market could reinforce system-wide resiliency and liquidity conditions.

### Corporate risk analysis
- Corporate defaults:
  - Corporate defaults would rise under the adverse scenario, consistent with the bank solvency stress test.
- Methodology:
  - Machine learning tools were used to model corporate default risk because they outperformed linear regressions in capturing non-linearities and interaction effects.
- Implication:
  - Model outcomes point to stronger corporate default paths in the adverse stress scenario and call for additional caution about the potential effects of greater financial tightening.

### Key policy recommendations (Table 1 summary)
- 1. Address bank idiosyncratic risk profiles by applying appropriately calibrated capital add-ons through Pillar II requirements; enhance prudential oversight over concentration risks and business models. Responsible: CNBV. Timing: NT. Priority: H.
- 2. Monitor the dynamics of contingent credit lines closely and assess relevant risks. Responsible: Banxico, CNBV. Timing: NT. Priority: H.
- 3. Incorporate the liquidity analysis in the Supervisory Review Process (SRP) to inform Pillar 2 capital requirements for banks. Responsible: Banxico, CNBV. Timing: MT. Priority: M.
- 4. Continue work on bridging data gaps (focusing on IFRS 9 and full IRB model application) and further enhance data reporting and modelling capacity. Responsible: CNBV, Banxico. Timing: MT. Priority: M.
- 5. Improve robustness checks in the liquidity stress test framework to address that part of retail deposits (of high net-worth individuals) that could behave as wholesale deposits. Responsible: Banxico. Timing: NT. Priority: M.
- 6. Utilize granular collected data to set up a regular maturity ladder template and use cash flow to further complement the current stress test framework. Responsible: Banxico. Timing: NT. Priority: M.
- 7. Consider incorporating and adjusting the system-wide liquidity analysis, ideally at the entity level, to monitor relative contributions of different agents to liquidity stress and identify/assess policies to strengthen resiliency. Responsible: Banxico. Timing: MT. Priority: M.
- Timing legend: C: continuous; I: immediate (<1 year); NT: short term (1–2 years); MT: medium term (3–5 years).
- Priority legend: H: high; M: medium; L: low.

*Technical Note prepared by Dimitrios Laliotis (lead), Priscilla Toffano, Xiaodan Ding, Kevin Wiseman, Padamja Khandelwal, Lu Zhang, and Sujan Lamichhane under the guidance of Vikram Haksar and Heedon Kang (IMF).*

### 3.     The structural current account balance

### 3.     The structural current account balance

### Current account and capital flows
- In 2020, strong U.S. demand, remittance inflows, and weak domestic demand led to a current account surplus of 2.5 percent of GDP.
- The current account returned to a deficit in 2021 and is expected to remain a deficit over the medium term.
- Portfolio flows have been volatile, with net capital flows (in billions of U.S. dollar) showing large swings across years (Text Chart indicators range shown between -120 and 200 on the left axis and -30 to 50 on the right-hand scale for the 4Q sum).
- Portfolio flows vulnerability is similar to other emerging markets (EMs).

### External financial integration and FX market characteristics
- Foreign investors hold about a sixth of the outstanding local currency government bonds, although the share has fallen sharply since 2017, with domestic banks picking up the slack.
- The sovereign had issued 7.2 percent of GDP in external debt as of end-2021.
- Mexican non-financial corporates (NFCs) have sizable debt issuance offshore: 13 percent of GDP, mainly long-term bonds.
- The Mexican peso (MXN) is widely used as a proxy for EM currencies due to high liquidity and a large global market for exchange traded MXN derivatives.
- Trading volumes of MXN on major exchanges are significantly higher than most other EM currencies and are comparable to those of major currencies.
- MXN exhibits a high correlation (beta) with global risk shocks; MXN volatility tends to increase the most among peers in periods of global risk aversion.

### Financial sector structure (size and composition)
- The financial system has total assets of about 100 percent of GDP, smaller than in EM peers.
- The banking sector (commercial and development banks) accounts for more than half of the financial system.
- Pension funds and insurers account for about 20 percent and 8 percent of total financial sector assets, respectively.
- Domestic debt markets are focused on sovereign securities; stock market capitalization stood at 31 percent of GDP, small relative to peers.

### Banking sector concentration and major institutions
- Mexico has 50 commercial banks with total assets at approximately 11 trillion pesos (43 percent of the financial sector’s assets) in 2021.
- The six domestic systemically important banks (D-SIBs) comprise almost ¾ of total banking sector assets and play a leading role within their respective financial conglomerate.
- Five of the D-SIBs are foreign subsidiaries that generate a large share of the parent groups’ profits.
- Table 2 (2016 vs 2021 snapshots) highlights:
  - Commercial banks: 2016 — 47 banks, 8,668 billion pesos, 48.1 percent of financial sector assets, 43.1 percent of GDP; 2021 — 50 banks, 11,078 billion pesos, 43.8 percent of financial sector assets, 42.2 percent of GDP.
  - Domestic banks: 2016 — 32 banks, 2,803 billion pesos, 15.5 percent of financial sector assets, 13.9 percent of GDP; 2021 — 30 banks, 3,643 billion pesos, 14.4 percent of financial sector assets, 13.9 percent of GDP.
  - Foreign subsidiaries: 2016 — 15 banks, 5,865 billion pesos, 32.5 percent of financial sector assets, 29.1 percent of GDP; 2021 — 20 banks, 7,435 billion pesos, 29.4 percent of financial sector assets, 28.3 percent of GDP.
  - D-SIBs: 2016 — 7 banks, 6,879 billion pesos, 38.1 percent of financial sector assets, 34.2 percent of GDP; 2021 — 6 banks, 8,099 billion pesos, 32.0 percent of financial sector assets, 30.8 percent of GDP.
  - Pension funds (Siefores): 2016 — 73 funds, 2,754 billion pesos, 15.3 percent of financial sector assets, 13.7 percent of GDP; 2021 — 117 funds, 5,236 billion pesos, 20.7 percent of financial sector assets, 19.9 percent of GDP.
  - Investment funds (Fondos de inversión): 2016 — 578 funds, 2,047 billion pesos, 11.4 percent of financial sector assets, 10.2 percent of GDP; 2021 — 613 funds, 2,795 billion pesos, 11.1 percent of financial sector assets, 10.6 percent of GDP.
  - Total assets: 2016 — 2,845 entities, 18,034 billion pesos, 100.0 percent of financial sector assets, 89.6 percent of GDP; 2021 — 2,337 entities, 25,271 billion pesos, 100.0 percent of financial sector assets, 96.2 percent of GDP.
- Memo: Financial holding companies (FHCs): 2016 — 10 FHCs, 6,546 billion pesos, 36.3 percent of financial sector assets, 32.5 percent of GDP; 2021 — 15 FHCs, 8,798 billion pesos, 34.8 percent of financial sector assets, 33.5 percent of GDP.

### Development banks and targeted finance
- Six development banks represent 9 percent of financial sector assets.
- They provide finance to long-term projects (e.g., infrastructures), SMEs, exporters, housing, and low-income populations via first-tier and second-tier loans and guarantees.
- Credit growth extended by development banks continues a long decline (Text Chart).
- The sovereign backstops their capital and fully guarantees their liabilities.

### Investment funds and pension funds trends
- From January 2020 to May 2022, the number of investment funds’ contracts increased by almost 60 percent.
- Assets under management of investment funds are currently at 2.8 trillion pesos (10 percent of GDP).
- Investment funds’ assets are mainly concentrated in liquid investments (e.g., government securities and repo agreements); equity investment represents only 10 percent of their portfolio.
- Pension funds increased assets from 2.7 trillion pesos (15 percent of GDP) in 2016 to 5.2 trillion pesos (20 percent of GDP) in 2021.

### Other nonbank financial institutions (NBFIs)
- Credit unions, sofomes, socaps, sofipos, and other financial companies account for 10 percent of the system’s assets.
- Regulated NBFIs represent around 2 percent of financial sector assets.
- These entities are generally small, non-deposit taking or have little interconnection with the banking system and are not systemic.

### Banking sector recent performance and inclusion
- Aggregate capital and liquidity ratios are high relative to EM peers; credit risk is moderate; profitability is high.
- Credit-to-GDP ratio is low at about 40 percent and has not reached pre-1994 crisis levels; the credit-to-GDP gap is negative.
- High informality and banks’ focus on prime corporate sector segments and high net-worth individuals help explain subdued credit growth.

### Loan composition, profitability, and margins
- The aggregate credit portfolio (2012–2021) is concentrated on corporate loans (about half of total exposure), followed by mortgages and consumption loans.
- Interest income from the credit portfolio and investments in securities make up almost 80 percent of banks’ revenue.
- Net interest margin, net of loan-loss provisions, was around 5 percent of interest-bearing assets in 2021.
- Profitability indices in 2021: ROA around 2 percent; ROE around 18 percent.

### Pandemic impact and policy responses
- During the pandemic, Mexico experienced capital outflows and a sharp exchange rate depreciation, but sovereign and corporate spreads remained low and market functioning was orderly.
- Banxico cut policy rates seven times from March 2020 through May 2021, 300 basis points in total; and hiked policy rate ten times since June 2021, 525 basis points so far.
- Central bank facilities during the COVID-19 crisis (In billions of Mexican peso):
  - Government securities term repurchase window: Envelope 150; Disbursed 465; Percent (B/A) 310; Expiration Sep. 2021.
  - Reduction of the Monetary Regulatory Deposit: Envelope 50; Disbursed 50; Percent (B/A) 100; Expiration Nov 2020.
  - Temporary securities swap window: Envelope 50; Disbursed 63; Percent (B/A) 126; Expiration Sep. 2021.
  - Swap of government securities: Envelope 100; Disbursed 15; Percent (B/A) 15; Expiration Feb. 2021.
  - Corporate Securities Repurchase Facility: Envelope 100; Disbursed 45; Percent (B/A) 45; Expiration Sep. 2021.
  - Provision of resources to banking institutions to channel credit to MSMEs and individuals affected by the COVID-19 pandemic: Envelope 250; Disbursed 14; Percent (B/A) 6; Expiration Sep. 2021.
  - Collateralized financing facility for commercial banks with corporate loans to finance MSMEs: Envelope 100; Disbursed 40; Percent (B/A) 40; Expiration Sep. 2021.
  - Total: Envelope 800; Disbursed 692; Percent (B/A) 86.
  - Total (percent of GDP): 3.1 (Envelope); 2.7 (Disbursed).
- Other financial sector measures included FX swap line with the U.S. Fed by December 2021 and the FCL arrangement with the IMF in November 2021; FX Hedging auction program (USD NDF auctions); temporary flexibilities on liquidity requirements; Special Account Criteria (SAC) for payment deferrals; credit restructuring measures; use of bank’s capital conservation buffer up to 50 percent; restrictions and later partial relaxation on dividend payouts; relief on minimum credit card payments by January 2021.
- Liquidity flexibilities were gradually undrawn by February 2022.

### Market developments and private sector leverage
- The peso depreciated sharply in April 2020 but returned to its pre-pandemic range.
- The yield curve has shifted up amid inflation concerns, tending to flatten.
- Sovereign and corporate risk spreads ticked up recently but remain low compared to peers.
- Credit extended by commercial banks contracted during the pandemic, particularly for corporates and consumer loans, and is gradually resuming to pre-pandemic levels.
- Private sector leverage and debt service burden remained low after the pandemic; the credit-growth increase seen in other EMs did not materialize in Mexico.

### Capital and liquidity buffers (post-pandemic)
- Aggregate capital adequacy ratio increased to 19.5 percent at end-2021.
- Higher capital levels reflect high profitability, preparation for TLAC implementation by D-SIBs, and buffers built due to pandemic-linked dividend payout restrictions.
- Aggregate Liquidity Coverage Ratio (LCR) and Net Stable Funding Ratio (NSFR) were above 200 and 120 percent at end-2021, respectively.
- Financial Soundness Indicators (In percent):
  - Regulatory capital to risk-weighted assets: 2016 14.9; 2017 15.6; 2018 15.9; 2019 16.0; 2020 17.7; 2021 19.5; 2022 Q2 18.7.
  - Regulatory Tier 1 capital to risk-weighted assets: 2016 13.2; 2017 14.2; 2018 14.2; 2019 14.4; 2020 16.1; 2021 18.1; 2022 Q2 17.3.
  - Capital to assets: 2016 9.9; 2017 10.4; 2018 10.7; 2019 11.0; 2020 10.7; 2021 11.8; 2022 Q2 10.8.
  - Nonperforming loans to total gross loans: 2016 2.1; 2017 2.1; 2018 2.1; 2019 2.1; 2020 2.4; 2021 2.0; 2022 Q2 2.3.
  - Provisions to Nonperforming loans: 2016 157.1; 2017 154.9; 2018 152.4; 2019 146.2; 2020 160.1; 2021 160.5; 2022 Q2 143.4.
  - Return on assets: 2016 1.7; 2017 2.0; 2018 2.2; 2019 2.2; 2020 1.2; 2021 2.1; 2022 Q2 1.9.
  - Return on equity: 2016 16.3; 2017 19.6; 2018 20.9; 2019 20.5; 2020 11.7; 2021 18.6; 2022 Q2 17.4.
  - Interest margin to gross income: 2016 73.8; 2017 73.3; 2018 74.5; 2019 74.3; 2020 76.0; 2021 72.7; 2022 Q2 76.0.
  - Trading income to total income: 2016 4.4; 2017 5.0; 2018 4.5; 2019 5.8; 2020 5.5; 2021 6.7; 2022 Q2 9.4.
  - Liquid assets to total assets: 2016 31.4; 2017 32.0; 2018 31.6; 2019 31.1; 2020 35.7; 2021 36.3; 2022 Q2 38.5.
  - Liquid assets to short-term liabilities: 2016 42.4; 2017 42.2; 2018 42.3; 2019 40.8; 2020 48.0; 2021 47.0; 2022 Q2 49.8.
  - Customer deposits to total loans, excluding interbank loans: 2016 88.9; 2017 91.4; 2018 89.3; 2019 90.7; 2020 100.2; 2021 105.2; 2022 Q2 99.5.
  - Net open position in foreign exchange to capital: 2016 0.8; 2017 1.1; 2018 1.2; 2019 2.9; 2020 1.8; 2021 0.1; 2022 Q2 2.2.

