## 1. Exchange Rate, Foreign Reserves, and Foreign Exchange Intervention

## Source details

**Canonical URL:** [1. Exchange Rate, Foreign Reserves, and Foreign Exchange Intervention](https://www.imf.org/-/media/files/publications/cr/2025/english/1albea2025002-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/cr/2025/english/1albea2025002-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/cr/2025/english/1albea2025002-print-pdf.pdf.json)

---

### Context and key facts
- Albania is a small open emerging market with an inflation targeting regime and a sustained appreciation trend of the Albanian lek amid a tourism boom.
- Bank of Albania (BoA) foreign exchange intervention (FXI) during the first 9 months of 2024 amounted to 2.6 percent of 2023 nominal GDP, almost triple the amount of the same period in 2023.
- Albania’s OTC FX market turnover is estimated at 11 billion USD yearly, compared with an emerging market average of 52 billion USD.
- Pass-through evidence: cumulative impact of about 0.2 percent (for a 1-percent appreciation) on headline CPI over four quarters (empirical analysis following Gopinath (2015)).

### Measures of market liquidity and balance-sheet exposures
- FX market shallowness
  - High volatility in UIP premia (especially since Q1 2023) and low turnover indicate shallow FX markets.
  - Shallow markets imply financial intermediaries have limited capacity to absorb excess currency demand/supply, increasing exchange rate deviations from fundamentals.
- Currency mismatches and hedging
  - Share of unhedged FX loans declined from 50 percent of total loans in 2014 to 25 percent in 2024.
  - Two-thirds of unhedged FX loans in 2024 are accounted for by the real estate sector.
- Inflation expectations
  - Inflation expectations are described as well-anchored in the medium and long run.
  - There is room for monetary policy to support a sustainable return of inflation to target without FXI.

### QIPF (Quantitative Integrated Policy Framework) model — specification and estimation
- Model features
  - New Keynesian open-economy QIPF model adapted for Albania, linearized following Chen et al. (2023) and extended to include trend productivity growth and a domestic financial sector.
  - Assumes FX intermediaries have limited risk-bearing capacity (as in Gabaix and Maggiori, 2015), which generates realistic exchange rate volatility and implies sterilized FX interventions have real effects through UIP premia.
- Estimation details
  - Estimated on a dozen quarterly series for Albania over 2009–2024 using Bayesian maximum likelihood.
  - Euro area block estimated separately; euro area variables included to capture foreign influences.
- Key estimated properties
  - Pass-through of nominal exchange rate movements to consumer prices is gradual due to local currency pricing-to-market behavior.
  - Interest rate policy gradually adjusts to deviations of expected one-year ahead core inflation from target and output deviations from trend.
  - Foreign exchange market is estimated to be often shallow, implying FX interventions can have sizeable macroeconomic effects when domestic policy rate is passive.

### Scenario analysis — Tourism-driven fundamental shock
- Scenario design
  - Stylized tourism boom driven by fundamentals: productivity (catch-up trend and temporary), relative export prices, and domestic demand shocks phased in gradually as MIT shocks.
  - Central bank modeled as gradually learning; does not have perfect foresight and initially understates the rise in potential output.
- Baseline vs alternative monetary policy responses
  - Baseline: monetary policy follows estimated historical rule; monetary policy eases because appreciation puts downward pressure on import prices and productivity lowers unit costs, causing inflation to fall below target.
  - Alternative (discretionary) policy: additional policy rate cuts assumed to offset the appreciation in the lek, narrowing the output gap and bringing inflation closer to target while avoiding volatile interest rate movements.
  - Illustrative welfare metric: the alternative policy delivers a 25 percent improvement over the historical policy response measured by a standard central bank quadratic objective in output gap and inflation (note: still leaves output gap negative and inflation below target, implying scope for even more stimulative policy).
- Policy implication from the tourism scenario
  - Monetary policy is a suitable tool to address fundamental appreciation pressures from a tourism boom; FXI is not necessary under this fundamental shock given monetary policy retains significant traction over exchange rates.

