## 1bwaea2024005 - EXECUTIVE SUMMARY

## Source details

**Canonical URL:** [1bwaea2024005 - EXECUTIVE SUMMARY](https://www.imf.org/-/media/files/publications/cr/2024/english/1bwaea2024005.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/cr/2024/english/1bwaea2024005.pdf.md)
- [Structured JSON version](/-/media/files/publications/cr/2024/english/1bwaea2024005.pdf.json)

---

### Background
- Botswana is a small, open economy with a highly concentrated financial sector comprising banks and sizeable non-bank financial institutions (NBFIs).
- The financial sector accounts for close to 130 percent of GDP.
- Banking sector structure:
  - Nine commercial banks; three largest banks account for 64 percent of banking sector assets.
  - Two D-SIBs account for 46 percent of banking sector assets.
  - Domestic ownership concentrated: largest pension fund holds 22 percent of bank shares.
- NBFIs and retirement funds:
  - Retirement funds account for 43 percent of financial system assets.
  - Retirement funds account for over 90 percent of NBFIs’ total assets as of June 2022.
- Recent dynamics:
  - Credit increased by 5.4 percent (y-o-y, Q2 2022).
  - BoB increased its monetary policy rate (MoPR) by a combined 151 basis points since April 2022.
  - Medium-term inflation objective: 3–6 percent.

### Financial sector vulnerabilities and risks
- Overall: financial institutions hold adequate capital and liquidity and show moderate profitability, but interconnectedness between banks and NBFIs and large exposures to unsecured household debt increase vulnerability.
- Three main systemic risks identified:
  - Geo-political developments slowing global growth and reducing diamond demand.
  - Sustained food and energy cost pressures pushing inflation.
  - Tightening global financial conditions as major economies increase policy rates leading to tighter domestic financial conditions.
- Additional concerns:
  - Recent regulatory changes for retirement funds could impact financial stability.
  - Banks’ balance sheets reflect a high concentration of lumpy short-term deposits from the non-bank sector, including NBFIs and corporates.

### Scope and approach of the assessment
- Institutional perimeter: eight commercial banks as of June 2022 (BBS Bank Limited licensed October 6, 2022 excluded).
- Stress-testing exercises:
  - Top-down solvency stress test (static balance sheet, 2023–2025).
  - Liquidity stress test (LCR-proxy 30-day and sensitivity analysis).
  - Contagion and interconnectedness stress test focusing on domestic banking interconnectedness.
- Scenarios include baseline and adverse macrofinancial scenarios incorporating global and domestic inflationary pressures, monetary policy tightness, and a major slowdown of economic activity.

### Stress test results — Solvency
- Aggregate capital depletion in the adverse scenario is relatively small: less than 0.02 percent of GDP.
- Solvency stress tests identify small capital shortfalls in two banks under the adverse scenario.
- System-level resilience drivers:
  - Net interest income (NII) contributes positively to capital under both baseline and adverse scenarios due to shorter maturities, higher share of variable-rate assets, and greater pass-through from policy rate on assets versus liabilities.
  - Market risk is low given limited securities exposures.
- D-SIBs:
  - Two D-SIBs represent 44 percent of system assets and remain resilient under both scenarios.
  - D-SIBs’ initial capital ratios: CAR 20 percent and Tier 1 13.6 percent.
  - Average dividend payout ratios: two D-SIBs 65.5 percent vs system average 31.2 percent.

### Stress test results — Liquidity (LCR-proxy)
- Baseline (narrow HQLA): banking sector liquidity is elevated; all banks comply with prescribed liquidity ratios with sufficient buffers.
- Adverse liquidity scenario:
  - Five banks would face a liquidity shortfall due to susceptibility to short-term wholesale funding sources (large deposits from NBFIs comprising insurance companies and retirement funds).
  - Expanding eligible liquid assets to include required reserves and domestic long-term government bonds (with appropriate haircuts) reduces banks facing shortfall to two.
- HQLA composition and concentration:
  - Peer 1 banks hold 71 percent of total HQLA; Peer 3 banks hold 8.5 percent.
  - Level 1 assets comprise 94 percent of total HQLA (broad definition); Level 2 assets comprise 6 percent.
- LCR stress parameters and outcomes:
  - Baseline total outflows = 21 percent of total banking sector assets (BWP 24 billion).
  - Adverse scenario anchors outflows/total assets to 26 percent and inflows/total assets to 9 percent; haircuts increased (e.g., 5 percent for government bonds <1 year; 15 percent for >1 year).
  - Under adverse (narrow HQLA): LCR ratios of four banks dropped by more than 50 percent; five banks fell below 100 percent threshold.
  - Expanding HQLA coverage (include required reserves and government bonds >1 year with 15 percent haircuts) improved resilience; only two banks remained below but close to 100 percent.

### NBFI market risk assessment
- Stress test of largest NBFIs to market risks under an adverse scenario indicates that the impact on NBFIs’ capital is limited.
- NBFIs’ cross-border exposures mainly reflect pension funds’ offshore investment, which account over 60 percent of pension funds’ total exposure portfolio.

### Data gaps identified
- Credit risk modelling:
  - Collect NPL inflows and outflows data.
  - Collect PD and LGD data from commercial banks by economic sectors.
- Market risk modelling:
  - Collect specific duration of securities for risk monitoring and management purposes.
- Interconnectedness analysis:
  - NBFIRA should add reporting requirements for bilateral exposures between banks and all NBFIs on a regular basis.

### Recommendations — Supervision, stress testing, and IRRBB
- BoB should introduce macro and micro-level stress tests based on multi-period scenario analysis and develop its framework to assess interest rate risk in the banking book (IRRBB).
- BoB should integrate outputs from ongoing IMF technical assistance within its macroprudential stress testing framework to challenge banks’ microprudential stress tests and validate IRRBB assessments.
- Develop supervisory methodologies for assessment of banks’ exposure to IRRBB and the potential impact on banks’ capital to better inform supervision and strengthen the Supervisory Review and Evaluation Process (SREP).
- Develop tools to validate banks' internal IRRBB measurements and internal limits; expand supervisory data on RSA and RSL by repricing maturity supported by analytical tools.

### Recommendations — Liquidity framework
- Prioritize planned transition to Basel III liquidity monitoring and assessment framework (LCR and NSFR).
- Expand qualifying liquid assets:
  - Include required reserves in the near term.
  - Include domestic long-term government bonds (maturity longer than one year) in the medium term with appropriate haircuts.
- Revise statutory reporting to improve granularity and quality to calibrate Basel III liquidity indicators.
- Calibrate regulatory weights on assets and funding structures and liquidity characteristics of Botswana’s banking system.
- Establish minimum requirements for Basel III liquidity standards once validation of parameters is conducted and material experience with Basel III supervisory monitoring is obtained.

### Recommendations — Capital buffers and macroprudential tools
- Implement additional capital buffers: impose tailored Pillar II capital buffer requirements for banks with low capital and high dividend payout ratios.
- Consider extending macroprudential tools such as debt-service-to-income (DSTI) ratio limits for households, or countercyclical capital buffers, to address underlying credit risks from loan concentrations.
- BoB should extend capacity with macro-level stress tests with multi-period scenarios and sensitivity assessments; continue efforts to fill data gaps to support calibration of macroprudential tools.

### Banking business model, credit composition, and asset quality
- Business model: intermediation of domestic deposits to provide private sector credit; assets largely denominated in Pula.
- Exposures and shares:
  - Bank loans to households = 27 percent of GDP; corporate loans = 10 percent of GDP.
  - Household loans accounted for around 41 percent of total commercial bank assets as of end-June 2022.
  - Sector average loan-to-deposit ratio: 85 percent (average).
- Credit risk metrics:
  - RWAs of credit risk account for 89 percent of total RWAs as of June 2022.
  - Loans comprise 83 percent of total assets.
  - Household loan composition: personal loans 70 percent, mortgages 23 percent, others remainder.
  - NPLs to total loans: 3.7 percent as of June 2022 (compared with 4.3 percent in 2020).
  - Total past-due loans: declined from 6.3 percent in 2020 to 5.4 percent as of June 2022.
  - Non-financial corporate loans account for over 30 percent of past-due loans.
  - Sectors with NPLs over 20 percent as of June 2022: construction, transportation, and business service sectors.
- Bank heterogeneity: widely differing household NPL ratios across banks indicate pockets of risk despite low system-wide levels.