*MEXICO — INTERNATIONAL MONETARY FUND*

### 15.     But some banks may be more vulnerable.

### 15.     But some banks may be more vulnerable.

### Deferred loans, restructurings, and current performance
- Deferred loans accounted for 13 percent of total outstanding loans at end-2021 and have been already phased-out.
- Banks have restructured about 17 percent of loans that benefitted from the deferral program; these restructured loans are currently mostly performing.
- Table 5 (end-2021, amounts in millions of Mexican pesos):
  - Total bank loans 1/: 5,459,257
  - Total Commercial: 2,560,732; Consumer: 837,478; Mortgage: 1,118,610
  - Bank loans benefitted from loan deferral program: 1,067,334
    - Commercial: 499,978; Consumer: 243,083; Mortgage: 324,273
  - Amount of loans voluntarily reduced by banks: 358,674
    - Commercial: 272,226; Consumer: 26,421; Mortgage: 60,027
  - Amount of loans under deferred loan category: 708,660
    - Commercial: 227,752; Consumer: 216,662; Mortgage: 264,246
    - Performing: 553,424 (Commercial: 166,428; Consumer: 156,518; Mortgage: 232,478)
    - Restructured: 121,024 (Commercial: 51,895; Consumer: 48,048; Mortgage: 21,081)
    - Nonperforming: 34,212 (Commercial: 9,429; Consumer: 12,096; Mortgage: 12,687)
- Shares of total commercial bank loans (percent, end-2021):
  - Bank loans benefitted from loan deferral program: 19.6 (Total), 19.5 (Commercial), 29.0 (Consumer), 29.0 (Mortgage)
  - Amount of loans voluntarily reduced by banks: 6.6 (Total), 10.6 (Commercial), 3.2 (Consumer), 5.4 (Mortgage)
  - Amount of loans under deferred loan category: 13.0 (Total), 8.9 (Commercial), 25.9 (Consumer), 23.6 (Mortgage)
    - Performing: 10.1 (Total), 6.5 (Commercial), 18.7 (Consumer), 20.8 (Mortgage)
    - Restructured: 2.2 (Total), 2.0 (Commercial), 5.7 (Consumer), 1.9 (Mortgage)
    - Nonperforming: 0.6 (Total), 0.4 (Commercial), 1.4 (Consumer), 1.1 (Mortgage)
- Performance of loans under deferred loan category (percent of total loans under deferred loan category, end-2021):
  - Amount of loans under deferred loan category: 100.0 (Total and by segment)
  - Performing: 78.1 (Total); 73.1 (Commercial); 72.2 (Consumer); 88.0 (Mortgage)
  - Restructured: 17.1 (Total); 22.8 (Commercial); 22.2 (Consumer); 8.0 (Mortgage)
  - Nonperforming: 4.8 (Total); 4.1 (Commercial); 5.6 (Consumer); 4.8 (Mortgage)

### FSAP systemic risk analysis components and methodologies
- Quantitative approaches included:
  - (i) bank solvency tests to assess resilience to credit, interest rate, and market risks;
  - (ii) bank liquidity stress tests;
  - (iii) a novel system-wide liquidity analysis focusing on liquidity interlinkages between agents;
  - (iv) interconnectedness and contagion risk analysis for cross-exposures between domestic institutions;
  - (v) machine learning techniques to assess corporate vulnerabilities and project corporate default rates.
- Sensitivity analyses included:
  - Impact of material contingent credit lines being drawn in a downturn.
  - Additional sharp rise in interest rates: simulated an additional shift by 150 bps in domestic interest rates on top of the adverse scenario.
- Climate risks in banking stress tests:
  - Physical risk shocks: estimated impact of increased hydrometeorological hazards from country-specific historical data used to anchor additional shocks.
  - Transition risks: considered across corporate bank exposures using the solvency stress testing machinery.
- System-wide liquidity analysis:
  - Used aggregated sectoral balance sheet data and cross-linkages between commercial and development banks, investment and pension funds, NFCs, and households.
  - Simulated liquidity and asset revaluation narratives to capture transmission of sovereign securities revaluation and liquidity outflow shocks.
- Contagion and funding risk analysis:
  - Used exposures among banks and NBFIs and BIS International Banking Statistics.
  - Simulated failure of one or several entities to assess cascade effects across domestic intercompany and cross-border networks.
- NFC vulnerabilities:
  - Firm-level data and machine learning estimation used to estimate firm distress shares and implied PD paths; results point to stronger corporate default paths for listed companies and even more severe paths when extrapolated to a larger sample.
- Digital payments sensitivity:
  - Hypothetical sensitivity analysis where banks experience erosion of net interest income (competing for retail sight deposits and paying higher rates) and reduced fee income from payments services due to increased competition from new forms of digital payments (e.g., CBDC).

### Macrofinancial scenarios and key adverse scenario parameters
- Scenario horizon: 2022–2024, starting from end-2021; baseline aligned with IMF projections as of June 15, 2022.
- Adverse scenario narrative:
  - Low growth and high inflation globally, significant stress in global financial markets, de-anchoring of inflation expectations in the United States, faster policy rate hikes in advanced economies, widespread risk-off, currency depreciation, rise in sovereign and corporate spreads, system-wide liquidity stress, and negative macrofinancial feedbacks.
  - In the adverse scenario, real GDP level falls about 11 percent below baseline by 2023, equivalent to 2¼ standard deviation cumulative 2-year growth rate shock relative to the historical mean.
- Table 6 selected indicators (2021–2024):
  - Real GDP (2021=100):
    - Baseline: 100.0, 102.4, 104.6, 106.0
    - Adverse: 100.0, 97.5, 93.9, 98.3
  - Real GDP Growth Rate (percent, y-o-y):
    - Baseline: 4.8, 2.4, 2.2, 1.4
    - Adverse: 4.8, -2.5, -3.7, 4.6
  - CPI Inflation Rate (percent, y-o-y):
    - Baseline: 5.7, 7.2, 4.4, 3.3
    - Adverse: 5.7, 9.6, 8.7, 5.8
  - Exchange Rate (Mexican peso per U.S. dollar, end of period):
    - Baseline: 20.6, 21.4, 21.6, 21.9
    - Adverse: 20.6, 24.1, 26.1, 25.8
  - Policy Rate (percent, year averages except 2021):
    - Baseline: 5.5, 7.9, 9.1, 8.1
    - Adverse: 5.5, 9.4, 10.1, 6.9
  - 10-year Sovereign Bond Yield (percent, year averages except 2021):
    - Baseline: 8.0, 9.4, 9.4, 9.4
    - Adverse: 8.0, 10.9, 11.2, 10.2
  - Equity prices (2021=100):
    - Baseline: 100.0, 106.7, 113.3, 119.7
    - Adverse: 100.0, 90.3, 78.1, 76.0
  - Housing prices (2021=100):
    - Baseline: 100.0, 106.7, 113.3, 119.7
    - Adverse: 100.0, 82.6, 82.6, 105.1
  - Price of Commodities (energy/oil, 2016 = 100):
    - Baseline: 184.4, 346.5, 262.4, 224.1
    - Adverse: 184.4, 443.9, 258.8, 191.9
- Long term yields assumed sustained at 11 percent on average during 2022-23 in the adverse narrative.

### Solvency stress tests: scope, sample, and core assumptions
- Coverage and sample:
  - Top-down solvency stress test used IMF internal models covering credit, market, net interest income, and non-interest income risks.
  - Ten largest commercial banks selected, capturing more than 80 percent of total commercial banking system assets; sample included all six D-SIBs.
  - Exercise conducted at highest consolidation level with a three-year projection horizon; cut-off date end-2021 using Dec 2021 regulatory reports.
  - Separate partial analysis by CNBV on development banks and largest non-deposit taking credit providers (NBFIs) using CNBV internal models; results not directly comparable.
- Technical infrastructure and approach:
  - Static balance sheet assumption: asset allocation and funding composition constant throughout the stress horizon (end-2021 levels); total credit exposures kept constant (sum of performing and nonperforming).
  - No increases in capital or managerial actions assumed; asset and liability compositions static for net interest income calculation.
- Hurdle rates and capital treatment:
  - Hurdle rates set to 4.5, 6, and 8 percent for CET1, T1 and total capital requirements respectively under both scenarios.
  - Baseline included an additional 2.5 percent Capital Conservation Buffer (CCB) in the minimum requirement.
  - Bank-specific D-SIB capital surcharge (as of 2023) included in hurdle rates for both scenarios.
  - Banxico staff projections on eligible capital instruments used for CET1, AT1, T2 for each year.
  - Note: D-SIB capital surcharge is not a buffer in Mexico and should be always met.
- RWAs, PDs, and provisioning:
  - End-2021 Basel III RWAs used as baseline; market and operational RWAs kept constant; credit risk RWAs allowed to change to reflect credit quality deterioration.
  - TTC PDs adjusted in Basel RWA formula for IRB approach; RWA densities for performing and non-performing exposures kept at cut-off date levels for STA portfolios.
  - A smoothing multiplier on the PiT PDs delta was used to account for the TTC nature of the impact.
  - IFRS9 introduced in Mexico in January 2022; historical stage transition rates not available for the exercise; regulatory mandated provisioning framework used as working assumption; regulatory expected loss approach used where applicable.
- Income, taxes, and dividends:
  - Total net income projections included net interest income, non-interest income and expenses, market risk revaluation of securities, credit loss provisions, and tax charges.
  - Effective tax rates during the cut-off year used as proxy for future applicable tax rates.
  - Dividend payout ratio assumptions: 40 percent for year 1 and 50 percent for years 2 and 3; zero-dividend payout ratio for loss-making years.
  - Rationale: increased payout ratio for years 2 and 3 reflects intention of banks to increase dividends as COVID19-linked dividend restrictions are lifted and given current material capital buffers.
- Operational risk:
  - Operational risk capital requirements were not stressed and were kept at cut-off year levels following standard FSAP practice.

*Sources: Banxico; CNBV; and IMF staff calculation.*

### 36.     A variety of approaches for modelling credit parameters was used under the baseline

### 1mexea2022007 - 36.     A variety of approaches for modelling credit parameters was used under the baseline

### Credit-parameter modelling approach
- Exposures allocated to asset types via an exposure segmentation scheme: commercial exposures, exposures to government and to financial institutions, retail mortgages, and retail consumer exposures. Segmentation mandated to map segments to available time series of credit parameters.
- Table 7 segmentation for a single geography (domestic) and two regulatory approaches (IRB and STA) includes five exposure classes: Commercial, Government, Financial Institutions, Retail/Mortgages, Retail/Consumer.
- For projections of PDs:
  - Satellite models project scenario-dependent forward paths.
  - For three most material segments (commercial, retail mortgages, consumer), models used historical point-in-time (PiT) default rates on a monthly basis back to 2008.
  - Bayesian Model Averaging (BMA) technique employed: BMA Panel FE regressions, pooling equations per dependent variable, assigning weights by predictive performance to yield a “posterior model” equation.
  - Further econometric details and calibration referenced in Appendix IV.
- For proxied segments:
  - Distance-to-default transformation used to obtain PD paths based on bank PD paths for the corporate (commercial) segment due to lack of historical time series and small materiality.
  - Scenario-dependent PD PiT paths used to generate default flows from initial segment exposures (EADs at the cut-off date).
  - Assumed non-performing state is absorbing/terminal (cure rates captured in PD PiT model).
  - Resulting PD paths used to drive NPL ratios by bank and portfolio segment (illustrated in Figure 48 of Appendix IV).
- LGD PiT paths:
  - Estimated at bank and portfolio level using Frye-Jacobs approximation.
  - Initial anchoring at the level implied by the starting point; implied default rates LGD PiTs projected for each bank, portfolio segment and regulatory approach.
  - LGD PiT paths used to model LGD evolution over the 3-year horizon.
  - Data constraints (collateral information) prevented granular repriced/stressed-collateral simulations; Frye-Jacobs approximation used instead.
  - Expert judgment noted as relevant for anchoring LGD paths to stress scenarios and structural characteristics of each exposure segment.

### Market risk analysis
- Valuation risks captured: changes in risk-free interest rates and credit spreads for interest-sensitive instruments, equity, commodity, and mutual fund exposures.
- Portfolios collected by maturity bucket and credit quality class (where available) for sovereign and corporate debt, equity, mutual fund and commodity portfolios to estimate P&L and OCI impact under scenarios.
- Interest-rate-related valuation losses derived using a modified duration approach:
  - Captured re-pricing losses from shocks to sovereign yield curves.
  - Accounted for valuation impact due to shocks to spreads of corporate debt securities.
  - Spread projections for corporate, bank and financial bonds proxied using average yield per maturity tenor from Bloomberg and anchored to scenario macrofinancial conditions.
  - Corporate spread path in adverse scenario estimated to peak at 650 bps in 2024.
  - Baseline corporate spread path kept between 300 and 350 bps on average.
- Scope and conservatism:
  - Analysis covered interest rate and spread risks on sovereign and corporate debt in all accounting portfolios.
  - Existing (and future) hedges assumed to be ineffective during scenario horizon for conservatism.
  - Counterparty credit risk and credit valuation adjustment (CVA) risk excluded due to data limitations.
- Equity and commodity price risks:
  - Accounted using scenario shocks.
  - All equity positions assumed domestic due to lack of country breakdown.
  - Commodity shocks aligned with oil shock in scenario.
  - Banking sector exposure to equity and commodity risks relatively limited versus interest rate and spread risks.
- FX net open position and inflation risks ignored due to low materiality for Mexican banks.

### Net Interest Income (NII) analysis
- Earnings-based measure approach used to project NII under both scenarios.
- Econometric approach to calculate effective interest rates for stock of each asset and liability segment due to lack of historical data for effective rates on new business and repricing ladder; details in Appendix IV.
- Asset segmentation: cash equivalents and interbank loans, investments in securities, commercial, mortgage and consumer loans, derivative assets, other assets, non-interest-bearing assets.
- Liability segmentation: sight and term retail deposits, wholesale deposits, issuance (variable and fixed rate bonds), derivative liabilities, other interest-bearing liabilities.
- Regulatory reports used to establish outstanding volumes by segment at cut-off date; econometric outputs applied to project NII deltas from cut-off.
- Pass-through conservatism under adverse scenario:
  - Floor constraint on pass-through rate for retail segments: delta in sight and term retail deposit rates floored at 35 and 70 percent of delta in wholesale funding rates respectively (a minimum pass-through of 35 and 70 percent percent).
- Partial recognition of income from non-performing exposures:
  - Baseline: 50 percent of the income is recognized for all banks.
  - Adverse scenario: 25 percent recognized.
  - Rationale: balance across diverse business models and avoid over-penalizing banks concentrated in high-PD consumer products that rely on new origination for income.

### Non-interest income analysis
- Non-interest income contribution:
  - Total non-interest income during 2021 stood at about 21 percent of total income for banks in the stress test sample.
- Treatment:
  - Non-interest income stressed only under the adverse scenario using a constrained approach; baseline assumed constant at cut-off year income.
- Constrained approach mechanics:
  - Historical mean and standard deviation of each revenue stream computed; income normalized by bank total assets.
  - For each scenario year, revenue stream projection produced as a multiple of standard deviation away from historical mean—streams with higher variability penalized more.
  - Additional constraints (floors and caps) may be imposed; example constraints include overall non-interest income not exceeding cut-off year level or a source not exceeding its cut-off year income reduced by a factor.
  - Specific multipliers used in core solvency stress test: a multiple of 1 standard deviation for year 1, 0.25 for year 2, and 0 for year 3 of the adverse scenario; 0 deviations for all years under the baseline.
- Non-interest expenses kept constant for all years in both scenarios.
- Net Trading Income (NTI) due to intermediation not modelled and kept constant for both scenarios; FV adjustments for FVTPL portfolio reported to NTI.
- Administrative and other operating expenses kept constant at 2021 levels.

### IMF Top-Down Stress Tests: Main results and attribution
- System-wide resilience:
  - Banking system remains broadly resilient in terms of capitalization under the adverse scenario; most banks have ample capital buffers relative to hurdles.
  - Aggregate capital shortfalls are relatively small (less than 0.4 percent of GDP) under the adverse scenario.
- Aggregate capital adequacy ratio (CAR) dynamics:
  - Aggregate CAR declines by 4.7 percentage points to 14.5 percent by 2022 under adverse scenario (comfortably above minimum hurdle rates).
  - In the baseline, aggregate CAR increases by 2.8 percentage points to 22 percent by 2024.
  - In the adverse scenario, aggregate CAR would decline by 4.1 percentage points to 15.1 percent in 2023.
  - The decline in 2023 is larger by almost 5.3 percentage points when compared with the baseline path in 2023.
- Drivers of capital depletion:
  - Two key drivers: material market losses from pronounced rise in interest rates and credit spreads, and increase in loan loss provisions due to credit deterioration.
  - Impact of NII is moderate due to solid net interest margins.
  - Static balance sheet assumption limits new loan origination and internal capital generation; securities exposures assumed constant.
- Credit losses and NII impacts:
  - Cumulative credit losses up to year 2: 6.1 percentage points of capital under the adverse scenario versus a 2.8 percentage point decline in the baseline scenario.
  - Net interest income in the adverse scenario lower on average by 1.3 percentage points of capital for the first two years.
  - In the adverse scenario, NII is reduced by approximately 14.2 percent during the first two years relative to the cut-off year.
    - Almost one eighth of that impact driven by interest rate shocks (more pronounced during year 1); remaining decrease attributed to forgone interest from non-performing exposures.
- Market risk contributions:
  - Cumulative NTI and OCI losses: 2.2 percentage points of capital under the baseline and 3.6 percentage points under the adverse scenario; main drivers of negative slope in 2022 capital path.
- Mitigants and heterogeneity:
  - Negative capital impact partially mitigated by reduced dividend distributions (loss-making years assumed no dividends) and tax outcome differences due to projected losses.
  - Aggregate leverage ratio remains well above regulatory minimum under both scenarios.
  - Substantial heterogeneity across banks: business model and portfolio concentration (e.g., consumer-focused banks) can lead to larger capital depletion; holdings of sovereign debt or weaker starting capital positions may also increase vulnerability, especially for smaller and less diversified banks.
- Income recognition assumptions and effects:
  - Exercise assumes 25 percent income recognition for non-performing exposures under adverse scenario and 50 percent under baseline.
  - Net interest income impact includes funding/lending repricing component and forgone interest from NPEs; pass-through on stable deposits can allow banks to increase net interest margins, more pronounced with larger deposit participation in funding mix.