### Scenario analysis — Non-fundamental (risk-on) capital flow shocks and trade-offs
- Role of non-fundamental shocks
  - Non-fundamental (“risk-on”) shocks can appreciably affect the real exchange rate in the presence of shallow FX markets by widening the UIP “wedge.”
  - Estimated contributions to long-run RER variance:
    - Around 5 percent if FXI is reacting to exchange rate movements.
    - Around 15 percent assuming no FX interventions are used to lean against exchange rate movements.
  - Cumulative contribution since 2021: non-fundamental shocks appear to have appreciated the lek by around 11 percent (subject to estimation and specification uncertainty).
- Policy trade-offs
  - Risk-driven exchange rate volatility can create a trade-off between output and inflation: an appreciating exchange rate from risk-on inflows can lower inflation while producing a persistent rise in output.
  - Interest rate policy primarily moves inflation and output in the same direction via aggregate demand and thus is not ideally suited to mitigate shocks that imply opposite movements in inflation and output.
  - In such cases, FXI may be useful to lean against temporary, non-fundamental capital in- and outflows, especially given market shallowness and limited intermediary capacity.

### Overall findings and policy recommendations
- Primary IPF friction in Albania: shallow FX markets that can justify FXI to moderate excessive exchange rate movements arising from temporary non-fundamental shocks.
- FX mismatches and de-anchored inflation expectations are broadly contained; remaining FX mismatch risks (notably in real estate) are better addressed by preemptive and targeted macroprudential measures (MPMs) rather than FXI.
- For fundamental shocks (e.g., tourism-driven productivity and demand changes), monetary policy is the appropriate tool; FXI is not warranted.
- For non-fundamental, sentiment-driven capital flow shocks in the context of shallow FX markets, FXI can be a useful complementary tool to lean against excessive exchange rate movements, given the limitations of interest rate policy to address the output–inflation trade-off these shocks create.
- Policymakers face real-time challenges in recognizing shifts in potential output and distinguishing fundamental from non-fundamental shocks; the QIPF model can help quantify trade-offs and guide the appropriate policy mix.

*Prepared by David Bartolini, Jakree Koosakul, Rebecca Huang, Jesper Linde, and Roland Meeks (IMF staff).*

### FX intervention (FXI): modeling and key findings
- Sterilized FX intervention (FXI) is modeled as an endogenous reaction to changes in the nominal exchange rate combined with gradual mean-reversion to restore the desired ratio of reserves to GDP.
- Under an inflow shock that makes the euro temporarily cheap, the FX rule calls for euro purchases, which:
  - Greatly reduces the lek exchange rate appreciation;
  - Eliminates the output-inflation trade-off observed without FXI;
  - Suggests that an easing of interest rate policy coordinated with FXI could realize greater stabilization benefits.
- Sensitivity: results depend on the strength of the FX policy response to the exchange rate and on which historical intervention data are used in the QIPF model; across alternatives FXI remains an effective tool for addressing risk shocks, though quantitative responses vary.
- Figure 6 (IMF staff estimates) contrasts central bank responses “With FXI” and “Without FXI” for:
  - Policy Rate (APR)
  - Output (% trend development)
  - Real Exchange Rate (Consumption based)
  - CPI Inflation (APR)

### Conclusion and policy considerations on exchange rate role
- The exchange rate should continue to play its role as a shock absorber with FXI reserved to moderate the impact of non-fundamental shocks.
- Empirical application of the quantitative IPF model to Albania indicates:
  - Appreciation of the lek has historically been driven largely by fundamentals;
  - Non-fundamental factors have played a modest role in some episodes.
- Policy recommendations:
  - Consider letting the exchange rate adjust more flexibly and rely on interest rate policy as the primary tool for price stability.
  - Use FX interventions in cases of non-fundamental shocks to lower output and inflation volatility.
  - Internalize potential adverse consequences of FXI and further reserve accumulation, including on risks to the central bank balance sheet, interest rate transmission, and financial market development.