### Household indebtedness (selected statistics and risks)
- End-2021 total household debt: P58.5 billion (30 percent of GDP), composition:
  - bank loans: 88.5 percent
  - micro-lender loans: 11 percent
  - hire purchase credit: 0.5 percent
- Collection features: for more than 70 percent of household loans in some banks, instalments deducted directly from salaries.
- Loan characteristics: majority medium-term (2 to 5 year), unsecured variable rate loans.
- Approximately 10 percent of household debt in 2021 is outside commercial banking system (micro-lenders).
- Household debt-service-to-income ratios for 2021:
  - banks: ranged between 40 and 70 percent (average 57 percent)
  - microlenders: ranged between 30 and 67 percent (average 44.5 percent)
- Assessment: household debt does not appear to represent an immediate system-wide risk given relatively low levels of indebtedness and NPL ratios, low and stable bank credit growth, and ample liquidity; pockets of risk exist due to unequal distribution by income, differential NPLs across banks, lower regulatory standards for micro-lenders, and growing lending to privately-employed individuals.

### Interest rate risk and pass-through
- Changes in MoPR have a one-to-one effect on prime lending rates.
- Approximately 66 percent of loans were priced at margins ranging from 2–10 percentage points above the prime rate as of June 2022.
- BoB set a spread of 411 basis points between the MoPR and the prime lending rate for a period of one year starting in April 2022; commercial banks can set final lending rates adjusted for customers’ risk profile.
- Unsecured household credit mainly variable rate; many loans susceptible to increases in lending rates when the BoB-imposed household loan interest rate reset moratorium expires in April 2023.
- A one-to-one pass-through implies increases in MoPR may increase loan rates and possibly push up NPLs.

### Interest rate risk in the banking book (IRRBB) and funding vulnerabilities
- Over 65 percent of bank assets are floating rate.
- Approximately 35 percent of liabilities are fixed rate and are subject to administered rates that typically lag policy rate changes.
- Bank assets have significantly shorter maturities than liabilities; rising rate environment may produce positive IRRBB exposures and a spike in NII.
- Deposits of NBFIs account for 23 percent of banking sector deposits and comprise sizeable deposits from a few large depositors.
- Largest banks hold close to 68 percent of total household and corporate call and savings deposits; smaller banks rely on price-sensitive fixed-deposits or interbank funding.

### Solvency stress test methodology (core assumptions)
- Static balance sheet assumption: assets and liabilities remain constant over the three-year stress test horizon (2023 to 2025).
- Capital ratios projected using scenario-conditional paths for capital and RWAs.
- Satellite models link macro scenarios to key bank variables via credit risk and P&L components.
- Credit risk PDs proxied using projected NPL ratio from satellite models; LGD paths based on bank-specific historical coverage ratios.
- Historical estimation periods:
  - Bank-specific PD/LGD estimation: 2012Q1 to 2022Q2.
  - Bank-level NPL panel model: quarterly data from 2000Q4 to 2022Q2.
- Market risk treatment:
  - AfS and HfT: mark-to-market; modified duration formula used to revalue exposures as function of residual duration and scenario bond yields.
  - HtM: credit risk approach with provisions to cover expected loss.

### Solvency stress test — selected quantitative results (table summaries preserved)
- Baseline scenario (capital ratios, percent):
  - Total capital adequacy ratio: 18.5, 19.6, 20.8, 22.1 (2022–2025)
  - Tier 1 capital ratio: 12.7, 13.8, 15.0, 16.2
  - Common/Core tier 1 ratio: 12.2, 13.3, 14.4, 15.7
- Baseline aggregated figures (In Pula millions):
  - Total Capital in the banking system: 14,344; 15,153; 15,980; 16,889 (2022–2025)
  - Sum of Risk Weighted Assets: 77,444; 77,200; 76,893; 76,602
  - Net Interest Income: 4,483; 4,482; 4,904; 4,663
  - Loss provisions (Credit Risk): -33; -487; -614; -582
- Adverse scenario (capital ratios, percent):
  - Total capital adequacy ratio: 18.5, 19.5, 19.9, 22.0 (2022–2025)
  - Tier 1 capital ratio: 12.7, 13.7, 14.0, 16.1
  - Common/Core tier 1 ratio: 12.2, 13.2, 13.5, 15.6
- Adverse aggregated figures (In Pula millions):
  - Number of banks failing the tests total: 0, 2, 2 (2022–2025)
  - Necessary Recapitalization (absolute): 0, 163.9, 163.8
  - Necessary Recapitalization (percent of GDP) — CAR: 0, 0.009, 0.013
  - Total Capital in the banking system: 14,344; 15,067; 15,107; 16,649
  - Sum of Risk Weighted Assets: 77,444; 77,123; 75,878; 75,599
  - Net Interest Income: 4,483; 4,506; 5,014; 5,642
  - Loss provisions (Credit Risk): -33; -640; -2,489; -560

### Sensitivity analyses — lending pass-through and household concentration
- Lending pass-through scenarios (additional to adverse):
  - Alternative pass-through assumptions: (1) model-projected effective rates on interest-earning assets; (2) lending pass-through equal to 0.
  - Results: CAR could drop up to 6 percentage point to 16 percent in 2025 under low pass-through; even with pass-through = 0, aggregate CAR remains above regulatory thresholds.
  - With pass-through = 0, three banks would face challenges meeting minimum capital requirements by 2025; aggregate recapitalization needs reach 0.4 percent of GDP.
- Household concentration single-factor shocks (10 percent and 20 percent of household performing loans become non-performing; provisioning excluded):
  - Baseline household NPL ratios (percent): Total 3.2; Peer 1 3.6; Peer 2 2.0; Peer 3 4.8.
  - Current Capital Ratios (percent): Total Capital — Total 18.5; Peer 1 18.9; Peer 2 18.2; Peer 3 17.1. Tier 1 — Total 12.7; Peer 1 13.1; Peer 2 13.3; Peer 3 10.4. Core Tier 1 — Total 12.2; Peer 1 13.1; Peer 2 10.7; Peer 3 10.4.
  - Outcomes (selected):
    - 10 percent shock — Total Capital: Total 15.1; Peer 1 15.9; Peer 2 11.7*; Peer 3 15.8. Tier 1: Total 9.3; Peer 1 10.1; Peer 2 6.7*; Peer 3 9.1. Core Tier 1: Total 8.9; Peer 1 10.1; Peer 2 5.0*; Peer 3 9.1.
    - 20 percent shock — Total Capital: Total 11.6*; Peer 1 12.9; Peer 2 5.1*; Peer 3 14.5. Tier 1: Total 6.3*; Peer 1 7.1*; Peer 2 2.7*; Peer 3 7.8. Core Tier 1: Total 6.3*; Peer 1 7.1*; Peer 2 2.7*; Peer 3 7.8.
  - Note: asterisks indicate numbers below the minimum requirements.
  - Interpretation: 10 percent household shock leaves system largely resilient though one bank has total capital shortfall; 20 percent shock causes several banks significant capital shortfalls and aggregated CAR could fall below minimum requirements, though most banks retain strong total capital buffers.

### Contagion and interconnectedness
- Exposures and funding:
  - Banking sector largest exposure (48.67 percent) to households through unsecured loans.
  - Major funding source: non-financial corporations comprising 49.76 percent of total banking deposit by May 2022.
  - Interbank market dominated by Peer 1 banks as domestic lending source.
- Contagion simulation assumptions:
  - LGD set to 85 percent for credit shocks.
  - When funding and credit shocks occur simultaneously: borrower banks can replace only 50 percent of previously borrowed funds and assets are fire-sold at a 35 percent discount.
- Simulation outcomes:
  - No single failure of a bank would trigger failure of another bank under simulated shocks.
  - Interbank positions are small compared to banks’ capital.
  - Peer 1 banks are the most contagious group; failure of any Peer 1 bank could cause up to 12.7 percent potential capital reduction for a counterparty bank.
  - Simultaneous credit and funding shocks with failure of all Peer 1 banks would generate total capital loss of 12 percent for the remaining system.
  - Peer 3 banks are among most vulnerable, experiencing capital loss of 11 percent from other banks' defaults.

### Macroprudential policy tools and governance
- Toolkit observations:
  - Botswana’s macroprudential toolkit broadly comparable to other EMDEs but smaller than advanced economies.
  - Existing tools include PRR (currently at 2.5 percent) and LAR (10 percent); LCR and NSFR not yet implemented as binding.
  - D-SIB framework: two banks designated D-SIBs since 2021; D-SIBs cover about 90 percent of banking system by assets (note: two D-SIBs account for 46 percent of banking assets as of June 2022).
- Recommended toolkit extensions:
  - Additional capital buffers: counter-cyclical capital buffers or systemic risk buffers.
  - Basel III liquidity tools: LCR and NSFR.
  - Household-sector measures: caps on DSTI ratios.
- Institutional arrangements:
  - NBFIRA Act, 2022; Financial Intelligence Act, 2022; Financial Stability Council (FSC) statutory committee; BoB primary macroprudential authority for banks.