*Source: IMF staff calculations, Mexico FSAP chapter.*

### 59.     The impact of market risk is very material in the overall adverse scenario result. As

### 1mexea2022007 - 59.     The impact of market risk is very material in the overall adverse scenario result. As

### Market risk and revaluation effects
- Revaluation of debt securities drives the market risk impact: "3.6 percent of RWAs during year 1 of the baseline and almost 6 percent under the adverse."
- Losses mainly concentrated during year 1, but cumulative revaluation (FV and OCI) remains important under both scenarios.
- Impact almost entirely driven by bond securities portfolios; equities contribute only a small amount.
- Contribution by accounting portfolios: "FVTP portfolio is almost double the size of the impact."
- Footnote scenario: analysis of AC/HtM portfolios illustrated relative impacts though "the exercise did not assume any losses coming from the AC/HtM portfolios, since such losses are accounted for under credit risk."

### Non-interest income and NII impacts
- Applied stress on non-interest income: "approximately 9.2 percent (as an average across all three years)."
- Contribution to capital depletion from non-interest income is modest: "0.80 percent in terms of starting RWAs versus the baseline scenario respective result during the first two years."
- Net interest income (NII): "the impact of forgone interest due to NPL formation is the major driver"; positive interest rate shocks tend to favor profitability because pass-through rates on retail deposits are small relative to benefits from positive term spread shocks.

### Business model dispersion and RWA density
- RWA densities in the exercise sample range from "30 to 90 percent" reflecting diverse business models, exposure to sovereign risk, and loan book concentrations.
- Observation supports use of bank-specific capital measures (Pillar 2) given uneven features across the 50 commercial entities in the banking ecosystem.

### Concentration risk and large exposures
- Stylized analysis suggests a relatively high percentage of banks would face material capital loss if their largest counterparty defaults (under a conservative assumption of "an LGD of 100 percent" and being agnostic on counterparty type).
- Recommendation: closely monitor exposure concentration risks and ensure prudential measures to mitigate impact.

### Sensitivity tests — contingent credit lines
- Contingent credit lines to NFCs and financial entities at end-2021: "1.6 trillion pesos" — almost "15 percent of total banking sector’s assets" or "31 percent of the total loan book."
- These lines are unevenly distributed across banks and are mostly revocable; CCFs are set to zero for revocable lines (no Pillar I capital charges).
- Central sensitivity scenario assumed "25 and 30 percent of the outstanding lines would be triggered through the three years of the adverse scenario."
- System-level impact: triggering such lines would make "landing capital ratio by 2024 almost 2 percentage points lower than the one observed under the normal adverse scenario."
- Prudential implication: concentrations merit closer monitoring and bank-specific Pillar II measures.

### Sensitivity tests — higher-than-anticipated interest rate shock
- Additional shock: parallel shift of "+150 bps" on top of adverse scenario.
- Capitalization impact: sample banks’ capitalization would decline by "2.4 percentage points by 2022" or "3.1 percentage points by the end of the horizon."
- Losses mainly from FV and OCI portfolios; credit losses increase but much can be mitigated by increased NII due to low pass-through on retail deposits.
- Concern: dispersion of capital impact across banks — evidence of sovereign exposure concentration; suggests adjusting Pillar I and II frameworks for business model/concentration risks.

### Sensitivity tests — digital payments / new forms of digital money
- Hypothetical stress: erosion of NII and payment-service income from competition (CBDC or private digital money).
- Most adverse private-money scenario: system total capital depletion of "34 basis points by 2024" relative to baseline solvency stress test results.
- Bank-level impacts: some banks could see capital ratios decline "up to 75 basis points by 2024."
- CBDC scenario with cap of "UDI 3000 (approximately 21,000 Mexican pesos)" limits impact: capital adequacy ratio drop of "25 basis points."
- Calibration notes: in private-money scenario cost of funding for sight deposits below 1 million Mexican pesos increases by "50 bps"; credit card fee income falls by "20 percent" at end of projection with a shock dynamic of "30 percent impact in the first year and a 70 percent in the second year."

### Policy recommendations (solvency-related)
- Ensure vulnerable banks with low capital levels or elevated business model risks implement capital raising plans without delay and reflect idiosyncratic risks in minimum capital levels.
- Apply bank-specific capital add-ons to address large credit exposure concentrations and business-model concentration; rely on an efficient Supervisory Review Process and appropriately calibrated Pillar 2 requirements.
- Closely monitor contingent credit lines; ensure exposures are identified, assessed, and mitigated; consider supervisory measures to avoid wrong incentives and uneven stress impacts across banks.
- Enhance regulatory reporting granularity: dynamically link exposures with updated LTVs, address weaknesses in collateral valuation relative to housing price fluctuations, and increase focus on IFRS9 and IRB metrics.
- Maintain adequate resources and organizational structure for stress testing given its complexity and resource intensity.

### Liquidity stress tests — overview and key findings
- Tests performed: Basel III LCR test over 30 days (by aggregate currency and significant currency), cash flow-based liquidity stress test over three months, and NSFR dynamics analysis.
- Regulatory ratios and aggregated positions by December 2021:
  - Aggregate LCR equal to "225 percent."
  - No bank had an LCR lower than the regulatory minimum of "100 percent."
  - System aggregate NSFR stood at "145 percent."
- Vulnerabilities identified:
  - Material contingent credit lines to corporates that could be withdrawn quickly or simultaneously, reducing liquidity buffers.
  - Part of retail deposits from high net-worth individuals may behave like wholesale deposits (e.g., through synthetic-repo products) and be prone to outflows despite current classification as stable deposits.
- LCR-based stress tests and three-month cash flow analysis showed some medium- and small-sized banks breached the "100 percent" LCR threshold in certain scenarios.
- Recommendation: authorities should mitigate risks from contingent credit lines and reassess deposit classification/behavior assumptions in liquidity frameworks.

*Source: IMF staff estimates and calculations from the referenced solvency and liquidity stress test chapter.*

### 81.     The LCR is designed to ensure that banks hold a sufficient reserve of HQLA to allow

### 1mexea2022007 - 81.     The LCR is designed to ensure that banks hold a sufficient reserve of HQLA to allow

### LCR design and objective
- The LCR requires that in normal times banks hold a stock of cash or unencumbered HQLA at least as large as the expected total net cash outflows over a period of significant liquidity stress lasting 30 calendar days.
- Purpose: promote short-term resilience of banks’ liquidity profile so that by converting HQLA into funding in private markets, banks can absorb shocks and reduce spillovers to the real economy.

### Evolution of Mexico’s LCR around the COVID-19 pandemic
- Average LCR for the banking system: 165 percent in January 2020 (pre-pandemic).
- By December 2021, aggregate LCR: 225 percent.
- As of December 2021, no bank in the system had an LCR below the 100 percent regulatory minimum.
- Pandemic dynamics:
  - Temporary reduction of liquidity during March-May 2020 linked to tapping of committed credit and liquidity lines held by corporates.
  - Corporate and retail customers triggered credit lines for precautionary purposes and placed them as deposits with banks, partially containing the drop.
  - Subsequent increase in HQLA was fostered by increased deposits from corporates and households and substantially reduced credit origination activity.

### Credit lines and liquidity vulnerability
- Undrawn credit lines (May 2022):
  - Revocable credit lines: 2.5 trillion pesos.
  - Irrevocable credit lines: 0.25 trillion pesos.
  - Liquidity lines: 0.15 trillion pesos.
  - Of the revocable credit lines: 1.2 trillion pesos to individuals, 1.2 trillion pesos to corporates, 0.3 trillion pesos to other financial entities.
- Risk: contemporaneous triggering of credit lines would expand banks’ assets and require parallel expansion of liabilities, stretching system liquidity—particularly if triggering is precautionary and replaces external funding during global tightening.
- Triggering of contingent credit lines represents a particular vulnerability for larger banks (D-SIBs), while smaller banks more often breach thresholds due to lower starting LCRs and greater exposure to wholesale outflows.

### Retail deposits classification and monitoring
- Some deposits classified as “retail” may behave more like “wholesale,” especially from high net-worth individuals seeking higher interest rates.
- Large difference between repo and reverse repo outstanding volumes may indicate material synthetic positions and yield pick-up products for high net-worth depositors—warrants monitoring.
- Classifying high net-worth deposits as retail could overestimate their stability and bias liquidity metrics positively.

### LCR stress-testing methodology (FSAP top-down test)
- Scope: all 50 commercial banks.
- Severity increases achieved by:
  - Increasing haircuts on banks’ counterbalancing capacity (HQLA used to obtain liquidity).
  - Increasing run-off rates applied to outflows.
- Combined scenarios: 3 HQLA haircut scenarios × 4 increased outflow scenarios = 12 combined scenarios.
  - Haircut scenario 1 = regulatory haircuts (aligned to Basel).
  - Haircut scenarios 2 and 3 = increasingly severe haircuts.
  - Outflow scenario 1 = regulatory run-off rates (aligned to Basel).
  - Outflow scenarios 2, 3, 4 = increased run-off rates on retail funding (O2), wholesale funding (O3), and both (O4).

### LCR stress-test results (system and banks)
- In all scenarios, average LCR for the banking sector remained above 100 percent, though some banks breached the threshold under some scenarios.
- Table 9: Bank Liquidity Stress Test Results (system-level)
  - Regulatory Scenario (1)
    - Aggregate LCR (In percent): 225
    - No. Banks with LCR<100: 0
  - Retail Shock (2)
    - Aggregate LCR (In percent): 157
    - No. Banks with LCR<100: 6
  - Wholesale Shock (3)
    - Aggregate LCR (In percent): 128
    - No. Banks with LCR<100: 13
- Drivers of breaches:
  - Smaller banks: lower starting LCRs, business models, exposure to wholesale outflows.
  - Larger banks: triggering of contingent credit lines is material vulnerability; D-SIBs started from an asset-weighted LCR of 245 percent and only breached threshold in the most severe scenario.

### Complementarity with Banxico’s Liquidity-at-Risk analysis
- Banxico conducts LCR-based stress tests with different estimation of haircuts and run-off rates:
  - Banxico uses solvency stress test scenarios and constructs stressed outflows using 95 percent Conditional Value at Risk (CVaR) of historic outflows back to 2006.
- Results were comparable overall, but FSAP identified more banks potentially breaching 100 percent because FSAP assumed more severe run-off rates for retail deposits and contingent credit lines.

### FX liquidity regulation and FX LCR results
- Mexico does not set regulatory minimum limits for LCR by single currency in line with Basel III, but has foreign currency liquidity regulation (introduced mid-1990s) consisting of three limits:
  - Limits to net open position: banks’ net open position in foreign currency limited to 15 percent of Tier-1 capital (including peso denominated products linked to the exchange rate).
  - Short-term liquidity requirement: banks must hold enough liquid assets to cover the sum of the largest gaps (liabilities minus assets) for up to 60 days and a percentage of all other liabilities up to 60 days not covered by same- or shorter-maturity assets.
  - Structural liquidity requirement: at the end of each day, no commercial bank can have Net Foreign Currency Liabilities greater than 1.83 times its core capital (CET1).
- FSAP LCR-based stress tests in domestic currency and USD (as of end-December 2021):
  - LCR in national currency: 182 percent.
  - LCR in U.S. dollars: 348 percent.
  - 33 out of 44 banks with material U.S. dollar operations had LCR above 100 percent.
  - Non-compliance drivers: LCR regulation limits recognition of inflows and does not consider deposits held by banks in other financial entities as eligible HQLA; relaxing LCR cap on inflows results in U.S. dollar LCRs above minimum threshold.

### NSFR implementation and levels
- NSFR aims to reduce funding risk over a longer horizon by requiring sufficiently stable funding sources.
- NSFR reporting has been in place since 2017; NSFR became binding for Mexican banks from March 2022.
- End-2021 average NSFR: 145 percent (up from end-2019: 128 percent), driven mainly by seven largest banks and increases in retail deposits and capital; corporate deposits also grew but were partially offset by decreases in government and financial institution deposits and securities issuance.

### Cash-flow analysis: approach and assumptions
- Horizon: three months (contractual maturity ladder for assets and liabilities provided by authorities).
- Four-step approach:
  1. Estimate banks’ maturing liabilities by funding segment over next three months.
  2. Apply progressive stress scenarios to liabilities to establish run-offs and create funding needs.
  3. Quantify counterbalancing capacity after haircuts to liquid assets.
  4. Compare funding needs (including contingencies) and counterbalancing capacity to estimate surpluses and shortfalls at bank and system level.
- Key assumptions:
  - Full collateral revaluation and triggering of committed credit lines (collateral release and additional pledging due to run-offs applied to counterbalancing capacity).
  - Contingent credit lines held by corporates with commercial banks were partially triggered.
  - Recognition of inflows from maturing loans decreases with scenario severity; roll-over rates of maturing retail and corporate loans progressively reach 100 percent in severe scenarios (i.e., banks are not allowed to counterbalance outflows by extending new credit).

### Cash-flow analysis findings
- Under mildly adverse scenarios, all banks could handle funding shortfalls with existing counterbalancing capacity.
- Under the most severe adverse scenario:
  - 21 banks (representing almost 70 percent of total banking sector assets) would encounter liquidity shortfalls.
  - Combined liquidity shortfall at bank-level: 0.3 trillion pesos (4 percent of total assets).
  - Aggregate (net) shortfall after netting surpluses and shortfalls: around 0.1 trillion pesos (1 percent of total assets).
  - Assessment: shortfalls appear manageable given Banxico’s ability to support the system via standard facilities or extraordinary measures.
- Comparison with LCR tests:
  - LCR corresponds to an intermediate severity between cash-flow mild and adverse scenarios.
  - Banks with liquidity shortfalls tend to be the same across LCR and cash-flow analyses, but exposures differ by balance-sheet structure:
    - Investment banks: particularly vulnerable to wholesale deposit outflows and loss of counterbalancing capacity.
    - Large and small banks: more vulnerable to triggering of credit lines.

### Distance to Liquidity Stress Indicator (DLSI)
- DLSI measures the stress factor required to bring a bank to liquidity shortfall (reverse stress-testing metric).
- FSAP specification anchors:
  - Mild scenario = stress factor of 0.25.
  - Severely adverse scenario = stress factor of 1.
- Interpretation:
  - DLSI below 1: bank would face liquidity shortfall at a stress level below the severely adverse scenario.
  - DLSI above 1: bank can withstand the severely adverse scenario and would require more severe stress to reach shortfall.
- Mexico results:
  - Half of the banking sector (in terms of asset size) has a DLSI value higher than 1.
  - For the other half, even in the most severe adverse scenario (stress factor = 1) liquidity shortfalls are small and manageable.