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

### Banking sector context and resilience (overview)
- Structural and performance indicators:
  - NPLs declined to just under 5 percent of total loans in Q1 2024.
  - Foreign ownership share of assets fell from 90 percent in 2013 to 65 percent in Q1 2024.
  - Return-on-assets (ROA) averaged around 1 percent between 2019 and 2022, rising to an average of 1.7 percent in 2023 and 2024.
  - Average capital adequacy ratio of 19.3 percent in July 2024 (minimum regulatory requirement: 12 percent).
  - The banking sector is predominantly bank-based: 11 commercial banks own over 90 percent of financial sector assets.
  - Four banks identified as D-SIBs, owning 65 percent of banks’ assets and 66 percent of banks’ capital.
  - Herfindahl-Hirschman Index (HHI) of 1428 based on Q1:2024 asset data (market considered competitive).
- Remaining vulnerabilities highlighted:
  - High sovereign securities holdings and sovereign-bank nexus risks.
  - Local-currency government bond and money markets remain illiquid.
  - High euroization of the economy; two-thirds of unhedged FX loans concentrated in the real estate sector, which saw rapid credit growth in 2024.

### Stress testing approach (capital and liquidity)
- Methodology:
  - Bottom-up balance-sheet approach per Cihak (2014) using bank-by-bank balance sheet data for Q1:2024.
  - Horizon: one year.
  - Banks considered undercapitalized only if CAR falls below 12 percent.
- Risks covered:
  - Credit risks: NPL-related shocks and large-exposure shocks.
  - Market risks: interest rate and exchange rate channels.
  - Liquidity resilience assessed via an exercise varying haircuts on liquid assets under the LCR standard.

### Credit risk stress test scenarios and assumptions
- General credit risk (NPL shock):
  - Three scenarios:
    1. Case A: 100-percent increase in NPLs (baseline, in line with 2023 Article IV Report assumptions).
    2. Case B: 200-percent increase in NPLs.
    3. Case C: 440-percent increase in NPLs (would bring NPL ratio back to 2014 historical high).
  - Provisioning rate assumed: 65 percent.
  - NPL increase distributed proportionally to current bank-level NPL shares.
- Large exposure risk:
  - Two scenarios:
    1. Failure of the largest borrower (Top 1).
    2. Failure of the top two largest borrowers (Top 2).

### Credit risk stress test results (capital adequacy)
- General credit risk results (capital adequacy ratio, percent):
  - Current weighted average CAR: 19.07
  - Case A (100% increase in NPLs): weighted avg CAR: 17.08
  - Case B (200% increase in NPLs): weighted avg CAR: 15.06
  - Case C (440% increase in NPLs): weighted avg CAR: 9.79
- Capital shortfall and injections:
  - Under Case A (100% NPL increase): all banks remain above minimum; average CAR around 17 percent.
  - Under Case B (200%): average CAR remains above 12 percent; three banks (including one D-SIB) experience capital shortfalls; required capital injection minimal at 0.2 percent of GDP.
  - Under Case C (440%): multiple banks, including three D-SIBs, experience significant capital shortfalls; required capital injection around 1.5 percent of GDP.
- Large exposure results:
  - System-wide CAR falls from around 19 percent to 15 percent if the largest borrower defaults.
  - If the two largest borrowers default, number of banks with capital shortfalls increases to six (including one D-SIB), with system-wide CAR falling almost below 12 percent.
  - Capital injection needed under the Top 2 scenario: 0.5 percent of GDP.

### Market risk stress test: scenarios and channels
- Interest rate risk:
  - Channels:
    - Flow channel: changes in interest income and interest expenses from the gap of interest-sensitive assets and liabilities.
    - Stock channel: impact of interest changes on marked-to-market (MtM) bond values.
  - Assumption: 25 percent of bonds in banks’ portfolios are MtM, rest held-to-maturity (HtM).
  - Scenarios:
    1. Baseline: 150-basis point (bps) increase in interest rate.
    2. Extreme: 250-bps increase.
- Exchange rate risk:
  - Channels:
    - Direct: changes in local currency value of banks’ net FX positions.
    - Indirect: exchange rate movements affecting corporates’ and households’ creditworthiness and NPLs.
  - Scenarios:
    1. Baseline: 25-percent depreciation of the lek vis-à-vis the euro (would bring the exchange rate roughly to the pandemic-era level).
    2. Extreme: 40-percent depreciation (would bring the exchange rate to its pre-pandemic level).