### Key policy recommendations (selected and prioritized)
- Standardize and enhance reporting frameworks for banks and NBFIs; improve data management to support stress testing and interconnectedness analysis.
- Conduct macro- and micro-prudential stress tests based on multi-period scenario analysis and sensitivity assessments.
- Develop supervisory methodologies for assessing IRRBB and potential impact on capital to strengthen SREP.
- Expand statutory liquid assets to include required reserves in the short term and domestic long-term government securities in the medium term (with appropriate haircuts).
- Revise statutory reporting to improve granularity for Basel III liquidity indicator calibration.
- Calibrate regulatory weights on assets and funding structures and liquidity characteristics tailored to Botswana’s banking system.
- Establish minimum requirements for Basel III liquidity standards after parameter validation and supervisory monitoring experience.
- Impose tailored Pillar II capital buffer requirements for banks with low capital and high dividend payout ratios.
- Consider macroprudential measures to address household credit concentration (e.g., DSTI caps, countercyclical buffers).

*Source: IMF staff analysis, Botswana Financial Sector Assessment (1bwaea2024005).*

### EXECUTIVE SUMMARY __________________________________________________________________________ 5

### 1bwaea2024005 - EXECUTIVE SUMMARY __________________________________________________________________________ 5

### BACKGROUND
- Botswana is a small, open economy with a highly concentrated financial sector comprising banks and sizeable non-bank financial institutions (NBFIs).
- The financial sector accounts for close to 130 percent of GDP.
- The banking sector comprises nine commercial banks, with the three largest banks accounting for 64 percent of banking sector assets; two are D-SIBs that account for 46 percent of banking sector assets.
- Retirement funds account for 43 percent of financial system assets; retirement funds account for over 90 percent of NBFIs’ total assets as of June 2022.
- Domestic ownership of banks is mainly through the largest pension fund that holds 22 percent of bank shares.
- Credit increased by 5.4 percent (y-o-y, Q2 2022).
- The BoB increased its monetary policy rate (MoPR) by a combined 151 basis points since April 2022.
- Medium-term inflation objective: 3–6 percent.

### FINANCIAL SECTOR VULNERABILITIES AND RISKS
- Financial institutions hold adequate capital and liquidity and show moderate profitability, but interconnectedness between banks and NBFIs and large exposures to unsecured household debt increase vulnerability.
- Three main systemic risks identified:
  - Geo-political developments may slow global growth and reduce diamond demand, adversely impacting economic performance in Botswana.
  - Sustained food and energy cost pressures could further push inflation.
  - Tightening of global financial conditions as major economies increase policy rates could lead to tighter domestic financial conditions.
- Recent regulatory changes for retirement funds could also impact financial stability.
- Banks’ balance sheets reflect a high concentration of lumpy short-term deposits from the non-bank sector, including NBFIs and corporates.

### SCOPE AND APPROACH OF THE ASSESSMENT
- Systemic risk assessment based on stress tests simulating severe yet plausible adverse scenarios including global and domestic inflationary pressures, monetary policy tightness, and a major slowdown of economic activity.
- Exercises covered eight commercial banks as of June 2022 (the BBS Bank Limited, licensed October 6, 2022, is excluded).
- Three types of stress test exercises performed:
  - Top-down solvency stress test.
  - Liquidity stress test (LCR-proxy and sensitivity analysis).
  - Contagion and interconnectedness stress test focusing on domestic banking interconnectedness.

### STRESS TEST RESULTS — SOLVENCY
- Solvency stress tests identify small capital shortfalls in two banks under the adverse scenario.
- Overall financial system appears resilient to a wide range of shocks.

### STRESS TEST RESULTS — LIQUIDITY
- Banking sector liquidity is elevated and allows all banks to comply with prescribed liquidity ratios with sufficient buffers in a baseline scenario.
- Under an adverse liquidity scenario, five banks would face a liquidity shortfall due to susceptibility to short-term wholesale funding sources (large deposits from NBFIs comprising insurance companies and retirement funds).
- Expanding the class of eligible liquid assets to be consistent with the Basel III HQLA definition would reduce the number of banks facing liquidity shortfall to two.

### NBFI MARKET RISK ASSESSMENT
- A stress test of the largest NBFIs to market risks under an adverse scenario indicates that the impact on NBFIs’ capital is limited.

### DATA GAPS IDENTIFIED
- For credit risk modelling: collect NPL inflows and outflows data; collect probability of default (PD) and loss given default (LGD) data from commercial banks by economic sectors.
- For market risk modelling: collect the specific duration of securities for risk monitoring and management purposes.
- For interconnectedness analysis: NBFIRA should add reporting requirements for bilateral exposures between banks and all NBFIs on a regular basis.

### RECOMMENDATIONS — SUPERVISION, STRESS TESTING, AND IRRBB
- BoB should introduce macro and micro level stress tests based on a multi-period scenario analysis and develop its framework to assess interest rate risk in the banking book (IRRBB).
- BoB should integrate outputs from ongoing IMF technical assistance within its macroprudential stress testing framework to challenge banks’ microprudential stress tests and validate IRRBB assessments.
- Develop supervisory methodologies for assessment of banks’ exposure to IRRBB and the potential impact on banks’ capital to better inform supervision and strengthen the Supervisory Review and Evaluation Process (SREP).

### RECOMMENDATIONS — LIQUIDITY FRAMEWORK
- Prioritize the planned transition to the Basel III liquidity monitoring and assessment framework (LCR and NSFR).
- Expand qualifying liquid assets by including required reserves in the near term and long-term government bonds with appropriate haircuts in the medium term.
- Revise statutory reporting to improve granularity and quality to calibrate Basel III liquidity indicators.
- Calibrate regulatory weights on assets and funding structures and liquidity characteristics of Botswana’s banking system.
- Establish minimum requirements for Basel III liquidity standards once validation of parameters is conducted and material experience with Basel III supervisory monitoring is obtained.

### RECOMMENDATIONS — CAPITAL BUFFERS AND MACROPRUDENTIAL TOOLS
- Implement additional capital buffers: impose tailored Pillar II capital buffer requirements for banks with low capital and high dividend payout ratios.
- Consider extending use of macroprudential tools such as debt-service to income ratio limits for households, or countercyclical capital buffers, to address underlying credit risks from loan concentrations.
- BoB should extend capacity with macro-level stress tests with multi-period scenarios and sensitivity assessments; continue efforts to fill data gaps to support appropriate calibration of macroprudential tools.

### KEY RECOMMENDATIONS TABLE (as summarized)
- 1. Standardize the reporting framework for banks and improve the data management systems to support stress testing and interconnectedness analysis [¶29,¶51]. — BOB, NBFIRA — ST
- 2. Conduct stress test at both the macro- and micro-prudential levels based on multi-period scenario analysis and sensitivity assessments to enhance supervisory oversight [¶30]. — BOB — ST–MT
- 3. Develop supervisory methodologies for assessment of banks’ exposure to IRRBB and the potential impact on capital to better inform supervision and strengthen SREP [¶31]. — BOB — MT–LT
- 4. Expand the use of macroprudential tools such as, debt-service to income ratio limits for households, or countercyclical capital buffers, to address underlying credit risks from loan concentrations [¶32]. — BoB — MT
- 5. Extend the coverage of statutory liquid assets to:
  - Include required reserves in the short-term [¶43]
  - Include domestic long-term government securities (maturity longer than one year) in the medium-term [¶44-46] — BoB — ST
- 6. Revise the statutory reporting framework to improve banks’ reporting with the level of granularity and quality to calibrate Basel III liquidity indicators [¶46]. — BoB — ST
- 7. Calibrate regulatory weights on the assets and funding structure and liquidity characteristic of Botswana’s banking system [¶46]. — BoB — MT
- 8. Establish minimum requirements for Basel III liquidity standards once validation of parameters is conducted and material experience with Basel III supervisory monitoring is obtained [46]. — BoB — MT–LT
- 9. Impose tailored Pillar II capital buffer requirements for banks with low capital and high dividends payout ratios [¶52]. — BOB — MT

*Prepared by Dan Cheng, Yuan Gao Rollinson, and Ian Stuart. The FSAP team expresses gratitude to the authorities for their cooperation in facilitating this exercise.*

### 4.      The accommodative monetary stance continues to be conducive to credit growth

### 4.      The accommodative monetary stance continues to be conducive to credit growth

### Banking business model and credit composition
- Banking business model centered on intermediation of domestic deposits to provide private sector credit.
- Assets largely denominated in Pula with largest exposures to unsecured household loans and to small- and medium-size corporates in services and other non-mining sectors.
- Bank loans to households account for 27 percent of GDP and corporate loans account for 10 percent of GDP.
- Household loans accounted for around 41 percent of total commercial bank assets as of end-June 2022.
- Sector average loan-to-deposit ratio: 85 percent (average).