*Source: IMF staff analysis based on Banxico data as presented in the provided content (1mexea2022007).*

### 99.     The authorities should continue to monitor the dynamics of contingent credit lines and

### 1mexea2022007 - 99.     The authorities should continue to monitor the dynamics of contingent credit lines and

### Liquidity risks from contingent credit lines (paras 99–101)
- Finding: Banks’ credit lines to corporates were material at 1.2 trillion pesos on average between January 2020 and May 2022.
- Event: In March 2020 part of these lines was withdrawn for precautionary reasons and were deposited with banks; the drop in liquidity for the system was generally contained.
- Observation: Only irrevocable and liquid credit lines were withdrawn (while revocable credit lines, that represent 90 percent of the total, were not).
- Risk characterization:
  - The impact on bank liquidity from withdrawal of credit lines could be sudden and significant.
  - During a crisis, revoking credit lines may generate large reputational costs for banks and amplify spillovers under tighter liquidity conditions.
- FSAP recommendation:
  - The authorities should continue to monitor the dynamics of contingent credit lines and assess with relevant supervisors how related risks are managed.
  - Design a framework of incentives (or capital charges) that would lead to a reduction of their outstanding amount.
- Additional supervisory tool:
  - Incorporate liquidity analysis and quantifiable metrics in the Supervisory Review Process (SRP) to inform Pillar 2 capital requirements, including for development banks.
  - Use liquidity positions of development banks to assess capital surcharges needed to reflect risk-taking or contribution to systemic risks.

### Retail deposits behavior and LCR robustness (para 101)
- Issue: Retail deposits of high net-worth individuals could behave as wholesale deposits in their search for yield.
- Recommendation:
  - Monitor evolution of such deposits (including further analyzing and monitoring the difference between repo and reverse repo transactions).
  - Conduct sensitivity analyses on the LCR to complement the standard LCR, providing a more realistic angle on banks’ liquidity positions.

### Maturity ladder and cash flow analysis (para 102)
- Progress: Authorities started to collect regulatory data allowing a maturity ladder template to produce cash flow analysis since March 2022.
- Recommendation:
  - Make production of the maturity ladder template more systematic.
  - Complement liquidity metrics with a full cash flow analysis-based liquidity indicator to efficiently supplement LCR-based stress tests and highlight differences in banks’ exposure to various sources of stress.

### Cross-border bank linkages and contagion (paras 103–106)
- Key concentrations (2021Q4):
  - Total foreign claims of Mexican banks: US$37 billion as of 2021Q4.
  - The United States comprises around 80 percent of total foreign claims.
  - Claims of foreign banks in Mexico: US$130 billion in 2021Q4.
  - The United States and Spain cover 55 percent of total claims on Mexico.
- Network analysis setup:
  - Covered 13 countries.
  - Credit shock LGD set to 0.5.
  - Funding shock assumption: borrower banks unable to replace 10 percent of funds previously granted by the creditor country in default, leading to a fire-sale; borrower bank’s assets sold at a 30 percent discount.
- Simulation results:
  - Inward and outward spillovers to and from Mexico are relatively modest.
  - A hypothetical financial distress in Mexico would impact banks in Spain the most through both channels (loss of approximately 5 percent of bank capital).
  - A hypothetical financial distress in the United States would have the most significant impact on banks in Mexico (loss of 30 percent of bank capital).

### Domestic financial system interconnectedness (paras 107–110)
- Data and scope:
  - Supervisory data included bilateral exposures among 42 commercial banks, 6 development banks, 25 brokerage firms, 4 credit unions, 6 insurance companies, 34 investment funds, 6 Sofomes, 1 nonbank financial institution, and 8 foreign countries, totaling 132 entities.
  - Data are for December 9, 2021.
- Balance-sheet role:
  - Development banks are the largest net borrowers from the system; commercial banks are key providers of funds.
  - Brokerage firms are among main creditors; credit unions, Sofomes and other NBFIs are borrowers; investment funds and insurance companies have bilateral asset and liability exposures but limited risk due to small exposures.
- Network shock assumptions:
  - Financial entity fails when loss of capital from shocks exceeds initial total regulatory capital (capital data collected for commercial banks, development banks, and brokerage firms only; other entities not assumed to act as contagion channels).
  - LGD assumptions for combined credit and funding shock:
    - Unsecured bilateral exposures LGD = 0.7.
    - Secured bilateral exposures LGD = 0.2.
    - Bilateral exposures through securities’ holdings LGD = 0.5.
  - Funding replacement in joint shock: a borrower can replace 70 percent of funds previously granted by the default creditor, causing assets to be traded at a 15 percent discount fire-sale.
- Simulation results and indices:
  - Total domestic bilateral exposures are only half of the size of the total regulatory capital for commercial banks, development banks and brokerage firms.
  - The failure of a domestic financial entity barely triggers failures of other entities, except for a few institutions.
  - Contagion:
    - Default of a development bank would cause an average loss of 1.8 percent of its counterparties’ capital.
    - Default of a commercial bank would cause an average loss of 0.2 percent of its counterparties’ capital.
  - Overall vulnerability index remains relatively low, indicating individual entities’ capital remain resilient during triggered failures.

### Corporate risk analysis — overview and data limitations (paras 111–119)
- Importance: Corporate sector health is pivotal for financial stability; deterioration can presage rising NPLs and deteriorating bank asset quality.
- Large publicly traded corporates:
  - Generally healthy and recently reversing deterioration from the previous decade.
  - A sample of 81 Mexican firms had among the lowest baseline and shock default probability levels in a sample of 24 advanced and emerging economies in a global corporate stress test.
  - Interest coverage ratios fell and leverage rose over the last decade and early pandemic, but trends recently reversed in early post-pandemic recovery and require continued monitoring.
- Coverage gaps:
  - The 84 corporates in the Tressel and Ding study represent 43 percent of total NFC debt, leaving a substantial fraction of financial sector exposure unobserved.
  - Databases covering a wider range of firms often have limited indicators, missing data, or data quality issues.
- Use of machine learning:
  - Penalized regressions and forest-based methods reduce overfitting and handle noisy, collinear, and skewed data.
  - Practices: imputing missing data, cross-fold validation, and out-of-sample testing to improve inference for firms outside high-quality samples.
- Firm microdata and EDFs:
  - Firm balance sheet data from Capital IQ covering more than 14,000 observations of firms from Brazil, Chile, Colombia, Mexico, and Peru; database covered about 860 firms in a typical recent year, of which about 109 are Mexican.
  - Expected default frequencies (EDFs) from Moody’s used as dependent variable; typically about 48 EDFs in Mexico of which 38 matched with Capital IQ firm data.
- Table of data sources and counts (exact entries preserved):
  - CIQ Microdata: Firms 863, Date Range ‘00-’21, Total Obs. 14,523, Mex Firms 109, Mex Obs 1,936.
  - EDFs: Firms 239, Date Range ‘99-’21, Total Obs. 5,352, Mex Firms 48, Mex Obs 1,142.
  - CIQ Microdata + EDFs: Firms 170, Date Range ‘00-’21, Total Obs. 3,481, Mex Firms 38, Mex Obs 799.
  - Orbis: Firms 22,159, Date Range ‘12-’20, Total Obs. 188,308, Mex Firms 420, Mex Obs 3,242.
  - Expansion 500: Firms 500, Date Range ‘10-’20, Mex Firms 313, Mex Obs 3,443.
- Country-sample considerations:
  - Including observations from other major Latin American emerging markets may improve estimation by increasing observations for macro variables but could be misleading due to institutional and structural differences; estimations with and without non-Mexican observations were evaluated.
- Variables used in estimation:
  - Firm-specific variables: 4 liquidity measures, two measures of leverage, 3 measures of earnings, and a measure of firm size relative to the average firm in a country-year.
  - Macroeconomic variables to capture real, external, and financial sector stresses.

*Source: Content unit 1mexea2022007.*

### 120.     Estimation methods

### Estimation methods

### Methodology and cross-validation
- Observations were divided into subgroups or ‘folds.’ Each subgroup was sequentially removed from the sample and the remaining observations were used to estimate a model; estimation was then evaluated on the hold-out subgroup.
- Estimation methods and specifications were evaluated by the simple average of the R2 on each of the subgroups.
- For models with hyperparameters, these hyperparameters were selected from a coarse grid in a nested cross-validation exercise within the non-hold-out sample.
- Cross-validation folds were constructed randomly and also grouped by country, industry (SIC 1-digit), time period (5 groups of contiguous years), and firm size (grouped by tercile) to assess performance on observations dissimilar to the estimation sample.
- EDFs were transformed by logit and all explanatory variables were normalized to mean zero and unit standard deviation.

### Models evaluated and data preparation
- Models estimated:
  - Ordinary least squares regression (OLS)
  - Penalized regressions (elastic net)
  - Decision trees
  - K-Nearest Neighbors
  - Random Forest regressions
- Rationale: these methods are commonly used for continuous dependent variables on small datasets and are often more robust than simple linear regression; more flexible methods like neural networks typically require substantially larger datasets.
- Hyperparameter selection implemented via nested cross-validation within the non-hold-out sample.

### Explanatory variables (corporate sector analysis)
- Firm-specific variables included: Quick Ratio; Current Ratio; Cash Equivalent to Current Liabilities; Leverage (D/A); Leverage (TL/E); EBIT to Current Liabilities; ST Interest Rate; LT Interest Rate; Interest Coverage Ratio; Return on Assets; Return on Equity; Revenue Growth; Leverage lags; Depreciation and Depreciation lag; Interest Coverage Ratio; Firm Size; and others as listed in the source table.
- Macroeconomic variables included: GDP Growth; GDP Growth Lag; Financial Conditions Index; Financial Conditions Index lag; ST Interest Rate; LT Interest Rate; Depreciation; and others as listed in the source table.

### Model interpretation: Shapley values
- Model assessments were decomposed using Shapley values, defined as the average marginal contribution of an independent variable across all possible orderings of independent variables.
- Shapley values provide an additive decomposition satisfying criteria like symmetry and independence of irrelevant alternatives.
- Computation time grows with the number of variables but can be reduced for penalized regression and tree-based methods; advances in computing power further reduce costs.

### Model performance and key results
- Using cross-fold validation with random division into folds, average R2s on the Mexico sample varied from about 0.35 for linear regression to about 0.65 for Random Forest.
- Estimations on the Mexico-only sample outperformed estimations on the larger LA-5 dataset across all methods; however, ranking of methods remained broadly unchanged and variation across samples was smaller than variation across models.
- Cross-validation by type (country, firm size, industry, year) showed materially worse performance for all models; OLS produced badly negative average R2 in several grouping cases and is not presented in that panel.
- Random Forest outperformed other methods in all cross-validation grouping cases.
- The Mexico-only sample showed very poor results when observations are grouped by time period, attributed to a very small number of observations for macroeconomic variables.

### Model insights (Random Forest and non-linearities)
- The Random Forest model emphasized:
  - Interest rates
  - Interest coverage ratio
  - Firm size
  - Weaker profitability
  - Higher leverage ratios
- Shapley-value diagnostics:
  - For interest coverage ratio, importance arises from very high risks associated with very low levels.
  - For leverage measured as total liabilities over equity, risks vary more continuously across the sample.
  - Firm size shows a strong relationship with risk at the low end: risk contributions fall rapidly between 0 and 0.05, and beyond 0.05 additional size does not substantially reduce assessed risk.
  - Vertical dispersion in Shapley-value plots reflects interaction effects: the same value of a variable can associate with different risk contributions depending on other characteristics.

### Alternative datasets and out-of-sample implications
- The Orbis dataset captured nearly 1,000 firms in 2017 and tends to include firms that are smaller, more highly levered, and more profitable.
- Expansion 500 firms are selected for size but show higher leverage.
- Adding unmatched CIQ firms produces a smaller and more profitable group with slightly lower leverage.
- Average corporate risk in the estimation sample (debt-weighted average of corporate EDFs) fell from 5 percent in the early 2000s to below 0.3 percent in the post-GFC era.
- Bank-based corporate PDs varied between 2 and 5 percent post-GFC.
- Applying the Random Forest estimation to the larger datasets produced EDFs varying between 2 and 4 percent in the last decade, without a substantial decline from the very early 2010s—more aligned with bank PDs.

### Scenario simulation and stress results
- The model estimated on the full LA5 database was used for scenario simulation due to superior cross-validation performance across dissimilar time periods and macroeconomic conditions.
- Macroeconomic variables were updated to reflect baseline and stress scenario paths; firm data from 2021 was simulated to evolve over 2022–24 according to simple rules (earnings and revenue growth track nominal GDP variations; interest costs track a weighted average of short- and long-term interest rates; profits evolve as the difference between earnings and interest cost growth).
- Model-implied EDFs under the solvency stress scenario:
  - Risks rose even in the baseline due to rising interest rates and normalizing post-pandemic growth.
  - Risks in the stress scenario rose roughly tripling their baseline levels in 2023 and 2024.
  - The modeled rise is worse than the increase estimated in the stress scenario of slightly less than double for these years.

### System-wide liquidity analysis (overview and vulnerabilities)
- Mexico’s openness:
  - Total flow of export and import ranked top at 83 percent of GDP.
  - Stock of FDIs recorded at around 600 billion US dollar as of 2021.
  - The United States accounts for nearly 80 percent of Mexico’s exports and 50 percent of FDIs.
  - The Mexican peso is the third most actively traded currency in the Americas (after the United States dollar and Canadian dollar), and the most actively traded currency in Latin America.
- Domestic financial system linkages:
  - Market agents act as funding providers or receivers via direct lending, issuance of debt securities, and secured and unsecured interbank transactions.
  - Indirect exposures arise from holdings of common assets, mostly sovereign and corporate debt securities, exposing agents to repricing risks from shifts in risk-free rates and sovereign and corporate spreads.
  - The offshore market provides liquidity to the domestic system, largely through buying and holding domestic sovereign bonds and corporate stocks.
- Historical evidence:
  - Liquidity shocks can be strongly correlated under stress, amplifying downside risks.
  - March 2020 exhibited strong co-movements among reduction in sovereign securities by international investors, commercial bank deposit outflows, and liquidation by foreign investors of Mexican corporate stocks—demonstrating synchronized liquidity strains that can build tail risks.
- Novel FSAP analytical approach:
  - Integrates liquidity and interconnectedness analyses to trace funding flows across all market agents.
  - Combines domestic and cross-border networks to allow simultaneous domestic and external shocks.
  - Complements solvency-focused contagion analysis by targeting the liquidity layer and incorporating second-round behavioral effects like asset liquidation.
  - Enables macroprudential perspective and multiple sensitivity and counterfactual analyses (for example, imposing or relaxing regulatory binding constraints on liquidity) to inform policy decisions.

*Source: IMF staff estimates and analysis as presented in the supplied content.*

### 135.     The objective of the system-wide liquidity analysis is manifold. First, it is essential to

### 1mexea2022007 - 135.     The objective of the system-wide liquidity analysis is manifold. First, it is essential to

### Objectives of the system-wide liquidity analysis
- Understand the extent of interconnectedness among agents and have a system-wide view of liquidity conditions because the resilience of an individual sector or institution cannot itself assure the stability of the entire system.
- Assess the contribution of each agent to system-wide liquidity stress to improve understanding of transmission channels and amplification mechanisms associated with willingness and capacity to intermediate in the market.
- Assess resilience against various adverse narratives pertinent to the Mexican financial system.
- Serve as a diagnostic tool to inform policy discussions with measurable and quantifiable data, aiming at ensuring sufficient liquidity buffers in the system.
- The framework is tailored to Mexico’s specific risks and vulnerabilities but can serve as a proof of concept and be extended or generalized to other economies.

### Overview of agents and stylized balance sheets
- Commercial and development banks:
  - Both conduct maturity transformation by leveraging short-term funding to finance longer-term holdings of sovereign securities and loan portfolios.
  - Funding profiles differ: commercial banks rely mostly on retail and wholesale deposits; development banks obtain wholesale funding from investment funds and nonfinancial corporations via short term repo transactions and short-term bond issuance, with minimal exposure to direct deposits from the public.
- Investment funds:
  - Finance almost exclusively via issued fund shares.
  - Invest mostly in sovereign securities.
  - Provide financing to other financial agents via reverse repos.
  - Hold large amounts of cash and other assets (such as equity investments).
  - Considered more liquid than commercial and development banks.
- Brokerage firms:
  - Act as market makers or agents in securities trading and offer investment advice.
  - Obtain short term repo-financing while investing in securities and participating in reverse repo transactions.
  - Excluded from the analysis due to simpler balance sheets and relatively small size.

### Bilateral exposures, sovereign holdings, and risks
- Commercial banks:
  - Hold most claims against corporates, households, and the government in the form of loans and sovereign securities.
  - Accept wholesale and retail deposits mainly from corporates and households.
  - Have little financial obligations to other agents.
- Development banks:
  - Obtain wholesale funding mostly from investment funds and corporates via repos or issuance of securities (both short-term), introducing higher funding risks.
  - Often seek funding from commercial banks and other nonbank agents.
  - Use funding to extend loans to SMEs or invest in sovereign securities.
- Sovereign securities exposure:
  - Development banks and investment funds hold a higher share of sovereign securities than commercial banks, but all agents are exposed.
  - Risks: sudden increases in sovereign yields, decline in market value of unencumbered collaterals, triggering of margin calls on encumbered collaterals.
  - Levels of encumbrance are elevated for sovereign securities, potentially limiting capacity to utilize available liquid assets under stress.
  - Sensitivity of market repricing is moderate: duration of the bulk of sovereign securities is between one to five years, with only a small share having maturity beyond ten years.
  - Corporate securities have a notable share at maturity beyond 10 years but are not expected to prompt system-wide market losses and liquidity stress due to significantly lower market holdings.

### Contingent credit lines and off-balance sheet exposures
- Off-balance sheet exposure extended by commercial banks to corporates and other private sectors is "around 3 trillion pesos" including both revocable and irrevocable credit and liquidity lines.
- As of May 2022 (breakdown provided in source):
  - Revocable credit lines were 2.5 trillion pesos.
  - Irrevocable credit lines were 0.25 trillion pesos.
  - Liquidity lines were 0.15 trillion pesos.
- Such credit lines can be an important source of liquidity risk if triggered en masse and displaced outside the system during heightened risk aversion or tightened global liquidity/financial conditions.

### Scope, data, and coverage
- The analysis includes the central bank and the government, commercial banks, development banks, investment funds, non-financial corporations, households, and foreign investors providing external funding to the domestic financial system.
- These agents collectively represent about 64 percent of total financial sector assets.
- Data compiled by Banxico as of December 2021 at the highest consolidation level and at an aggregate balance-sheet level (by agent type), collected in a data template designed by IMF staff.
- Data elements collected include:
  - Agent-specific balance sheet composition and bilateral exposures (who-to-whom holdings): loans, debt securities, reverse repo (assets); deposits, issuance of debt securities and shares, repo financing (liabilities).
  - Collateral data: encumbered and unencumbered, split into central bank eligible and non-eligible, by issuer type and remaining maturities.
  - Margin positions covered with debt securities under derivative transactions.
  - Haircut information in repo transactions by maturity buckets split into central bank operations and secondary market transactions.
- Due to data limitations, pension funds and insurance companies are not included in the analysis and account for the majority of the remaining assets of the system.

### Workflow and methodology
- Three distinct steps:
  1. Narrative design: formulate four narratives (designed for Mexico) each simulating a unique liquidity stress event.
  2. Shock generation: generate a series of liquidity shocks impacting market agents’ balance sheets according to each narrative; many scenario shocks are defined with values for each layer.
  3. Monte Carlo simulation: quantify the net liquidity position for each market agent after each simulation to capture direct funding and market stress impacts plus second-round revaluation effects (calls on encumbered collateral for funding or margin positions).
- Novel feature: shocks are drawn from correlated distributions specified by a copula (multivariate distribution with pre-defined correlation factors and marginal distribution shapes and bounds).
  - Correlation factors can be adjusted to match historical stresses or desired correlation levels between shock pairs.
  - Sensitivity analysis recommended where historical data are limited.

### Policy-use and simulation capabilities
- Framework can inform liquidity-relevant policy decisions:
  - Example: Expanded access of investment funds to the repo market could enhance fund liquidity positions and improve systemic resiliency by relieving supply pressures in the sovereign bond market without withdrawing residual liquidity from the financial system.
  - Framework can impose or relax regulatory constraints (e.g., simulate various LCR constraints), simulate behavioral responses, and quantify system-wide impacts.
  - Identified liquidity surpluses or shortfalls can inform design and calibration of supportive policy measures under stress, such as central bank emergency lending assistance (ELA) or changes in the perimeter of eligible collateral or eligible counterparts.

### Narrative design — four layers of tailored stress events
- Layer 1: Global tightening triggering investor selling off sovereign and corporate bonds; margin calls for existing funding and derivatives positions are triggered.
  - High share of foreign holdings of domestic sovereign securities and corporate shares: "currently at around 20 and 33 percent".
  - Trigger channels: tighter US monetary policies, deteriorating fiscal position, subdued foreign investor risk appetite, negative growth outlook, geopolitical reasons; March 2020 cited as reminder.
- Layer 2: Tighter global funding conditions triggering credit and liquidity lines of corporates.
  - At "around 3 trillion pesos", corporate credit lines may be triggered at a high rate during global liquidity stress as firms seek alternative funding.
- Layer 3: Capital outflows via wholesale deposit run-off.
  - Wholesale deposit run-offs occur when firms move deposits on-shore to off-shore amid fears of deteriorating domestic conditions, weakened currency, persisting inflationary pressure, further U.S. monetary policy tightening, new COVID-19 waves, or refinancing needs abroad.
- Layer 4: Redemption shocks triggering investment funds liquidity strains.
  - Redemptions can pressure development banks, commercial banks, and other nonbank institutions via loss of repo financing or refinancing options for maturing short-term funding from investment funds.
  - Impact can be mitigated by collateralized nature of transactions if counterparties find other willing institutions.

### Shock specification and calibration
- Shocks are calibrated for variables designated as triggering points within each narrative layer.
- Concept: use a beta distribution with symmetric bell-curve shape (close to normal) and upper and lower bounds; correlation factor of 0.9 applied to simulate high correlation under stress.
- Variable ranges (units in fraction — not percent, as noted in source):
  - Credit line triggering rate: [0,40]
  - Bond selling rate: [0,40] based on historical maximum selloff
  - Sovereign bond yield shock: [0,4] with average shock at 200 bps
  - Corporate bond yield shock: [0,8] with average shock at 400 bps
  - Wholesale deposit run-off rate: [0,50]
  - Retail deposit run-off rate: [0,20]
  - Share redemption: [0,40]
  - Short term debt phase-out rate: [0,40]

### Market valuation approach and channels of impact
- Market valuation shocks on debt securities:
  - Initial calibration of yield shocks simulates parallel shifts of yield curves across maturity buckets.
  - Higher upper bound shift for corporate securities; bank bond yield shocks assumed higher than sovereign and lower than corporate securities.
  - Granular holdings by agent, encumbrance status, eligibility for central bank operations, and maturity buckets used with calibrated yield shocks to derive market valuation impacts using a modified duration approach.
- Two main channels by which market impacts affect agent balance sheets:
  - Reduced market value (higher market discount) on unencumbered collateral.
  - Margin calls on encumbered collateral underlying funding and derivative margin transactions.
- Simultaneous occurrence of both channels broadly increases tail risks of a system-wide liquidity stress.

### Market clearing pecking order and behavioral assumptions
- Pecking order mimics behavioral responses under stress:
  1. Use cash and cash equivalents first to absorb liquidity outflows.
  2. If insufficient, phase out (or not roll over) outstanding reverse repo transactions or short-term bond investments.
  3. If liquidity gaps remain, pledge unencumbered collateral via repo transactions for additional liquidity support.
- Starting assumption: utilization of counterbalancing capacity is very accommodative; agents withdraw liquidity from others only after exhausting own buffers.
- Commercial banks and the central bank are the main counterparties for repo transactions to backstop the system.

*Source: IMF staff; data compiled by Banxico as of December 2021 (as presented in the original chapter).*

### 150.     The sovereign securities that are sold off by foreign investors under layer 1 of the

### The sovereign securities that are sold off by foreign investors under layer 1 of the

### Shock absorption and repo mechanics
- Sovereign securities sold off by foreign investors under layer 1 are absorbed pro-rata by all market agents based on their existing holdings of sovereign securities and available liquidity after the initial liquidity shock.
- Purchases are voluntary:
  - Development banks and investment funds may halt purchases once their cash and cash equivalents are fully depleted.
  - Commercial banks can continue more flexibly through repo arrangements with the central bank; a CB repo is treated as a back-to-back transaction even if commercial banks lack sufficient cash at hand.
- Phase-out of a reverse repo contract is considered liquidity neutral:
  - Maturity or revocation withdraws cash and returns underlying collateral (mostly debt securities) to the counterparty.
  - This converts encumbered assets back to unencumbered assets by reversing the original haircut and then applying the discounted market price specified by the shock, increasing the counterparty’s liquid buffer.
  - Released encumbered collateral on repo termination is allocated into unencumbered collateral pro-rata based on the relative share of the starting-point encumbered corporate and sovereign securities for each agent.
- Limited information on repo encumbrance composition leads to pro-rata allocation of released collateral.

### System-wide liquidity analysis: main results
- Overall resilience:
  - The financial system remains resilient against the four narratives with commercial banks backstopping liquidity needs of all agents (Figure 42).
  - Under the most severe test with combined shocks and assuming no binding regulatory constraints for all agents, commercial banks show only marginal liquidity shortfalls (a thin negative tail in their liquidity distribution), mainly driven by triggering of contingent credit lines and wholesale deposits’ outflows, while acting as shock absorbers by providing liquidity via repo transactions.
- Role of other agents:
  - Development banks and investment funds can withstand significant liquidity outflows, though risks depend on commercial bank behavior.
  - Access of investment funds to the repo market could enhance their liquidity position and improve system resiliency.
- Correlation effects:
  - Higher correlation of liquidity shocks under stress intensifies downside risks for system-wide liquidity.
  - A higher correlation factor yields a flatter distribution for commercial banks with a fatter left tail, indicating amplified liquidity stress (Figure 43).
- Transmission channels to commercial banks:
  - Corporates contribute most to decline in commercial banks’ net liquidity position due to wholesale deposit exposure and contingent credit/liquidity lines, only marginally offset by small reduction in margin calls and increase in unencumbered assets from phase-out of short-term repo financing provided by corporates.
  - Investment funds rank second in outflow contribution but also bring roughly equal inflows to commercial banks—most transactions are repo or direct sale of debt securities and are liquidity neutral (exchange of cash and liquid debt securities).
  - Development banks place little deposits into or obtain credit lines from commercial banks; they have an illiquid asset profile with lending to private sector and sovereign securities holdings, a high share already encumbered for short-term funding.
  - Government actions (price impact and transactions in sovereign securities) can indirectly reduce commercial banks’ liquid asset holdings (negative net impact); purchases of sovereign securities by commercial banks are liquidity neutral.
  - Households’ contribution is minimal as they place mostly stable retail deposits into commercial banks.

### Quantified contributions and exposures
- Development banks’ vulnerability:
  - High contribution to development banks’ liquidity outflows from corporates and investment funds: 45 and 40 percent, respectively.
  - Share of short-term financing extended by corporates and investment funds to development banks: 35 and 27 percent out of total short-term financing as of end-2021.
- Banking sector coverage and data:
  - Institutions included: All D-SIBs (6 banks) and other important banks (4 additional mid-tier commercial banks) — 10 commercial banks within scope represent approximately 84 percent of banking sector assets.
  - Data as of December 2021 (cut-off); end-2019 data may be used for comparisons and sensitivity analysis.
- Stress test horizon and macro adverse calibration:
  - Stress test horizon: Three years (2022 Q1 – 2024 Q4).
  - Adverse scenario severity: deviation of Mexico real GDP from its baseline of 11.3 percent by 2023, with a 2.3 Standard Deviation move in two-year cumulative real GDP growth rate, and a 6.3 percent increase in unemployment rate from its baseline.
  - Domestic layer assumes monetary policy remains accommodative with short-term interest rates decreasing towards the 4 percent effective lower bound.

### Sensitivity analysis and policy experiments
- Commercial banks’ behavioral constraints:
  - Freezing repo activities of commercial banks once they reach a 100-percent LCR limit (policy experiment) can create knock-on effects to development banks, pushing part of the development banks’ liquidity distribution into negative territory (Figure 45 left panel).
  - If commercial banks pull back on short-term funding instead of freezing repo activities (e.g., deposit withdrawal or refusal to refinance short-term debt of development banks), the net liquidity distribution for development banks shifts left as well, though not as severely as the repo-freeze experiment (Figure 45 right panel).
  - Pullback on short-term funding is considered a more likely behavioral response under stress because it increases banks’ liquidity buffers, while freezing repo operations is liquidity neutral in theory.
- Policy implications from experiments:
  - Commercial banks act as final shock absorbers by providing liquidity through repo transactions; their liquidity shortfalls are marginal even under severe narratives.
  - Development banks are more vulnerable due to funding concentration; vulnerability is amplified when binding liquidity constraints (e.g., mandatory LCRs for commercial banks or minimum liquidity buffers for investment funds) are present or when banks behave conservatively under stress.
  - Expanding access of investment funds to the repo market could strengthen system-wide resiliency (simple policy assessment experiment).

### Stress-testing framework highlights (top-down assumptions)
- Methodology and channels:
  - Balance sheet approach with static balance sheet assumption; projections for key balance sheet, income, and capital items.
  - Credit risk, market risk, net interest income, and non-interest income projections for baseline and macro adverse scenarios.
  - Five loan segments: corporates, mortgages, financials, government, and consumer credit.
  - PD PIT satellite models, LGD PiTs via Frye-Jakobs method, and Bayesian Model Averaging (BMA) techniques to control model uncertainty.
  - RWAs adjusted for changes in credit quality; operational expenses over total assets kept at 2021 levels.
- Scenario features:
  - Baseline scenario: April 2022 World Economic Outlook (WEO) projections.
  - Macroeconomic adverse scenario calibrated using the Global Macrofinancial Model (GFM) with supply-side disruptions, higher inflation, and a Fed tightening faster than expected (about one percentage point within the first year), triggering capital outflows, currency depreciation, and higher long-term rates in emerging markets.
  - Domestic confidence shocks apply downward pressure on domestic demand, aggravating corrections on real estate and equity prices, curtailing bank profitability and tightening interbank market conditions.
- Sensitivity analyses:
  - Targeted sensitivity analysis on triggering of credit lines to NFCs and financial entities.
  - Solvency sensitivity analysis to capture more pronounced interest rate shifts.
  - Partial credit and market risk analysis on development banks and the twenty largest credit-providing NBFIs using CNBV’s top-down models and infrastructure.