### Credit risk and asset quality
- Credit risk is the largest risk in Botswana’s banking system: RWAs of credit risk account for 89 percent of total RWAs as of June 2022.
- Loans comprise the largest part of total assets: 83 percent.
- By sector, loans are mostly concentrated in households, followed by real sector and public non-financial sector.
- Household loan composition: personal loans 70 percent, mortgages 23 percent, others remainder.
- NPLs to total loans: 3.7 percent as of June 2022 (compared with 4.3 percent in 2020).
- Total past-due loans declined from 6.3 percent of total loans and advances in 2020 to 5.4 percent of total loans as of June 2022, due largely to loan recoveries and write-offs.
- Non-financial corporate loans account for over 30 percent of past-due loans.
- Sectors with NPLs over 20 percent as of June 2022: construction, transportation, and business service sectors.
- Banks have widely differing household NPL ratios, indicating pockets of risk despite low system-wide levels.

### Household indebtedness (Box 1) — structure and risks
- In 2021 commercial banks (accounting for 40 percent of total financial sector assets) lent primarily to households, which accounted for 66 percent of bank lending.
- At end-2021, total household debt amounted to P58.5 billion (30 percent of GDP), comprising:
  - bank loans: 88.5 percent
  - micro-lender loans: 11 percent
  - hire purchase credit: 0.5 percent
- For a large share of household loans (more than 70 percent in the case of some banks), lenders can deduct instalments directly from borrowers’ salaries.
- Share of household debt held by higher income earners (>P20 000 per month) has been increasing.
- Majority of household debt is medium-term (2 to 5 year), unsecured variable rate loans.
- Approximately 10 percent of household debt in 2021 is outside the commercial banking system (micro-lenders).
- Household debt-service-to-income ratios for 2021:
  - banks: ranged between 40 and 70 percent (average 57 percent)
  - microlenders: ranged between 30 and 67 percent (average 44.5 percent)
- Household debt does not appear to represent an immediate system-wide risk because:
  - Levels of indebtedness and NPL ratios are relatively low by international standards.
  - Bank credit growth has been fairly low and stable, and there is ample liquidity in the banking sector.
  - Large share of loans are collected at source and most borrowers are employed in the public service.
- However, pockets of risk exist due to unequal distribution by income, differential NPLs across banks, lower regulatory standards for micro-lenders, and growing lending to privately-employed individuals which could make NPLs more cyclical.
- Data gaps (household income, real estate prices, unemployment, debt service costs) and changes in NPL reporting hinder strong statistical relationships between NPLs and economic conditions.

### Interest rate risk and pass-through
- Changes in the MoPR have a one-to-one effect on prime lending rates.
- Approximately 66 percent of loans were priced at margins ranging from 2–10 percentage points above the prime rate as of June 2022.
- BoB set a spread of 411 basis points between the MoPR and the prime lending rate for a period of one year starting in April 2022; commercial banks can set final lending rates adjusted for customers’ risk profile.
- Unsecured household credit consists mainly of variable rate loans; many loans are susceptible to increases in lending rates when the BoB-imposed household loan interest rate reset moratorium expires in April 2023.
- A one-to-one pass-through implies that increases in the MoPR may increase loan rates and possibly push up NPLs.

### Interest rate risk in the banking book (IRRBB) and liquidity/funding vulnerabilities
- Over 65 percent of bank assets are floating rate.
- Approximately 35 percent of liabilities are fixed rate and are subject to administered rates that typically lag policy rate changes.
- Bank assets have significantly shorter maturities than liabilities; in a rising rate environment banks may experience a spike in net interest income (NII) and positive IRRBB exposures.
- Deposits of NBFIs account for 23 percent of banking sector deposits and comprise sizeable deposits from a few large depositors.
- Largest banks hold close to 68 percent of total household and corporate call and savings deposits, leaving smaller banks reliant on price-sensitive fixed-deposits or interbank funding.

### Homogeneity and peer group exposures
- Household loans dominate asset portfolios for Peer 1 and Peer 2 banks; smaller banks niche in corporate loans.
- Peer groups determined by share of total assets to banking system assets: Peer 1-2-3 represent 64-22-14 percent, respectively, as of end-June 2022.
- Differences in asset quality and provisioning coverage across banks raise concerns about underlying loan-book risks.

### Prudential forbearance and reporting
- Regulatory forbearance during the pandemic: restructured loans were 8.5 percent of total loans and special mention accounts were 1.3 percent of total loans; banks were allowed to treat COVID-19 related restructured loans as performing.
- Changes in prudential treatment and reporting standards may mask higher NPLs.

### Scope of stress tests and risk analysis
- Stress tests cover eight commercial banks (including two D-SIBs), representing 93.2 percent of total banking sector assets as of June 2022.
- Tests rely on supervisory data from the Bank of Botswana and a commercial bank survey by the FSAP mission.
- Stress testing exercises:
  - Solvency stress test: balance-sheet approach under baseline and adverse macrofinancial scenarios over 2023–2025, with sensitivity analysis on concentration of household credit and pass-through of policy rates to loan rates.
  - Liquidity stress test: 30-day liquidity stress based on proxy LCR for pula positions, different categories of HQLA and bank responses.
  - Interconnectedness and contagion: mapping domestic interconnections among banks and NBFIs; contagion analysis based on Espinoza and Sole (2010) simulating credit and funding shocks.

### Macro scenarios used in stress tests
- Baseline: Botswana recovers from COVID-19, growing by approximately 4 percent on average over the medium term; BoB assumed to continue increasing MoPR to address high inflation; inflation declines towards objective band over course of 2023; strong demand for diamonds supports reserves; bank lending supported by wage settlement for government employees.
- Adverse scenario: simulates an extreme but plausible recession with:
  - lower growth in 2022 of 1.58 percent below the baseline
  - a 10 percent drop in GDP in 2023 (equivalent to a 2-year cumulative shock of 2.3 standard deviations from the historical mean estimated from 1990 to 2021)
  - GDP path more severe than the pandemic, returns to potential in medium term
  - inflation spike from volatility in food and energy prices, tighter domestic monetary policy, decrease in diamond prices, and moderate nominal exchange rate depreciation against euro and USD
- Macro shock projections relied on the FSGM (Flexible System of Global Model) and WEO October 2022 baseline; details presented in Appendix IV.

*Source: IMF staff analysis, Botswana Financial Sector Assessment (chapter content).*

### 17.      The solvency stress test is based on a static balance sheet assumption and the

### 17.      The solvency stress test is based on a static balance sheet assumption and the

### Methodology and core assumptions
- Static balance sheet assumption: assets and liabilities remain constant over the three-year stress test horizon (2023 to 2025).
- Capital ratios projected using scenario-conditional paths for capital and RWAs (equation (2)).
- Satellite models link macroeconomic scenarios to key bank variables via credit risk and profit & loss (P&L) components.
- Capital evolution equation (as provided): 퐶퐶퐶... (equation (1)) and capital ratio definition (equation (2)) preserved as in source.

### Credit risk modeling and PD/LGD assumptions
- Probability of default (PD) proxied using projected NPL ratio obtained from satellite models.
- Loss given default (LGD) paths assumed based on bank-specific historical coverage ratios.
- Estimation/sample periods:
  - Bank-specific PD/LGD estimation: 2012Q1 to 2022Q2.
  - Bank-level NPL panel model: quarterly historical data from 2000Q4 to 2022Q2.
- NPL model specification: bank-level NPL ratios modelled as a logit function of lagged NPL ratios, nominal GDP growth, the real prime rate, and exchange rates per USD.
- Historical write-off and recovery treatment:
  - System average write-off and recovery rates of 1.4 percent applied where bank-specific data missing.
  - Minimum of PD assumed to be 0.1 percent.
- Empirical notes: economic activity found to have low explanatory impact on NPL evolution in Botswana; exchange rate impact on NPLs is indirect given limited foreign loan exposures.

### Market risk and valuation adjustments
- Market risk module evaluates valuation adjustments from changes in asset prices, interest rates, and exchange rates for securities portfolios and open FX positions.
- Treatment by security type:
  - AfS and HfT: mark-to-market approach. Modified duration formula used to revalue exposures as a function of reported residual duration and scenario bond yield assumptions.
  - Trading losses from HfT treated as realized losses, affect net income, and are subject to taxes and dividend payout.
  - Unrealized gains/losses from AfS affect OCI and are not subject to taxes; valuation changes in AfS affect capital one to one.
  - HtM: credit risk approach with provisions to cover expected loss as asset quality deteriorates.
- Valuation changes in open foreign positions estimated from scenario exchange rate fluctuations.

### Interest rate risk in the banking book (IRRBB)
- IRRBB assessed on interest-earning assets and interest-bearing liabilities.
- Projected effective interest rates for baseline and adverse derived from effective rates of total stock of interest-earning assets / interest-bearing liabilities (Appendix II).
- Main stress results incorporate:
  - Pass-through assumption on the funding rate based on effective rate on interest-bearing liabilities.
  - Full pass-through assumption on lending rate for main result.
- Sensitivity analyses include alternative lending rate pass-through assumptions: 0 and model-projected effective rates on interest-earning assets.