*Source: Banxico and IMF staff estimate; IMF staff (text as provided).*

### 4. Risks

### 4. Risks

### Risks/factors assessed
- Credit risk captures all on-balance/off-balance sheet exposures at amortized cost by regulatory exposure sector. Exposures are largely domestic; therefore, no scenarios and parameter paths would be required for geographies outside Mexico.
- Market risk is reflected in valuation effects of FVTPL and FVOCI positions, as well as net open financial positions (i.e., equities, funds, and inflation-linked instruments exposures). Scenario-based Interest rate curves are used to infer reference interest rate changes. The adverse macro scenario is further augmented to include financial variables that are needed to produce accurate projections for fair value positions (like corporate spread rate shock or bank issued bonds spread shock).
- Net interest income is affected by projecting effective interest rates by asset/liability class. Policy rates and wholesale/interbank rates will directly follow the macroeconomic scenario paths and a panel econometric approach will be used to define the velocity of passthrough rates to all remaining asset and liability segments.
- Shocks to non-interest income are simulated to capture varying degrees of market-sensitive components of non-interest income.
- Projected RWA densities are also capturing a twofold impact: deterioration of credit quality and partial/full unwinding of relevant policy support measures.

### Behavioral adjustments
- Under the static balance sheet assumption exposures remain constant and do not evolve in accordance with credit growth assumptions of scenarios.
- For NII, maturing assets/liabilities are assumed to be replaced by instruments of the same type, maturity but at current rates.
- There is no recognized interest on non-performing exposures.
- If banks’ capital falls below regulatory requirements, no prompt corrective action is assumed.
- Banks are assumed to pay 30 percent of their profits as tax. Dividend payout ratio is assumed to be the maximum of 40 percent or the payout ratio of the cut-off year unless the capital conservation buffer falls below 2.5 percent.

### Calibration of risk parameters
- Currently the banking system is regulated under a full Basel III prudential framework.
- Accounting provisions are set by CNBV regulations (IFRS 9 was only implemented in January 2022 and CNBV has the mandate to set requirements for the accounting loan loss provisioning). In this context the stress test analysis will follow regulatory definitions of PDs and LGDs where applicable.
- Currently credit exposure portfolios are under the Standardized (STA) and the Advanced Internal Rating Based (A-IRB) regulatory approach.
- Risk-weighted asset densities are either assumed to remain constant for STA portfolios and following the PD PIT path (making use of a smoothening factor for the TTC effect).

### Regulatory/accounting and market-based standards
- In the baseline, hurdle rates include the regulatory minimum (CET1: 4.5 percent, Tier1: 6 percent, Total Capital: 8 percent) and any applicable capital buffers (CCB, D-SIB surcharge, P2R). D-SIB charge ranges from 0.6 percent to 1.5 percent for the banks within scope.
- In the adverse scenario, the regulatory minimum (including D-SIB surcharge and P2R) is assumed to be the hurdle rate, as banks can draw down the CCB. Note that D-SIB surcharge is not considered as a buffer in Mexico.
- Hurdle rates are based on the CET1, Tier1, and Total Capital ratios.

### Reporting Format for Results (solvency)
- System-wide evolution of aggregate CET1 and capital ratios.
- Distribution of banks’ capital positions.
- Contribution to key drivers to system-wide net income and capital position, including differences between the baseline scenario and the adverse scenario.
- Share of institutions with capital below the hurdle rates.

---

### Banking Sector: Liquidity Risk

### 1. Institutional Perimeter
- Institutions included: The fifty commercial banks in Mexico at the highest level of consolidation.
- Market share: 100 percent of commercial banking sector assets.
- Data and baseline date:
  - Banxico’s regulatory reports monitoring the Liquidity Coverage Ratio and the Net Stable Funding Ratio and the additional (synthetically constructed) monitoring report capturing liquidity contractual maturity ladder.
  - Data as of December 2021; December 2019 data will also be used to highlight the impacts of the pandemic on liquidity positions of banks.
  - Scope of financial consolidation: group-wide at the highest level.

### 2. Channels of Risk Propagation — Methodology
- The exercise is based on three types of tests—LCR test, cash-flow analysis and NSFR revision.
- The LCR test is in line with the standard Banxico (and Basel compliant) monitoring tool, featuring total consolidated liquidity and liquidity in significant currencies (mainly USD).
- A set of scenarios for LCR outflows and HQLA haircuts is used to produce stressed LCR ratios (by currency and at the consolidated level).
- For the LCR test, the stress test horizon is 30 days.
- The cash-flow analysis analyzes the net cash balance (as a proxy of banks’ resiliency to liquidity stress events), accounting for available unencumbered assets, contractual cash inflows and outflows, and behavioral flows.
- For the cash-flow analysis, a range of scenarios featuring funding run-off rates, liquid assets haircuts and assumptions on inflows and outflows of increased severity for different durations of liquidity stress are explored (a stress-horizon of 3 months is used as the central assumption). Positive counterbalancing capacity post-scenario implies bank resiliency, negative is an indication of liquidity stress.
- For the cash-flow analysis, asset haircuts reflect two components: (i) shocks to interest rates and asset prices as captured the macrofinancial scenarios; and (ii) additional haircuts required by counterparties to accept specific assets as collateral for secured funding transactions.
- The NSFR became a binding requirement for Mexican banks in March 2022. For monitoring purposes, banks have been reporting NSFR calculations to competent authorities since 2017.

### 3. Risks and Buffers
- Risks:
  - Funding liquidity risk is reflected in funding run-off rates and asset roll-over rates, the latter providing cash inflows related to non-renewal of maturing assets.
  - Market liquidity risk is reflected in asset haircuts, which could be influenced by market movements, potential fire sales and collateral supply considerations.
- Behavioral adjustments:
  - Liquidity from the central bank’s emergency lending assistance (ELA) is not considered.
  - Inflows from maturing loans are ignored (cash-flow analysis, after a certain level of scenario severity) capturing a systemic liquidity stress scenario vs a bank-idiosyncratic one.
  - The cash-flow analysis may consider some behavioral assumptions about a counterparty’s ability or willingness to transact based on banks’ solvency and liquidity conditions.

### 4. Tail shocks — Scenario analysis
- For the LCR test, 12 scenarios are considered as a combination of: (i) three scenarios on liquid assets shocks (regulatory, mild, and severe), and (ii) four scenarios on liability outflows; regulatory, one reflecting retail outflows, one reflecting higher wholesale outflows, and one combining the retail and wholesale outflows.
- For the cash-flow analysis, a series of scenarios are considered, with a range from mild to severely adverse liquidity conditions. The cash-flow analysis considers both funding and market liquidity risks.

### 5. Regulatory and Market-Based Standards and Parameters
- Calibration of risk parameters:
  - LCR tests are based on regulatory and stress parameters.
  - Cash-flow analysis may incorporate relevant second-round effects.
  - Stress funding run-off rates, asset roll-over rates, and asset haircuts are calibrated based on empirical evidence and relevant international experiences.
- Regulatory/accounting and market-based standards:
  - LCR per Basel III; the hurdle at 100 percent.
  - Net cash balance for the cash-flow analysis; to pass, a non-negative net cash balance is required, where the balance reflects net funding outflows and counterbalancing capacity.
  - NSFR per Basel III; the hurdle at 100 percent.

### 6. Reporting Format for Results (liquidity)
- Changes in the system-wide liquidity position, including important drivers for cash outflows, cash inflows and counterbalancing capacity.
- Distribution of banks’ liquidity positions.
- Number of institutions with LCR/NSFR below 100 percent and/or negative net cash balance.
- Amount of liquidity shortfalls, including by currencies.

### 7. Sensitivity Analysis
- The analysis would cover policy support measures and will identify how such measures have impacted regulatory liquidity metrics.
- The analysis will also assess how the gradual measure unwinding will have affect liquidity positions of banks.

### 8. Infrastructure
- For the LCR test, Banxico’s infrastructure to run the scenario developed by IMF staff and Banxico’s Liquidity at Risk tests. For cash flow analysis, fully comprehensive infrastructure developed by IMF staff using newly introduced (March 2022) Banxico’s regulatory reports as a data repository. MATLAB and Excel based.

---

### Financial System: Contagion Risk

### 1. Institutional Perimeter
- Institutions included: All commercial and development banks, brokerage houses, investment and pension funds and the largest credit providing NBFIs (subject to data availability) in Mexico, at the highest level of consolidation.
- Market share: Almost the entire system in terms of asset coverage.
- Data and baseline date:
  - Source: Supervisory data and ad-hoc data request.
  - Data as of December 2021 (random day cut-off, to avoid window dressing effects).
  - BIS consolidated banking statistics, data as of end-Sept 2021.

### 2. Channels of Risk Propagation — Methodology
- Interbank and cross-border network model by Espinosa-Vega and Solé (2010).

### 3. Risks and Buffers
- Risks: Credit and funding losses related to interbank/inter-entity cross-exposures (and cross-border banking exposures).
- Buffers: Banks’ and brokerage houses’ own capital buffers; other entities are not assumed to default in the simulation (internal loss absorption).

### 4. Tail shocks — Size of the shock
- Pure contagion: default of individual institutions.
- Several types of cross-entity exposures considered: secured, unsecured, crossholdings of debt instruments, settlement exposures. Different LGDs might be used, depending on exposure type.
- Simulation of multiple concurrent defaults may also be examined.

### 5. Reporting Format for Results (contagion)
- Contagion and vulnerability indicators.
- System-wide capital shortfall.
- Bank-level capital shortfall.
- Number of undercapitalized and/or failed institutions, and their shares of assets in the system.
- Evolution and direction of spillovers.

---

### Financial System: System-Wide Liquidity (SWL) Analysis

### 1. Institutional Perimeter
- Entities included:
  - Central Bank
  - Government
  - Commercial Banks
  - State-owned banks
  - Investment Funds
  - NFCs
  - Households
  - Foreign investors
- Data and baseline date:
  - Ad-hoc data request template provided by the FSAP team to Banxico, capturing:
    - Available collateral (encumbered and unencumbered) by asset class, remaining maturity bucket and eligibility for CB operations.
    - Existing collateralized funding and margin positions for all agents.
    - Composition of the most important segments of B/S assets and liabilities by agent type, as well as bilateral exposure between agents informed by who-to-whom holdings.
  - Data as of December 2021, at the aggregate B/S level and on a best effort basis.
  - Scope of financial consolidation: group-wide at the highest level.

### 2. Channels of Risk Propagation — Methodology
- Analysis conducted at the aggregated B/S data for each type of economic agent.
- For each scenario, the liquidity counterbalancing capacity for each type of economic agent is measured, in response to direct shocks (funding and market) and after considering second round effects due to calls on available collateral for existing funding and margin positions.
- Shocks are generated based on correlated distributions (copula) with flexibility of adjusting ranges of the distributions and correlation factors between distributions to reflect different level of severity.
- Cash and unencumbered collateral are considered as accessible liquidity buffers.
- Pecking order of the utilization of liquid assets: 1. Cash and equivalences 2. Short term assets including short term paper and outstanding reverse repos 3. Repos using unencumbered assets.
- Willingness and capacity to roll-over existing funding positions across agents are assessed after measuring liquidity excess or shortfalls.
- The resilience of the system (and of individual agents) is assessed based on the net liquidity distribution across the number of simulated scenarios (shortfall probability density).
- Agents will be classified in accordance with their liquidity shortfall propensity and with respect to their contribution to the overall system-wide resiliency or vulnerability.
- Existing counterbalancing capacity of unencumbered collateral is measured against severe tail events as the point in the distribution that would force Banxico to increase the perimeter of eligible collateral.

### 3. Risks and Buffers
- Risks:
  - Funding liquidity risk is reflected in funding run-off rates, capital outflows, share redemption and offshore switching.
  - Market liquidity risk is reflected in asset haircuts, influenced by market movements, potential fire sales and collateral supply considerations.
- Buffers:
  - Available unencumbered collateral (CB eligible and non-eligible), cash position and capacity to absorb pressure in all market segments considered (sovereign, repo, and derivatives markets, etc.).

### 4. Behavioral Assumptions
- Liquidity from the central bank’s emergency lending assistance (ELA) or any other increase in the perimeter of eligible collateral or eligible counterparts is not considered.
- Pecking order in the way agents with excess (insufficient) liquidity decide to (not) roll-over funding positions may be important.
- Binding liquidity requirements (LCR constraints) can be switched on/off.

### 5. Tail shocks — Scenario analysis
- The analysis narrative would entail the simulation of a material number of scenarios consisting of a series of random (but correlated) layers of shocks:
  - Sovereign market repricing shocks due to capital outflows and risk premia reassessment.
  - Drawdown of existing credit and liquidity facilities by NFCs due to global tightening funding conditions.
  - Run-offs on wholesale and retail deposits and switch to offshore accounts due to rebalancing of funding requirements.
  - Investment Fund redemption shocks and associated short-term funding stress (e.g., via the repo market).
  - FX depreciation and shocks attributed to the shortage of sufficient FX reserves (implemented but muted).
  - Shocks attributed to dislocated derivatives markets and margin requirements and derivative basis shocks (implemented but muted).

### 6. Sensitivity analysis — Shock severity and policy experiment
- Single factor sensitivity analysis by increasing of correlation factor between shock parameters.
- Mute repo or pull back other short-term funding (deposits or short-term paper) from commercial banks to other agents as commercial banks reach liquidity regulatory threshold (e.g., LCR).
- Allow expanded access of investment fund to repo market to assess benefit of repo participation.

### 7. Regulatory and Market-Based Standards
- Regulatory Standards:
  - LCR and other liquidity constraints are not used for the identification of bank pass/failure since the analysis is performed at the aggregate level (not entity specific).

### 8. Reporting Format for Results (SWL)
- Probability distribution of excess/shortfall for the system and by agent type.
- Impact attribution by agent type in the overall resiliency or vulnerability.
- Shortfall thresholds for different agents.
- Contribution of each layer of shocks to the overall liquidity shortfalls.

### 9. Infrastructure
- Fully comprehensive and novel infrastructure developed by IMF staff using the ad-hoc data request as a data repository. MATLAB based.

---

### Appendix III. Credit and Market Risks in Development Banks and Nonbank Financial Institutions
- CNBV has conducted a stress-test analysis of credit and market risks for all commercial banks, development banks and the largest NBFIs using the FSAP adverse scenario. While the FSAP team has independently conducted a fully-fledged top-down solvency stress test for the top-10 commercial banks (see Section IV. B), collaboration with CNBV has allowed the team to partially expand the analysis by assessing the impact on credit and market risk for commercial banks, the six development banks and the twenty largest NBFIs within CNBV’s regulatory perimeter.
- Development banks and largest NBFIs represent only 17 percent and 1.4 percent of total assets respectively, but are highly interconnected and could transmit shocks to the rest of the system.
- The results show that the impact of market and credit risks is limited (Figure 47). Market risk is contained and driven mainly by the revaluation of bonds and the impact on P&L from derivatives’ exposures for both commercial and development banks (NBFIs do not have material market risk exposures in their portfolio).
- Reflecting the different credit quality of the loan portfolios, expected losses (loan loss provisions) as a share of risk weighted assets under the adverse scenario are higher for NBFIs and development banks compared to commercial banks, and increasing in the scenario horizon.
- A direct comparison with the results of the FSAP team’s exercise for the overlapping banks is not feasible, given major differences in the methodology used, particularly regarding satellite model estimation, granularity of data sources and modeling differences.
- Note: The term NBFI is hereby used to denote the segment of smaller non-deposit taking credit provisioning entities and does not include insurance companies or pension and investment funds.

---

### Appendix IV. The Estimation of Satellite Models — Credit Risk
- A series of panel econometric models were estimated to produce scenario dependent forward paths for Point-in-Time (PiT) default probabilities (PD) for different exposure segments.
- The selection of the modelling approach was largely driven by the quality and availability of bank level historical data for calibration purposes.
- Challenges noted:
  - Obtaining reliable and informative bank specific default rates is a challenging task because any approach that would back out default rates from impairment flows might suffer from the presence of significant outliers because of spike of individual exposure impairments or random write-off decisions might introduce distortionary impact on the inferred historical PD rates.
  - The lack of cyclicality or volatilities for corporate and mortgage portfolios over time may compound the complexity and challenges in the estimation and projections of bank specific probability of default.

*Source: 1mexea2022007 - 4. Risks*

### 2.     The analysis used bank specific PD data obtained from country authorities and covers

### 1mexea2022007 - 2.     The analysis used bank specific PD data obtained from country authorities and covers

### Data and scope
- Quarterly PD time series starting 2007 and ending at cutoff date 2021Q4 were obtained from country authorities for each bank in the stress test.
- Covered portfolio segments: corporate, household retail (consumer), and household mortgage.
- Rationale for cutoff: maximize sample coverage and reflect that most policy measures, including debt moratoria, had been phased out by end-2021 and banks’ credit quality had deteriorated since the pandemic onset.

### Methodology: PD modeling and projection
- A model averaging technique was employed to model and project default rates at individual bank and portfolio levels.
- Approach: panel fixed effect regression on a pool of equations per dependent variable; weights assigned based on relative predictive performance to produce a “posterior model” equation.
- Pool of equations considered all possible combinations of predictors from a candidate set including real GDP growth, unemployment rates, housing prices, short- and long-term interest rates, and others.
- To ensure PD predictions in [0,1] and capture nonlinearities, a logit transformation was applied to the original PD.
- Logit-transformed PDs were modeled as a linear function of exogenous macro-financial factors; lags or rolling sums of explanatory variables were allowed.
- Conditional PiT forecast for each segment and each bank were generated based on estimated coefficients and fixed effects under both the baseline and adverse scenarios.

### Explanatory variables, data sources and calibration
- Input variables sourced from IMF World Economic Outlook (WEO), Datastream and Haver Analytics.
- Main explanatory variables used: real GDP growth, unemployment rate, short-term interest rate, long-term sovereign bond yield, term spread, housing price growth.
- Transformation rule: variables other than interest rates and unemployment were subject to annual growth transformation; interest rates and unemployment taken in original level in percent.
- Calibration sample period: 2007–2021 at quarterly frequency.

### Model selection criteria
- Users can vary model specifications (number of explanatory variables under permutation, number of lags).
- Main information criteria used: R-square, adjusted R-square, AIC, quality of in-sample forecast, and size of impact in forecasting period.
- Ideal candidate model characteristics: relatively high R-square, small root-mean-square-error, historically consistent size of impact under stress.

### Estimated PD models (Table 1 excerpts and diagnostics)
- Dependent variables: probability of defaults in logit form for Corporate Loans, Consumer Loans, Mortgage Loans.
- Coefficient highlights (entries preserved as in source table):
  - GDP growth, percent, yoy: Corporate -0.002; Consumer -0.007*; Mortgage 0
  - Unemployment rate, percent: Corporate 0.179*; Consumer 0.124*; Mortgage 0.337*
  - Short term interest rate, percent: Corporate 0; Consumer 0.096*; Mortgage 0
  - Long term sovereign bond yield, percent: Corporate 0; Consumer 0; Mortgage 0.035*
  - Term spread, percent: Corporate 0; Consumer 0.17*; Mortgage 0.020
  - House price growth, percent, yoy: Corporate 0; Consumer 0; Mortgage -0.002
  - Intercept: Corporate -3.941*; Consumer -3.12*; Mortgage -4.448*
- Sample and fit statistics:
  - Number of observations: Corporate 585; Consumer 600; Mortgage 600
  - R square: Corporate 0.35; Consumer 0.33; Mortgage 0.62
  - Fixed effect: Yes for all segments

### Key findings on PD determinants
- Unemployment rate plays a significant role across all portfolios (high P values and sizable coefficients).
- Real GDP growth exhibits a negative relationship with PDs.
- Rise in unemployment and short-term interest rates drive PDs up due to lower affordability and higher borrowing costs.
- Term spread has a significant positive impact on consumer loans.
- Housing price growth is negatively correlated with mortgage PDs.

### PD projections and scenario behavior
- Scenario-dependent PiT PD projections reveal larger shocks for household retail portfolios and are broadly in line with historical stress episodes (e.g., GFC).
- Relative severity under stress (higher PDs): household retail > mortgage > corporate, reflecting lack of collateralization in retail portfolios and idiosyncratic differences across segments.
- Forward-looking PD paths were generated under baseline and adverse scenarios (figures in source show segment-specific projected PD trajectories from 2007Q1 through forecast horizon).

### Interest rate risk: data and computation
- Interest rates on assets and liabilities approximated by effective rates for each bank.
- Authorities provided total outstanding amounts of various asset and liability items and periodic interest income and expense flows on a quarterly basis from 2008Q1 to 2021Q4.
- Effective lending and funding rates computed for front and back book and used as inputs for satellite models.
- Macroeconomic inputs for interest rate models sourced from IMF WEO, Datastream and Haver Analytics; adverse scenario macro series followed the stress-test scenario.

### Interest rate modeling approach
- Satellite models estimated aggregate funding and lending rates at individual bank and portfolio level using same model selection and averaging techniques as credit risk models.
- Modeled rates include: interest rates on corporate, household retail, mortgage loans, securities holdings (assets); overnight retail deposits, term retail deposits, wholesale deposits, securities issuance (liabilities).
- Projected bank-specific interest rate paths constructed by attaching period changes of effective interest rates in forecasting horizon to bank-specific starting points.
- Sequential estimation: funding rate estimated first, then used as input for lending rate projection to allow partial passthrough from funding cost to lending rate.

### Interest rate model findings (Table 2 excerpts and diagnostics)
- Dependent variables: interest rates in percent for Corporate Loans, Consumer Loans, Mortgage Loans, Assets Debt Securities, Overnight Retail Deposits, Term Retail Deposits, Wholesale Deposits, Liabilities Debt Securities.
- Coefficient highlights (entries preserved as in source table):
  - Unemployment rate, percent: Corporate 0; Consumer 1.679*; Mortgage 0.572*; Assets Debt Securities 0; Overnight Retail Deposits 0; Term Retail Deposits 0; Wholesale Deposits 0; Liabilities Debt Securities 0
  - Short term interest rate, percent: Corporate 0; Consumer 0; Mortgage 0.124*; Assets Debt Securities 0; Overnight Retail Deposits 0.273*; Term Retail Deposits 0.596*; Wholesale Deposits 0.803*; Liabilities Debt Securities 0.371*
  - Long term sovereign bond yield, percent: Corporate 0.015; Consumer 0; Mortgage 0.073; Assets Debt Securities 0.914*; Overnight Retail Deposits 0.055; Term Retail Deposits -0.279; Wholesale Deposits 0.387; Liabilities Debt Securities 0
  - Term spread, percent: Corporate 0.195*; Consumer 0.966; Mortgage 0; Assets Debt Securities 0; Overnight Retail Deposits -0.165; Term Retail Deposits 0.224; Wholesale Deposits -0.317; Liabilities Debt Securities 0
  - Inflation, percent, yoy, lagged: Corporate 0.092*; Consumer 0; Mortgage 0; Assets Debt Securities 0; Overnight Retail Deposits 0.004; Term Retail Deposits 0; Wholesale Deposits 0.539*; Liabilities Debt Securities 0
  - Overnight retail deposit, percent: Corporate 0.182*; Consumer 0.497; others 0
  - Term retail deposit, percent: Corporate 1.023*; Consumer 0.080; others 0
  - Intercepts: Corporate 3.442*; Consumer 23.671*; Mortgage 7.245*; Assets Debt Securities -3.349*; Overnight Retail Deposits 0.845*; Term Retail Deposits 3.691*; Wholesale Deposits 1.157*; Liabilities Debt Securities 22.458*
- Sample and fit statistics:
  - Number of observations: Corporate 560; Consumer 549; Mortgage 560; Assets Debt Securities 560; Overnight Retail Deposits 560; Term Retail Deposits 560; Wholesale Deposits 560; Liabilities Debt Securities 464
  - R square: Corporate 0.90; Consumer 0.64; Mortgage 0.16; Assets Debt Securities 0.51; Overnight Retail Deposits 0.78; Term Retail Deposits 0.15; Wholesale Deposits 0.15; Liabilities Debt Securities 0.33
  - Fixed effect: Yes for all reported models

### Interest rate projection insights
- Projected interest rate paths align with portfolio characteristics:
  - Liabilities: larger impact on long-term and unsecured debt portfolios (term deposits, wholesale deposits) relative to highly liquid short-term funding (overnight deposits).
  - Assets: potential increase in lending rates could be hindered by rising PDs of existing borrowers; a partial passthrough from funding to lending rates was applied to be conservative.
- Main contributors in projections of bank interest income and funding cost: unemployment rate and short-term money market rate.
  - 3-month money market rate explains majority of movement in interest expense.
  - On lending side, unemployment rate is more important than short-term interest rate for consumer and mortgage loans.
  - Term spread, inflation, and pass-through from overnight and term retail deposits are significant for corporate loans, with sizable impact particularly from term retail deposits.

*Source: IMF staff estimation; data from Banxico, IMF World Economic Outlook, Haver, Datastream, and IMF staff calculations.*

### Appendix V. LCR-Based Stress Scenario Parameters

### Appendix V. LCR-Based Stress Scenario Parameters

### Haircuts (Unencumbered assets; Level 1, Level 2A, Level 2B)
- Cash (not to include the cash balances allocated to cover operating costs): 100% | 100% | 100%
- Deposits at Bank of Mexico — Monetary Regulation Deposits: 100% | 100% | 100%
- Other unencumbered deposits (single account balance, deposits: IEIR and OMO or any other deposit at Bank of Mexico): 100% | 100% | 100%
- Demand deposits at other central banks: 100% | 100% | 100%

Holdings of unencumbered debt securities and shares (received in repos or loans, not encumbered elsewhere)
- Debt securities with credit risk weighting of 0% assigned (issued or backed by Government of Mexico, Bank of Mexico, and the Bank Savings Protection Institute): 100% | 95% | 90%
- Mexican development banks: 100% | 95% | 90%
- Foreign governments, foreign central banks, and decentralized agencies of foreign governments: 100% | 100% | 100%
- International bodies (BIS, IMF, European Commission, multilateral development agencies): 100% | 100% | 100%

Debt securities with credit risk weighting other than 0% (degree of risk ≤ 2)
- Governments or central banks in local currency of the country: 100% | 95% | 90%
- Governments or central banks in foreign currency (if currency matches institution's liquidity needs): 100% | 95% | 90%

LEVEL 2A ASSETS (debt securities to which a credit risk weighting is assigned)
- Decentralized agencies of the federal government: 85% | 80% | 75%
- Mexican federal entities and municipalities and bodies under them: 85% | 75% | 65%
- Foreign governments, foreign central banks, and decentralized agencies of foreign governments: 85% | 80% | 75%
- Multilateral development agencies: 85% | 75% | 65%
- Mexican development banks and Mexican public development [fomento] funds and trusts: 85% | 75% | 65%
- Debt securities eligible as Level 2A issued by nonfinancial institutions other than sovereigns, central banks, and public sector entities: 85% | 75% | 65%

LEVEL 2B ASSETS
- Debt securities eligible as residential mortgage-backed Level 2B assets: 75% | 70% | 65%
- Debt securities eligible as Level 2B assets, issued or backed by nonfinancial institutions other than sovereigns, central banks, and public sector entities: 50% | 40% | 30%
- Shares of nonfinancial enterprises included in the main index of the Mexican Stock Exchange (BMV) with high or medium marketability and without historically cumulative decline > 40% over 30 days: 50% | 25% | 0%
- Debt securities eligible as Level 2B assets issued or backed by foreign governments or foreign central banks: 50% | 40% | 30%
- Assets that fulfill Level 2A criteria but have historical cumulative decline > 10% and ≤ 20% over 30 days: 50% | 40% | 30%

### Run-off rates for Outflows — Unsecured funding (Retail, Wholesale, Combined scenarios)
Retail deposits (fully insured by IPAB or corresponding deposit insurance; individuals, individual business activity, non-financial entities other than sovereigns/central bank/public sector)
- Deposits in local currency paying interest ≤ 28-day IEIR, and foreign currency paying ≤ SOFR
  - In transactional accounts or amount that fulfils operational purposes:
    - Payable on demand: 5% | 10% | 5% | 10%
    - Maturity: 5% | 10% | 5% | 10%
  - In accounts other than transactional accounts and that do not fulfil operational purposes:
    - Payable on demand: 10% | 20% | 10% | 20%
    - Term deposit: 10% | 20% | 10% | 20%

- Deposits in local currency paying interest > 28-day IEIR, and foreign currency paying > SOFR
  - In transactional accounts or amount that fulfils operational purposes:
    - Payable on demand: 5% | 10% | 5% | 10%
    - Term deposit: 5% | 10% | 5% | 10%
  - In accounts other than transactional accounts and that do not fulfil operational purposes:
    - Payable on demand: 10% | 20% | 10% | 20%
    - Term deposit: 10% | 20% | 10% | 20%

Deposits not fully insured by IPAB (received from individuals including business activity)
- Part covered by IPAB:
  - In transactional accounts:
    - Payable on demand: 5% | 15% | 5% | 15%
    - Term deposit: 5% | 15% | 5% | 15%
  - In nontransactional accounts:
    - Payable on demand: 10% | 20% | 10% | 20%
    - Term deposit: 10% | 20% | 10% | 20%
- Part NOT covered by IPAB:
  - In transactional accounts:
    - Payable on demand: 10% | 20% | 10% | 20%
    - Term deposit: 10% | 20% | 10% | 20%
  - In nontransactional accounts:
    - Payable on demand: 10% | 20% | 10% | 20%
    - Term deposit: 10% | 20% | 10% | 20%

Wholesale deposits
- Demand deposits — local currency pay ≤ IEIR, foreign currency pay ≤ SOFR — amount that fulfils operational purposes
  - Part covered by IPAB (sovereigns, central banks, federal entities and municipalities, public sector entities, government entities, development banking sector, public development funds and trusts): 5% | 5% | 15% | 15%
  - Nonfinancial institutions not included in other categories, not fully secured by IPAB: 5% | 5% | 15% | 15%
- Part NOT covered by IPAB — amount that fulfils operational purposes
  - Sovereigns, central banks, federal entities and municipalities, public sector, development banking sector, public development funds and trusts: 25% | 25% | 35% | 35%
  - Domestic and foreign financial entities (excluding development banking sector and public development funds and trusts): 25% | 25% | 35% | 35%
  - Nonfinancial institutions not included in other categories, not fully secured by IPAB: 25% | 25% | 35% | 35%

- Amount that does not fulfil operational purposes — examples:
  - Amount fully secured by IPAB (sovereigns, central banks, federal entities and municipalities, public sector entities, government entities, development banking sector, public development funds and trusts): 20% | 20% | 40% | 40%
  - Part not covered by IPAB received from nonfinancial institutions not included in other categories: 40% | 40% | 60% | 60%
  - Domestic and foreign financial entities (excluding development banking sector and public development funds and trusts): 100% | 100% | 100% | 100%
  - Nonfinancial institutions not included in other categories, not fully secured by IPAB: 40% | 40% | 60% | 60%

Term deposits (various payer categories)
- Sovereigns, central banks, federal entities and municipalities, public sector entities, government entities, development banking sector, public development funds and trusts — amount fully insured by IPAB: 20% | 20% | 40% | 40%
- Part not covered by IPAB: 40% | 40% | 60% | 60%
- Domestic and foreign financial entities (excluding development banking sector and public development funds and trusts): 100% | 100% | 100% | 100%
- Nonfinancial institutions not included in other categories, not fully secured by IPAB: 40% | 40% | 60% | 60%