### Dividend payout and provisioning assumptions
- Dividend payout ratios: bank-specific 10-year average; assumed consistent across baseline and adverse scenarios.
- Dividends assumed paid only when net profit is positive; when net profit is negative, no dividends and capital used to cover losses.

### RWA and Basel II SA adjustments
- Basel II standardized approach requires RWA adjustments aligned with credit risk evolution.
- Change in RWAs for credit risk determined by:
  - (i) RWA growth in line with balance sheet growth (set to 0 under static balance sheet).
  - (ii) Decline in risk weights (to zero) generated by flow of provisions related to new NPLs.
  - (iii) Increase in risk weights from the non-provisioned segment of new NPLs.

### Regulatory hurdle rates (Box 2)
- Minimum Requirement (In percent):
  - Total Capital 12.5
  - Tier 1 7.5
  - CET1 4.5

### Key stress test results (Static balance sheet)
- Aggregate capital depletion in the adverse scenario is relatively small: less than 0.02 percent of GDP.
- Net interest income (NII) contributes positively to capital under both baseline and adverse scenarios due to:
  - shorter maturities,
  - higher share of variable-rate assets,
  - greater pass-through from policy rate on assets versus liabilities in a high-interest environment.
- Credit risk increases significantly under the adverse scenario in 2024, but NII increase outweighs capital ratio impact from rising NPLs.
- Market risk is low given limited securities exposures.

- System concentration and resilience:
  - Two D-SIBs represent 44 percent of system assets and remain resilient under both scenarios.
  - D-SIBs have high dividend payout and high credit risk under adverse scenario, yet CAR and Tier 1 remain above minimums due to high initial capital ratios: CAR 20 percent and Tier 1 13.6 percent.
  - Average dividend payout ratios: two D-SIBs 65.5 percent vs system average 31.2 percent.

- Market risk capital share:
  - RWA for market risk was 1.04 percent of total RWA as at end December 2021, reduced to 1.02 percent as at end June 2022.
  - Banks had no commodity or equity exposures as of December 2021.

### Sensitivity analyses — lending pass-through (Single factor)
- Full pass-through assumption on lending rate explains significant NII.
- Alternative pass-through scenarios tested on top of the adverse scenario:
  - (1) Model-projected effective rates on asset-earning assets.
  - (2) Lending pass-through equal to 0.
- Results:
  - CAR could drop up to 6 percentage point to 16 percent in 2025.
  - Even with pass-through = 0, aggregate CAR remains above regulatory thresholds.
  - However, three banks would face challenges meeting minimum capital requirements by 2025; aggregate recapitalization needs reach to 0.4 percent of GDP.

### Concentration risk in household loan portfolio (Single-factor sensitivity)
- Assumptions: shocks where 10 percent and 20 percent of household performing loans become non-performing (provisioning excluded).
- Baseline household NPL ratios and capital (in percent):
  - NPL ratios in household sector: Total 3.2; Peer 1 3.6; Peer 2 2.0; Peer 3 4.8.
  - Current Capital Ratios:
    - Total Capital: Total 18.5; Peer 1 18.9; Peer 2 18.2; Peer 3 17.1.
    - Tier 1: Total 12.7; Peer 1 13.1; Peer 2 13.3; Peer 3 10.4.
    - Core Tier 1: Total 12.2; Peer 1 13.1; Peer 2 10.7; Peer 3 10.4.
- Single-factor shock outcomes (selected results, in percent):
  - 10 percent household performing loans become non-performing:
    - Total Capital: Total 15.1; Peer 1 15.9; Peer 2 11.7*; Peer 3 15.8.
    - Tier 1: Total 9.3; Peer 1 10.1; Peer 2 6.7*; Peer 3 9.1.
    - Core Tier 1: Total 8.9; Peer 1 10.1; Peer 2 5.0*; Peer 3 9.1.
  - 20 percent household performing loans become non-performing:
    - Total Capital: Total 11.6*; Peer 1 12.9; Peer 2 5.1*; Peer 3 14.5.
    - Tier 1: Total 6.3*; Peer 1 7.1*; Peer 2 2.7*; Peer 3 7.8.
    - Core Tier 1: Total 6.3*; Peer 1 7.1*; Peer 2 2.7*; Peer 3 7.8.
  - Note: asterisks indicate numbers below the minimum requirements.
- Interpretation:
  - Under a 10 percent household shock the system remains largely resilient though one bank has total capital shortfall.
  - Under a 20 percent household shock several banks would face significant capital shortfalls and aggregated CAR could fall below minimum capital requirements; nonetheless most banks have strong total capital buffers.

### Macroprudential policy tools (Box 3) — summary observations
- Botswana’s macroprudential toolkit is broadly on par with other EMDEs but smaller than advanced economies.
- Concerns: high exposure to household debt, lumpy financing, balance-sheet concentration, potential excess liquidity from higher diamond prices, and reduced overseas exposure limits for retirement funds increasing deposit concentration risk.
- Recommended toolkit extensions to reduce solvency and liquidity risks:
  - Additional capital buffers: positive-cycle neutral counter-cyclical capital buffers or systemic risk buffers (releaseable in stress).
  - Basel III liquidity tools: LCR and NSFR to manage liquidity vulnerabilities.
  - Household-sector measures: caps on debt-service-to-income (DSTI) ratios to limit household debt risks.

### Policy recommendations (Section F)
- Data improvements:
  - Collect more granular PDs and LGDs from banks that use them for internal risk management.
  - Obtain NPL flow data by economic sector (inflows and outflows) rather than only beginning and closing balances.
  - Collect specific durations for existing loan and security portfolios (current statutory reports only capture securities duration below 1 year and over one year).
- Stress-testing framework enhancements:
  - Expand to multi-period scenario analysis at macro and micro levels.
  - Develop models for credit risk, market risk, and interest rate risks leveraging ongoing IMF TAs.
  - Compare macro-level stress outcomes with banks’ micro-level stress tests to better monitor system and bank-specific risks.
- Supervisory methodologies:
  - Develop supervisory tools to assess banks' exposure to IRRBB and sensitivity of net interest margins to strengthen the SREP.
  - Current supervisory data on RSA and RSL by repricing maturity should be supported by analytical tools to validate banks' internal IRRBB measurements and internal limits.
- Macroprudential measures for household credit concentration:
  - Consider complementary sectoral tools (e.g., DSTI caps) to manage household loan vulnerabilities.
  - Deploy macroprudential capital buffers (counter-cyclical or systemic risk buffers) to protect against a range of shocks.

### Top-down bank liquidity stress test — overview and methodology
- The BoB statutory liquid assets ratio (LAR) is defined as the ratio of a bank’s holding of statutory liquid assets to short-term deposits; statutory LAR requirement is 10 percent (daily requirement).
- LAR is a prudential liquidity regulation but less informative than LCR which assesses liquidity from a flow perspective.
- BoB plans to introduce Basel III LCR and NSFR framework to enhance liquidity monitoring.
- Liquidity stress test methodology:
  - Top-down LCR-proxy test using granular Basel II statutory report data, conducted in local currency over a 30-day period.
  - BoB has not implemented Basel III LCR and NSFR as binding requirements; assumptions were made to align with standard Basel III LCR components (Appendix V).
  - Detailed data mapping from BoB’s Basel II items to Basel III LCR available in Appendix VII.

*Source: IMF staff estimates and Botswana Financial Sector Assessment material as provided in the content unit.*

### 35.      Banks’ HQLA are concentrated in Peer 1 banks (71 percent of total) and are dominated

### 35.      Banks’ HQLA are concentrated in Peer 1 banks (71 percent of total) and are dominated

### HQLA composition and distribution
- Peer 1 banks collectively hold 71 percent of total HQLA.  
- Peer 3 banks collectively hold 8.5 percent of total HQLA.  
- Level 1 assets comprise 94 percent of total HQLA (based on the broad definition).  
- Level 2 assets comprise 6 percent of total HQLA, mostly marketable securities and corporate bonds; four banks hold no level 2 assets.  
- Banks’ level 1 assets consist mostly of government securities and balances due from the BoB (including current and repo accounts, standing deposit facility, and required reserves).  
- Peer 3 banks hold a greater proportion of required reserves in their HQLA holdings than other groups.  
- Peer 1/2/3 banks’ share of total assets in the banking system is 64/22/14 percent, respectively, as of end-August 2022.