Loans (outflows)
- Sovereigns, central banks, federal entities and municipalities, government entities, and public sector entities: 40% | 40% | 60% | 60%
- Development banking sector and public development funds and trusts:
  - Call money loans: 100% | 100% | 100% | 100%
  - Other loans: 40% | 40% | 60% | 60%
- Domestic and foreign financial entities (excluding development banking sector and public development funds and trusts): 100% | 100% | 100% | 100%
- Nonfinancial institutions not fully insured by IPAB: 40% | 40% | 60% | 60%

Debt securities issued by the entity (includes securities with secondary market)
- Debt from money market (debt securities other than those indicated in concept 10025): 100% | 100% | 100% | 100%
- Subordinated debt in circulation: 100% | 100% | 100% | 100%

Transactions by brokerage houses in same ownership as commercial bank
- Financing received through repo transactions backed with debt securities with residual term > 30 days issued by the commercial bank: 100% | 100% | 100% | 100%

### Secured funding (run-off rates for funding backed by asset levels)
- Amount of funding backed with Level 1 assets: 0% | 0% | 0% | 0%
- Amount of funding backed with Level 2A assets:
  - with Bank of Mexico: 0% | 0% | 0% | 0%
  - with counterparties other than Bank of Mexico: 15% | 15% | 25% | 25%
- Amount of funding backed with Level 2B assets (eligible residential mortgage-backed):
  - with Bank of Mexico: 0% | 0% | 25% | 25%
  - with the federal government, government entities, federal entities and municipalities, public sector entities, development banking sector, and public development funds and trusts: 25% | 25% | 50% | 50%
  - with counterparties other than the foregoing: 25% | 25% | 50% | 50%
- Amount of funding backed with Level 2B assets other than residential mortgage-backed:
  - with Bank of Mexico: 0% | 0% | 0% | 0%
  - with the development banking sector and public development funds and trusts: 25% | 25% | 50% | 50%
  - with the federal government, government entities, federal entities and municipalities, and public sector entities: 25% | 25% | 50% | 50%
  - with counterparties other than the foregoing: 50% | 50% | 100% | 100%
- Amount of funding backed with nonliquid assets:
  - with Bank of Mexico: 0% | 0% | 0% | 0%
  - with the development banking sector and public development funds and trusts: 25% | 25% | 100% | 100%
  - with the federal government, government entities, federal entities and municipalities, and public sector entities: 25% | 25% | 100% | 100%
  - with counterparties other than the foregoing: 100% | 100% | 100% | 100%
- Premiums and interest deliverable for operations of secured financing received: 100% | 100% | 100% | 100%

### Other outflows: foreign exchange, derivatives, guarantees, structured vehicles
- Currency to be provided for foreign exchange value-dated transactions (24, 48, 72, and 96 HOURS): 100% | 100% | 100% | 100%
- Sum of outflows from offsetting currency receivable with currency to be provided for each FX transaction: 100% | 100% | 100% | 100%
- Sum of outflows from offsetting currency with securities of value-dated securities purchase/sell transactions: 100% | 100% | 100% | 100%

Derivatives (OTC)
- Outflows for contractual payments pending settlement: 100% | 100% | 100% | 100%
- Contingent outflow for transactions with derivative financial instruments (Look Back Approach, LBA): 100% | 100% | 100% | 100%
- Total contractual outflows scheduled next 30 days for OTC derivatives that may NOT be offset (not part of a master clearing agreement), net of Level 1/2A/2B guarantees delivered: 100% | 100% | 100% | 100%
- Total contractual outflows scheduled next 30 days for derivatives offset by inflows (part of a master clearing agreement), net of Level 1/2A/2B guarantees delivered: 100% | 100% | 100% | 100%

Market value of Level 1, 2A, and 2B guarantees delivered
- Sum for derivatives with master clearing agreement: (presented as sum)
- Sum for derivatives without master clearing agreement: (presented as sum)
- Due to rating deterioration (increase in liquidity needs related to derivatives and financing as a result of decline in institution credit rating): 100% | 100% | 100% | 100%

Market value of guarantees provided in derivatives and other transactions
- Level 1 guarantees provided in derivatives transactions: 0% | 10% | 10% | 10%
- Level 1 guarantees provided in other transactions: 0% | 10% | 10% | 10%
- Level 2A guarantees provided in derivatives transactions: 20% | 35% | 35% | 35%
- Level 2A guarantees provided in other transactions: 20% | 35% | 35% | 35%
- Level 2B guarantees provided in derivatives transactions: 20% | 35% | 35% | 35%
- Level 2B guarantees provided in other transactions: 20% | 35% | 35% | 35%
- Guarantees other than Level 1/2A/2B provided in derivatives and other transactions: 20% | 35% | 35% | 35%
- Due to unsegregated surplus guarantees held by the institution that could contractually be demanded by the counterparty: 100% | 100% | 100% | 100%
- Due to a shortfall in guarantees: 100% | 100% | 100% | 100%
- Due to substitution of guarantees: 100% | 100% | 100% | 100%

Participation in structured vehicles (administrator, originator, issuer, or provider of implicit/explicit support)
- Liabilities generated by securitizations and any other structured product issued by the institution: 100% | 100% | 100% | 100%
- Contingent liabilities associated with securitizations and special-purpose vehicles with initial maturity ≤ one year: 100% | 100% | 100% | 100%

### Credit commitments: credit lines and liquidity lines (run-off rates)
Irrevocable credit lines granted to:
- Individuals and SMEs: 5% | 15% | 30% | 30%
- Corporates, sovereigns, central banks, public sector entities: 10% | 20% | 40% | 40%
- Commercial banks: 40% | 60% | 100% | 100%
- Other financial entities not included in other categories: 40% | 75% | 100% | 100%

Revocable credit lines granted to:
- Individuals and SMEs: 5% | 15% | 10% | 15%
- Corporates, sovereigns, central banks, public sector entities: 5% | 10% | 20% | 20%
- Commercial banks: 10% | 20% | 40% | 40%
- Other financial entities not included in other categories: 10% | 20% | 40% | 40%

Liquidity lines provided to:
- Individuals and SMEs: 5% | 50% | 50% | 50%
- Corporates, sovereigns, central banks, public sector entities: 30% | 60% | 80% | 80%
- Commercial banks: 40% | 100% | 100% | 100%
- Other financial entities not included in other categories: 100% | 100% | 100% | 100%

Other instruments and outflows
- Guarantees by endorsement [avales] provided: 30% | 40% | 50% | 50%
- Letters of credit: 0% | 10% | 10% | 10%
- Other international trade instruments: 0% | 10% | 10% | 10%
- Other cash outflows not included in other categories:
  - Contractual: 100% | 100% | 100% | 100%
  - Noncontractual: 100% | 100% | 100% | 100%

*Source: Appendix V. LCR-Based Stress Scenario Parameters.*

### Appendix VI. Cash Flow Analysis Scenario Parameters

### Appendix VI. Cash Flow Analysis Scenario Parameters

### Outflows parameters (liabilities and secured lending liabilities)
- Liabilities resulting from securities issued (if not treated as retail deposits) — Outflows:
  - Unsecured bonds due: Value 0, Min Value 1, Max Value 1
  - Regulated covered bonds: Value 0, Min Value 1, Max Value 1
  - Securitizations due: Value 0, Min Value 1, Max Value 1
  - Other: Value 0, Min Value 1, Max Value 1
- Liabilities resulting from secured lending and capital market driven transactions collateralized by Level 1 tradable assets — Outflows:
  - Level 1 excluding covered bonds: Value 0.1, Min Value 0.3, Max Value 1.02
  - Level 1 (CQS 1): Value 0, Min Value 0.3, Max Value 1.02
  - Level 1 (CQS2, CQS3): Value 0.1, Min Value 0.5, Max Value 1.02
  - Level 1 (CQS4+): Value 0.2, Min Value 0.5, Max Value 1.02
  - Level 1 covered bonds (CQS1): Value 0.2, Min Value 0.5, Max Value 1.02
- Level 2A tradable assets — Outflows:
  - Level 2A tradable assets: Value 0.2, Min Value 0.5, Max Value 1.02
  - Level 2A covered bonds (CQS1, CQS2): Value 0.2, Min Value 0.5, Max Value 1.02
  - Level 2A public sector (CQS1, CQS2): Value 0.2, Min Value 0.5, Max Value 1.02
- Level 2B tradable assets — Outflows:
  - Level 2B ABS (CQS1): Value 0.2, Min Value 0.5, Max Value 1.05
  - Level 2B covered bonds (CQS1-6): Value 0.2, Min Value 0.5, Max Value 1.05
  - Level 2B: corporate bonds (CQ1-3): Value 0.2, Min Value 0.5, Max Value 1.05
  - Level 2B shares: Value 0.35, Min Value 1, Max Value 1.5
  - Level 2B public sector (CQS 3-5): Value 0.35, Min Value 1, Max Value 1.5
- Other assets/tradable assets — Outflows:
  - Other tradable assets: Value 0.35, Min Value 1, Max Value 1.5
  - Other assets: Value 0.35, Min Value 1, Max Value 1.5
- Liabilities not reported in 1.2, resulting from deposits received (excluding deposits received as collateral) — Outflows:
  - Stable retail deposits: Value 0.05, Min Value 0.1, Max Value 1
  - Other retail deposits: Value 0.1, Min Value 0.2, Max Value 1
  - Operational deposits: Value 0.05, Min Value 0.25, Max Value 1
  - Non-operational deposits from credit institutions: Value 0.2, Min Value 1, Max Value 1
  - Non-operational deposits from other financial customers: Value 0.2, Min Value 1, Max Value 1
  - Non-operational deposits from central banks: Value 0, Min Value 0.25, Max Value 1
  - Non-operational deposits from non-financial corporates: Value 0.2, Min Value 0.4, Max Value 1
  - Non-operational deposits from other counterparties: Value 0.2, Min Value 0.4, Max Value 1
- Other specific outflows — Outflows:
  - FX-swaps maturing: Value 0, Min Value 0, Max Value 1
  - Derivatives amount payables other than those reported in 1.4: Value 0, Min Value 0, Max Value 1
  - Other outflows: Value 0, Min Value 0, Max Value 1
- Total outflows — Outflows: (Value fields listed above combine into Total outflows)

### Inflows parameters (secured lending and loans and advances)
- Monies due from secured lending and capital market driven transactions collateralized by Level 1 tradable assets — Inflows:
  - Level 1 excluding covered bonds: Value 0.1, Min Value 0.3, Max Value 1.02
  - Level 1 (CQS 1): Value 0, Min Value 0.3, Max Value 1.02
  - Level 1 (CQS2, CQS3): Value 0.1, Min Value 0.5, Max Value 1.02
  - Level 1 (CQS4+): Value 0.2, Min Value 0.5, Max Value 1.02
  - Level 1 covered bonds (CQS1): Value 0.2, Min Value 0.5, Max Value 1.02
- Level 2A tradable assets — Inflows:
  - Level 2A tradable assets: Value 0.2, Min Value 0.5, Max Value 1.05
  - Level 2A covered bonds (CQS1, CQS2): Value 0.2, Min Value 0.5, Max Value 1.05
  - Level 2A public sector (CQS1, CQS2): Value 0.2, Min Value 0.5, Max Value 1.05
- Level 2B tradable assets — Inflows:
  - Level 2B ABS (CQS1): Value 0.2, Min Value 0.5, Max Value 1.05
  - Level 2B covered bonds (CQS1-6): Value 0.2, Min Value 0.5, Max Value 1.05
  - Level 2B: corporate bonds (CQ1-3): Value 0.2, Min Value 0.5, Max Value 1.05
  - Level 2B shares: Value 0.35, Min Value 1, Max Value 1.5
  - Level 2B public sector (CQS 3-5): Value 0.35, Min Value 1, Max Value 1.5
  - Other tradable assets: Value 0.35, Min Value 1, Max Value 1.5
  - Other assets: Value 0.35, Min Value 1, Max Value 1.5
- Monies due not reported in 2.1 resulting from loans and advances granted to — Inflows:
  - Retail customers: Value 0, Min Value 1, Max Value 1
  - Non-financial corporates: Value 0, Min Value 1, Max Value 1
  - Credit institutions: Value 0, Min Value 1, Max Value 1
  - Other financial customers: Value 0, Min Value 1, Max Value 1
  - Central banks: Value 0, Min Value 1, Max Value 1
  - Other counterparties: Value 0, Min Value 1, Max Value 1
- Other inflows — Inflows:
  - FX-swaps maturing: Value 0, Min Value 1, Max Value 1
  - Derivatives amount receivables other than those reported in 2.3: Value 0, Min Value 1, Max Value 1
  - Paper in own portfolio maturing: Value 0, Min Value 1, Max Value 1
  - Other inflows: Value 0, Min Value 1, Max Value 1

### Withdrawable central bank reserves (CBL) and eligible collateral parameters
- Withdrawable central bank reserves — CBL: Value 0, Min Value 0, Max Value 1
- Level 1 tradable assets — CBL:
  - Level 1 excluding covered bonds: Value 0, Min Value 0.1, Max Value 1
  - Level 1 (CQS 1): Value 0, Min Value 0.1, Max Value 1
  - Level 1 (CQS2, CQS3): Value 0, Min Value 0.1, Max Value 1
  - Level 1 (CQS4+): Value 0, Min Value 0.1, Max Value 1
  - Level 1 covered bonds (CQS1): Value 0, Min Value 0.2, Max Value 1
- Level 2A tradable assets — CBL:
  - Level 2A tradable assets: Value 0.05, Min Value 0.2, Max Value 1
  - Level 2A corporate bonds (CQS1): Value 0.05, Min Value 0.2, Max Value 1
  - Level 2A public sector (CQS1, CQS2): Value 0.05, Min Value 0.2, Max Value 1
- Level 2B tradable assets — CBL:
  - Level 2B tradable assets: Value 0.1, Min Value 0.2, Max Value 1
  - Level 2B ABS (CQS1): Value 0.1, Min Value 0.2, Max Value 1
  - Level 2B corporate bonds (CQ1-3): Value 0.1, Min Value 0.2, Max Value 1
  - Level 2B shares: Value 0.1, Min Value 0.2, Max Value 1
  - Level 2B public sector (CQS 3-5): Value 0.1, Min Value 0.2, Max Value 1
  - Other tradable assets: Value 0.1, Min Value 0.2, Max Value 1
- Central government and other eligible assets — CBL:
  - Central government (CQS 2 & 3): Value 0, Min Value 0.2, Max Value 1
  - Shares: Value 0, Min Value 0.2, Max Value 1
  - Covered bonds: Value 0, Min Value 0.2, Max Value 1
  - ABS: Value 0, Min Value 0.2, Max Value 1
  - Other tradable assets: Value 0, Min Value 0.2, Max Value 1
  - Non tradable assets eligible for central banks: Value 0, Min Value 0.2, Max Value 1

### Undrawn committed facilities, facilities by level, and intragroup/other counterparties
- Undrawn committed facilities — CBL:
  - Level 1 facilities: Value 0.8, Min Value 1, Max Value 1
  - Level 2B restricted use facilities: Value 0.8, Min Value 1, Max Value 1
  - Level 2B IPS facilities: Value 0.8, Min Value 1, Max Value 1
  - From intragroup counterparties: Value 1, Min Value 1, Max Value 1
  - From other counterparties: Value 1, Min Value 1, Max Value 1

### Committed credit facilities, contingencies, and outflows due to downgrade triggers
- Outflows from committed facilities — Contingencies:
  - Committed credit facilities considered as Level 2B by the receiver: Value 0.15, Min Value 0.3, Max Value 1
  - Other: Value 0.15, Min Value 0.4, Max Value 1
  - Liquidity facilities: Value 0.5, Min Value 1, Max Value 1
  - Outflows due to downgrade triggers: Value 0.5, Min Value 1, Max Value 1

*Appendix VI. Cash Flow Analysis Scenario Parameters — IMF content unit*

---


_Source: https://www.imf.org/-/media/files/publications/cr/2022/english/1mexea2022007.pdf_