### Baseline and adverse scenario design (LCR testing)
- The baseline scenario adopted a “narrow” coverage of HQLA to be more consistent with the BoB’s statutory liquid assets; the BoB’s regulatory framework excludes the primary reserve requirement (PRR) and government bonds with maturities beyond one year from its statutory liquid assets. These excluded items make up 28 percent of the total HQLA in the banking system.  
- Baseline parameters largely follow Basel III LCR factors, with two exceptions:  
  - Run-off rate for parastatals (government and development banks) set at 25 percent.  
  - Run-off rate for NBFIs’ deposits set at 40 percent.  
- Total outflows in the baseline scenario represent 21 percent of total banking sector assets (BWP 24 billion).  
- BoB statutory haircut schedule used in baseline: 1.5 percent haircut on government bonds with maturities of less than one year; 4 percent haircut on government bonds with maturities greater than one year.  
- Adverse scenario: higher run-off rates on outflows (50 percent and 15 percent) increased outflows by BWP 6 billion from the baseline. Parameters anchor banks’ outflows/total assets ratio to 26 percent, and inflows/total assets ratio to 9 percent. Haircuts in adverse scenario: 5 percent for government bonds with maturities of less than one year and 15 percent for government bonds with maturities greater than one year.

### Stress test results (LCR-proxy)
- Under the baseline (narrow HQLA), LCR ratios for all commercial banks were substantially above the 100 percent threshold.  
- Peer 3 banks had the highest average LCR because of sizeable short-term liabilities: Peer 3 banks’ average 30-day inflows to total assets was 26 percent (after applied LCR factors), compared to Peer 1 banks' average of 16 percent.  
- Peer 3 banks’ average 30-day inflows to 30-day outflows was 193 percent, compared to 72 percent for Peer 1 banks; the inflow cap set at 75 percent of outflows was binding for Peer 3 banks.  
- In the adverse scenario using the baseline (narrow) HQLA:  
  - LCR ratios of four banks dropped by more than 50 percent.  
  - Five banks fell below the 100 percent threshold.  
- Expanding eligible liquid assets to include required reserves and government bonds with maturity over one year (with 15 percent haircuts) improved resilience: LCRs moved closer to 100 percent for most banks; only two banks remained below but close to 100 percent.  
- Peer 3 banks saw a larger increase in LCRs from the expanded coverage due to their higher holdings of required reserves.

### LCR sensitivity analysis and interpretation
- Peer 1 banks are more vulnerable to short-term inflow shocks due to sizeable long-term maturities of assets (securities holdings and loan lending).  
- Banks with heavy reliance on short-term household and wholesale funding are more vulnerable to outflow shocks.  
- Peer 3 banks are resilient to inflow shocks due to substantial investment in short-term liquidity (inflows were capped at 75 percent of outflows in all adverse scenarios 0–3).  
- Banks with higher LAR ratios did not necessarily have higher LCR ratios; LCR focuses on 30-day stress, applies haircuts and run-off rates, and considers off-balance-sheet commitments, making it more informative for monitoring short-term liquidity risk.

### Contagion and interconnectedness findings
- Banking sector has the largest exposure (48.67 percent) to households through unsecured loans.  
- NBFIs’ cross-border exposures mainly reflect pension funds’ offshore investment, which account over 60 percent of pension funds’ total exposure portfolio.  
- Major funding source in banking sector is from non-financial corporations, comprising 49.76 percent of total banking deposit by May 2022.  
- Interbank market: Peer 1 banks are the predominant domestic lending source, followed by medium banks.  
- Contagion simulation assumptions: loss given default (LGD) set to 85 percent for credit shocks; when funding and credit shocks occur simultaneously, borrower banks can replace only 50 percent of previously borrowed funds and assets are fire-sold at a 35 percent discount.  
- Simulation outcomes:  
  - No single failure of a bank would trigger the failure of another bank under the simulated credit and funding shocks.  
  - Interbank positions are small compared to banks’ capital.  
  - Peer 1 banks are the most contagious group, followed by Peer 2. The failure of any bank in the Peer 1 group could result in a potential capital reduction of up to 12.7 percent for a counterparty bank.  
  - In the case of simultaneous credit and funding shocks and failure of all banks in Peer 1, the remaining banking system would experience a total capital loss of 12 percent.  
  - Peer 3 banks are among the most vulnerable group: they experienced capital loss of 11 percent resulting from other banks' defaults.  
  - Peer 2 group vulnerability driven by a bank particularly vulnerable to the other two banks in the system.

### Policy recommendations (selected)
- Include the PRR as statutory liquid assets, particularly for future LCR implementation; Basel III guidance allows reserve requirement among liquid assets “to the extent the central bank policies allow their use in time of stress.” Commercial banks’ balances held in the PRR account meet three pre-conditions for Basel III LCR compliance.  
- In the medium term, include long-term government bonds as statutory liquid assets with appropriate haircuts; currently government bonds with maturities over one year are eligible collateral for BoB refinancing facilities but not eligible as statutory liquid assets.  
- Calibrate appropriate haircuts for long-term government bonds; the BoB’s existing haircut schedule (up to 4 percent on government bonds with maturity over 10 years) may not reflect current market risks given limited secondary trading—BoB should monitor and improve calibration over the medium term.  
- Develop a clear plan for introducing the Basel III LCR and NSFR liquidity framework that considers:  
  - Improvements in banks’ statutory reports to include net loans and advances to different counterparties in different maturity buckets, balances due to domestic and foreign banks over a 30-day horizon, for parameter calibration;  
  - Developing regulatory guidance for banks on the application of the LCR and NSFR in Botswana’s context and how to use them for internal liquidity management;  
  - Appropriate calibration and application of regulatory weights on assets, funding structure, and liquidity characteristics of Botswana’s banking sector;  
  - A path for supervisors to gain experience with LCR and NSFR reporting and validation by introducing a sufficiently long period where banks report ratios for monitoring purposes only before imposing prudential limits.  
- Provide more technical guidance from the BoB to banks, particularly small banks, on implementation and accurate calculation of LCR and NSFR; currently four banks report LCR monthly or quarterly, two banks use LCR with LAR or internal models for market liquidity monitoring, and only one of two banks reporting NSFR also report LCR.

*Source: IMF staff analysis in the FSAP technical note (Botswana).*

### 51.      The interconnectedness analysis highlights the needs for NBFIRA to enhance the data

### 1bwaea2024005 - 51.      The interconnectedness analysis highlights the needs for NBFIRA to enhance the data

### Interconnectedness, data gaps, and supervisory reporting
- The interconnectedness analysis highlights the needs for NBFIRA to enhance the data quality and streamline the data sharing with BoB.
- Recommendation: Upgrade supervisory reporting templates to collect standardized information on bilateral exposures between banks and NBFIs (including inshore and offshore exposures).
- Purpose of added data:
  - Contagion analysis among financial institutions.
  - Identify the most vulnerable institutions across the financial sector to enhance supervision.

### Macroprudential policy recommendation: tailored Pillar II capital buffers
- Recommendation for BoB: Consider imposing tailored Pillar II capital buffer requirements for specific banks.
- Rationale:
  - Banks with a high index of vulnerability concurrently fail to meet the capital requirement under adverse scenario in solvency stress test.
  - These banks typically exhibit low capital ratios and paid high dividends based on historical patterns.
  - Implementing tailored Pillar II buffers would enable banks with high index of vulnerability to enhance their ability to withstand adverse economic shocks and address the possibility of a “domino effect”.
- Context note: The assessment and definition of D-SIBs by the BoB have made the use of capital buffers an important part of the macroprudential toolkit.

### NPL Ratios Satellite Model: methodology and key parameters
- Data used:
  - Banks' NPL ratios quarterly from 2000Q4 to 2022Q2.
  - Loan recoveries and write-offs data available for a shorter period, depending on the bank.
- Transformation:
  - Logit transformation applied to NPL ratios to ensure predictions between 0 and 1 and capture nonlinearities.
- Model specification:
  - Linear model on logit-transformed NPL ratios with lagged dependent variable and exogenous macroeconomic and financial regressors.
  - Bayesian Model Averaging (BMA) OLS panel approach used to eliminate model uncertainty and allow sign restrictions.
- Table I.1. Key long run multipliers (with sign restriction) and fit:
  - L.Logit_NPL: 0.54*
  - nGDP growth: 0.001
  - Real Prime Rate: 0.015
  - Exchange Rate: 0.17*
  - R-squared: 0.58
  - Note: *denotes a higher posterior inclusion probability than the prior.

### Proxy PDs: approximation methodology
- BoB does not collect PDs and LGDs; PDs approximated using NPL stock, recoveries, and write-offs.
- Proxy PD equation (preserve notation as in source):
  - PPeeI PP P P t = [NNll ii −(1− ᾱt )NNll ii−1] NPPP RRRR ii−1
  - Where ᾱt represent the proportion of NPLs that are either written-off or recovered to performing.
- Change of NPLs between consecutive periods expressed and simplified assuming constant alpha:
  - ΔNPNN t = NPP C t − NPP C t−1 = NNNN NN t − αt * NPP NN t−1
- Assumption used for ᾱ:
  - ᾱ = (recoveries+ write-offs)/ NPP NN t−1
- Visual: Figure I.1. Botswana: NPL Ratio and PD Projections (sources: Bank of Botswana, and IMF staff estimates).

### Satellite models — Interest Income and Interest Expense
- Separate estimations for funding rates and lending rates based on historical interest rates for loans and deposits.
- BMA OLS panel methodology used.
- Results: high explanatory power of nominal prime rates and exchange rate.

- Interest Income Satellite Model (Table II.1) — Long run multipliers:
  - L.Logit_Lending Rate: 0.41*
  - inflation: -0.01
  - L.inflation: 0.012
  - Nominal Prime Rate: 0.035*
  - Exchange Rate: 0.17*
  - R-squared: 0.29
  - Note: *denotes a higher posterior inclusion probability than the prior.

- Interest Expense Satellite Model (Table II.2) — Long run multipliers:
  - L.Logit_Funding Rate: 0.48*
  - inflation: 0.002
  - L.inflation: -0.0002
  - Nominal Prime Rate: 0.065*
  - Exchange Rate: 0.03*
  - R-squared: 0.31
  - Note: *denotes a higher posterior inclusion probability than the prior.

### Solvency stress test: methodology and scope
- Framework: Top-down by FSAP Team using IMF Solvency Stress Test Workbox (Balance-sheet model).
- Institutional perimeter:
  - Eight commercial banks (including two D-SIBs).
  - Market share: 93.2 percent of assets in the Botswana banking system as of June 2022.
- Data and baseline date:
  - Supervisory data (balance sheet and income statements).
  - Commercial Bank Survey (bank-level average duration of securities in banking book and trading book, dividend payout ratio by banks in the past 10 years).
  - Starting position: June 2022.
- Channels and modeling:
  - Credit Risk: Satellite models link credit risk variables with macro variables bank-by-bank to estimate loan losses.
  - Market risk: Mark to Market (MtM) for HfT and AfS securities when shocks to risk-free rates and credit spread; repricing and credit spread risk formulas preserved in source.
  - Net Interest Income / IRRBB: assessed using interest income rate/interest expense rate satellite models and maturity gap analysis.
- Stress test horizon: 3 years (2023-2025).
- Scenarios: Two macro scenarios — baseline and adverse.

### Solvency stress test: Baseline scenario summary results (Table III.1, In Pula millions unless otherwise indicated)
- GDP:
  - 2022: 220,464,000
  - 2023: 246,853,541
  - 2024: 268,206,372
  - 2025: 290,063,479
- Nominal GDP change (percent): 12, 8.7, 8.1
- Capital ratios (percent):
  - Total capital adequacy ratio: 18.5, 19.6, 20.8, 22.1
  - Tier 1 capital ratio: 12.7, 13.8, 15.0, 16.2
  - Common/Core tier 1 ratio: 12.2, 13.3, 14.4, 15.7
- Number of banks failing the tests total: 0, 0, 0
- Total Capital in the banking system: 14,344; 15,153; 15,980; 16,889
- Tier 1 Capital in the banking system: 9,867; 10,675; 11,503; 12,412
- Common/Core Capital in the banking system: 9,467; 10,275; 11,102; 12,012
- Sum of Risk Weighted Assets: 77,444; 77,200; 76,893; 76,602
- Change of Total capital in the banking system: 808; 828; 909
- Change of RWA: -243; -307; -291
- Net Profit: 2,406; 1,728; 1,781; 1,949
- Net Interest Income: 4,483; 4,482; 4,904; 4,663
- Loss provisions (Credit Risk): -33; -487; -614; -582
- Net non-interest income: -1,415; -1,460; -1,486; -1,498
- Losses/Gains on market portfolios: 0; -13; -12; -11
- Taxes: -628; -497; -515; -562

### Solvency stress test: Adverse scenario summary results (Table III.2, In Pula millions unless otherwise indicated)
- GDP:
  - 2022: 220,464,000
  - 2023: 244,252,065
  - 2024: 243,362,500
  - 2025: 266,456,047
- Nominal GDP change (percent): 10.8, -0.4, 9.5
- Capital ratios (percent):
  - Total capital adequacy ratio: 18.5, 19.5, 19.9, 22.0
  - Tier 1 capital ratio: 12.7, 13.7, 14.0, 16.1
  - Common/Core tier 1 ratio: 12.2, 13.2, 13.5, 15.6
- Number of banks failing the tests total: 0, 2, 2
  - CAR failures: 0, 1, 1
  - T1R failures: 0, 1, 2
  - Core T1R failures: 0, 1, 1
- Necessary Recapitalization (absolute): 0, 163.9, 163.8
- Necessary Recapitalization (percent of GDP):
  - CAR: 0, 0.009, 0.013
  - T1R: 0, 0.009, 0.005
  - Core T1R: 0, 0.058, 0.059
- Total Capital in the banking system: 14,344; 15,067; 15,107; 16,649
- Tier 1 Capital in the banking system: 9,867; 10,589; 10,630; 12,171
- Common/Core Capital in the banking system: 9,467; 10,190; 10,230; 11,772
- Sum of Risk Weighted Assets: 77,444; 77,123; 75,878; 75,599
- Change of Total capital in the banking system: 723; 40; 1,542
- Change of RWA: -320; -1,245; -280
- Net Profit: 2,406; 1,609; 844; 3,163
- Net Interest Income: 4,483; 4,506; 5,014; 5,642
- Loss provisions (Credit Risk): -33; -640; -2,489; -560
- Net non-interest income: -1,415; -1,440; -1,566; -1,510
- Losses/Gains on market portfolios: 0; 28; -11; -78
- Taxes: -628; -470; -309; -892

### Risk Assessment Matrix (RAM) — selected risks and impacts
- Global Risks:
  - Abrupt global slowdown or recession: Overall Level of Concern: Medium; Relative Likelihood: —; Expected Impact if Materialized: High.
    - Key impacts: Lower demand for diamond and other commodity exports; lower SACU revenue; pressure on external and fiscal balances; pressure on the pula.
  - Rising and volatile food and energy prices: Overall Level of Concern: High; Expected Impact: Medium.
    - Key impacts: Higher inflation → higher interest rates; potential losses for retirement funds and insurance; liquidity pressures.
  - De-anchoring of inflation expectations and stagflation: Overall Level of Concern: High; Expected Impact: High.
    - Key impacts: Increases in funding costs; tighter financial conditions; credit risks from loan re-pricing; valuation losses for retirement funds and non-bank sector.
  - Natural disasters related to climate change: Overall Level of Concern: Low; Expected Impact: Medium.
    - Key impacts: Infrastructure damage, supply disruptions, higher NPLs, higher loan loss impairment.
  - Cyber attacks: Overall Level of Concern: Medium; Expected Impact: High.
    - Key impacts: Loss of confidential data, financial losses, business disruption, reputational damage.
- Domestic Risks:
  - Delays in implementing fiscal consolidation: Overall Level of Concern: Medium; Expected Impact: Medium.
    - Key impacts: Amplified if coupled with sovereign credit rating downgrade; crowding out of private sector credit; lower growth and higher NPLs.

### Adverse scenario assumptions (high-level)
- External & domestic drivers:
  - Use IMF’s Flexible System of Global Models for external context; calibrated based on previous crisis episodes.
  - Driven by combination of external shocks amplified by domestic characteristics (see RAM).
- Macroeconomic shock profile:
  - Botswana experiences a U-shaped growth path with a stagflation scenario.
  - Assumes a lower growth in 2023 (1.52 percent below the baseline) and a drop in GDP of -9.47 percent in 2024.
  - This is equivalent to a shock to 2-year cumulative growth of 2.3 standard deviation from historical mean (estimated from 1990 to 2021).
  - GDP growth returns to its potential in year three.
  - Unemployment rate over the 3-year horizon: 25.77 percent, 30.58 percent, and 30.5 percent.
  - Pula depreciation against USD most significant in year 2, with 13.25 percent depreciation below baseline.
  - Inflation: 14.72 percent and 12.39 percent in first two years, dropping to 6.03 percent in year 3.
  - Monetary policy rate will stay high to fight inflation.
  - Diamond prices: 9 percent, 24.1 percent, and 11.5 percent below baseline over the 3-year horizon.

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

### 4. Risks and

### 4. Risks and Buffers

### Risks and factors assessed
- Credit risk (provision costs): determined by changes in estimated new NPL flows.
  - EEPPCNINICNL NNIeeee = PeII PP P P P * NLLP * ERP (formula preserved as in source).
- Market risk: impact of financial variable’s evolution on sovereign bonds and FX open positions.
- Interest rate risk in the banking book.
- Net fee and commission income, other income and non-interest expense evolve with macroeconomic conditions.
- Behavioral adjustments:
  - Balance sheet growth assumption: Static balance sheet assumption (Balance sheet growth is zero).
- Model note: It is a multi-region, forward-looking, semi-structural dynamic stochastic general equilibrium (DSGE) model.

### Calibration of risk parameters (PDs, LGDs, EADs)
- PDs: proxy based on estimated new NPL flows over performing loans. PD is derived in the following steps (formulas preserved exactly as in source):
  - NNNNNN NP N_t = NP C_t − NP C_{t−1} + RNI I IeneecNe_t + ReecC N I r r e_t
  - ΔNP N_t = NP C_t − NP C_{t−1} = NNNNNN NP N_t − RNI I IeneecNe_t − ReecC N I r r e_t
  - ΔNP N_t = NP C_t − NP C_{t−1} = NNNNNN NP N_t − α_t * NP N_{t−1}
  - NNNNNN NP N_t = NP C_t − (1−α_t) NP C_{t−1}
  - NNNNNN NP N_t = NP C_t − (1− α̅) NP C_{t−1}
  - PeII PP P P P_t = [ NP C_t − (1− α̅) NP C_{t−1} ] / PPNeerIIeII CIL NI CIIe_t−1
- Where α̅ = (recoveries+ write-offs)/ 〖NPL〗_{t−1}, and it is bank specific. Because write-off and recoveries data is only available for shorter period, average alpha is calculated by using existing data and applied to historical NPLs to estimate PDs.
- LGDs: proxy based on average historical provision coverage ratio by banks (provisions/ NPLs).
- EADs: proxy as performing loans in t-1.

### Regulatory, accounting, and market-based standards and parameters
- Basel II standardized approach.
- The hurdle rates are based on minimum capital requirements: 7.5 percent for Tier1 Capital, 4.5 percent for CET1, and 12.5 percent for total capital.
- RWAs evolve with credit growth, net of increase in provisions.

### Reporting / Output presentation
- Capital ratios pre- and post-shock and capital shortfall for the aggregate banking system.
- Decomposition of the key drivers of changes in capital ratios system wide.
- Distribution of capital ratios system wide over the scenario horizon.

### Banking Sector: Liquidity Test (methodology and scenarios)
- Domain: TD by FSAP Team.
- Institutional perimeter:
  - Number of banks: Eight commercial banks (including two D-SIBs).
  - Market share: 93.4 percent of assets in the Botswana banking system as of end-August 2022.
- Methodology:
  - LCR (Liquidity Coverage Ratio) – proxy test, using the bank-level Basel II statutory reports data at end of August 2022.
  - Necessary assumptions are made to be consistent with standard Basel III LCR framework when breakdown data are unavailable (e.g., proportions of breakdown items).
  - Banks’ 30-day “Net loans and Advances” (M-LIQGAP) lacks breakdown data by counterparties; therefore, ratios of “Gross Loans and Advances” of all maturities were applied to proxy the breakdown data.
- Scenarios:
  - Baseline: coverage of HQLA close to the BoB’s statutory liquid assets, excluding required reserves and government bonds with maturities beyond one year.
  - Severe: same HQLA coverage as baseline but with larger haircuts on HQLA, lower inflows from retail and non-financial corporations, and larger run-off on banks’ wholesale and retail deposits funding.

### LCR data mapping (selected mapping notes and baseline factors)
- Baseline HQLA mapping highlights (items and assigned percentages preserved as in source):
  - Coins and bank notes: 100%.
  - Treasury Bills equal to or less than 6 months: 100%.
  - Bank of Botswana Certificates (less those pledged as security): 100%.
  - Other government obligations (with maturity less than one year): 100%.
  - Government Bonds - (with maturity less than one year): 98.5%.
  - Claims on Sovereign or CBs - Exposure assigned credit quality grade of AAA to AA-: 100%.
  - Qualifying central bank reserves (Due from Bank of Botswana): 100%.
  - Level 2A Assets: 85% for certain claims on Sovereign or CBs rated A+ to A- and related items.
  - Level 2B Assets: Qualifying Mortgage Backed Securities 75%; Qualifying corporate debt securities rated between A+ and BBB- 50%; Qualifying common equity shares 50%.

- Inflows baseline factors (examples preserved):
  - Level 1 assets inflows: 0%.
  - Level 2a assets inflows: 15%.
  - Level 2b assets inflows: Eligible Residential Mortgage Backed Se 25%, Other 50%, Margin lending backed by all other collateral 50%, All other assets 100%.
  - Retail counterparties inflows: Households 50%; parastatals (govt & dev banks) 100%; corporations 50%; NBFIs 50%.

- Outflows baseline factors (examples preserved):
  - Retail Deposits: Stable deposits 5%; Less stable retail deposits 10%.
  - Unsecured Wholesale Funding: Stable deposits 5%; Less stable deposits 10%; Operational deposits generated by clearing, custody 25%; Portion covered by deposit insurance 5%.
  - Secured Funding: 0% for Balances due to Bank of Botswana; Secured funding backed by Level 2A assets 15%; Debt securities in issue (< 5Y) 25%; Repurchase agreements (with Fin corporations) 25%; Funding backed by RMBS eligible for Level 2B 25%; Funding backed by other Level 2B assets 50%; Other secured funding transactions 100%.
  - Net derivative cash outflows: 100%.

### Macroprudential governance, toolkit, and calibration
- Institutional assessment dimensions:
  - Willingness to act: clear mandate and accountability, communication tools.
  - Ability to act: obtaining necessary information, activating regulatory constraints, changing regulatory perimeters.
  - Cooperation: effective inter-agency cooperation in risk assessment and mitigation.
  - Toolkit and calibration: availability and calibration of macroprudential tools for the banking sector.

- Progress and institutional arrangements:
  - NBFIRA Act, 2022 establishes context and oversight for safety, soundness, and stability of NBFIs.
  - Financial Intelligence Act (2022) strengthens oversight of anti-money laundering activities.
  - Financial Stability Council (FSC) established in 2019; Bank of Botswana Amendment Act (BoBAA) elevates FSC to a statutory committee with specific mandate and expanded remit.
  - The BoB assumes the primary macroprudential policy function for banks in Botswana and has implemented a wide range of prudential tools.

- Powers and cooperation:
  - BoB and NBFIRA have broad powers to collect information to monitor systemic risks.
  - FSC has “semi-hard” powers: can make formal recommendations with a “comply or explain” mechanism.
  - Memorandum of Understanding (MoU) underpins information sharing, cooperation, and communication; FSC membership has broadened to include BoB, MoF, FIA, and representation from the impending DIS; as of 2021, Botswana Stock Exchange has observer status.

- Macroprudential toolkit and international comparison:
  - Under Category V (liquidity tools) banks are required to hold liquidity for prudential reasons through the PRR (currently at 2.5 percent) and a LAR (at 10 percent). LCR and NSFR have not been implemented.
  - Category II (banking sector as a whole): Botswana has loan-loss provision and Capital Adequacy Ratio. Forward looking provisioning introduced in 2018 under IFRS9, but not comprehensively reviewed or tested by the BoB.
  - Countercyclical buffers and limits on leverage ratio are being considered but not activated.
  - The BoB has designated D-SIBs: Two banks out of eight have been identified as D-SIBs since 2021 based on BCBS methodology. The D-SIBs cover about 90 percent of the banking system by assets.

### Selected macroprudential instruments and implementation dates (Table VI.1 highlights)
- Limit on large exposures:
  - Exposures shall not exceed 30 percent of a bank’s unimpaired capital, and total large exposures shall not exceed 800 percent of unimpaired capital.
  - Implementation Date: Nov-09.
- Capital Adequacy Ratio:
  - Implementation of Basel III definition of capital: Unimpaired Capital to Risk-Weighted Assets of minimum 15 percent (CET1 – 4.5%, AT1 – 3%, T2 – 7.5%).
  - Implementation Date: Jan-16.
- Household tools / Liquidity Risk Monitoring (prudential):
  - Bank reporting requirements: risk-weighted reporting based on LTV.
  - Implementation Date: Jan-16.
- Liquidity Tools:
  - Liquid asset ratio: current ratio is 10% (1995).
  - Prudential Reserve Requirement: current ratio is 2.5% (2019).
  - Liquidity gap profile (monitoring only); Top 20 depositors (monitoring only).
- D-SIB framework:
  - Domestic Systemically Important Banks (D-SIBs) framework partially implemented in 2021, with D-SIB designation but no associated additional capital requirement.
  - Implementation Date: Jan-21.
- Financial Stability Council:
  - Comprises Ministry of Finance (MoF), the BoB, NBFIRA, Financial Intelligence Agency (FIA); primarily for monitoring and risk assessment, overseeing and guiding macro-prudential policy framework and implementation.
  - Implementation Date: Feb-19.

*Source: 1bwaea2024005 - 4. Risks and (IMF staff).*

---


_Source: https://www.imf.org/-/media/files/publications/cr/2024/english/1bwaea2024005.pdf_