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

### Interest rate risk: main findings
- Under both baseline and more extreme scenarios, banks’ capital ratio is found to be little affected by interest rate shocks.
- Two channels operate in opposite directions:
  - Income (repricing) channel: banks generally have more interest-sensitive assets than interest-sensitive liabilities and therefore appear to benefit from an interest rate increase.
  - Repricing (mark-to-market) channel: an interest rate rise leads to a fall in the MtM value of bonds in banks’ portfolios, reducing their capital ratio.
- Under the baseline assumption that only 25 percent of banks’ bond portfolio is MtM, the quantitative effects of the two channels almost completely offset each other.
- If all bonds in banks’ portfolios are instead assumed to be MtM, the repricing channel produces a sizable negative effect such that the overall impact of an interest rate rise is negative and sizable.
- Under the baseline scenario of a 150-bps increase:
  - All banks remain capitalized above the minimum requirement.
  - Two banks (including one D-SIB) are found to experience a mild shortfall despite the system-wide CAR remaining at 15.4 percent.
- Note: Stress test scenarios include 1) a 150 bps interest rate increase, and 2) a 250 bps interest rate increase.

### Interest rate risk: illustrative statistics
- Current CAR weighted average: 19.07
- CAR after 150 bps: 19.24
- CAR after 250 bps: 19.35

### Exchange rate risk: main findings
- There is some vulnerability to exchange rate shocks.
- Under the baseline scenario of a 25-percent lek depreciation against the euro:
  - The system-wide CAR drops by around 3 percentage points.
  - One bank falls short of the required capital ratio, although it remains well above the 12-percent requirement.
- Under a more extreme scenario of a 40-percent depreciation:
  - The system-wide CAR remains above the minimum requirement.
  - Three banks experience capital shortfalls; none of these banks is systemically important.
- Decomposition of the total impact indicates the direct channel is slightly more significant, responsible for around 60 percent of the reduction in the system-wide capital ratio.

### Exchange rate risk: illustrative statistics
- Exchange Rate Risk - Capital Adequacy Ratio:
  - Current weighted average CAR: 19.07
  - After 25% depreciation weighted average CAR: 16.01
  - After 40% depreciation weighted average CAR: 14.43

### Liquidity resilience and LCR: context and methodology
- LCR was introduced under Basel III to ensure banks hold adequate HQLA to meet liquidity needs over a 30-calendar day stress scenario.
- HQLA must be convertible into cash easily and immediately with minimal price impact in secondary markets.
- The analysis focuses on LCR in lek; foreign-currency-denominated HQLAs (for example, euro bonds) are deemed appropriately liquid.
- Albanian government securities are treated with a 0 percent haircut and constitute the lion’s share of Albanian banks’ HQLA in lek due to large holdings amid a lack of other financial instruments.
- The analysis uses the March 2024 LCR dataset provided by the Bank of Albania (BoA).

### Liquidity stress scenarios examined
- Scenario A: Access to BoA’s liquidity facility with an assumed haircut of 20 percent on lek-denominated sovereign government bonds.
- Scenario B: No access to BoA’s liquidity facility with an assumed haircut of 90 percent on lek-denominated sovereign government bonds (bank would need to liquidate in secondary market subject to 90-percent haircut).
- Rationale: 90-percent haircut scenario represents absence of access to BoA facility (e.g., concurrent liquidity and solvency stress) and reflects the illiquid secondary bond market; the assumed liquidation amount remains well above actual 30-day turnover of the secondary market.

### Liquidity resilience: main findings
- Under the current LCR assumption of a zero haircut:
  - Almost all banks (except for one outlier) maintain a very comfortable LCR in lek well above 100 percent.
  - System-weighted average LCR: 302.82 percent.
- Under the 20-percent haircut assumption (access to BoA facility):
  - System-weighted average LCR falls marginally to 265.89 percent.
  - No additional bank experiences a liquidity shortfall.
- Under the 90-percent haircut assumption (no access to BoA facility):
  - System-weighted average LCR drops to 75.03 percent, below the minimum 100 percent requirement.
  - Eight banks experience a liquidity shortfall, including all four D-SIBs.
- Implication: Banks’ ability to meet liquidity needs under stress depends crucially on access to the BoA’s facility in the absence of a liquid secondary bond market.

### Policy implications and recommended actions
- Continued efforts to develop the secondary bond market and the secured money market are important to improve market-based resilience to liquidity risks.
- Ensuring reliable access to the BoA’s liquidity facility is critical for system-wide short-term liquidity resilience given the current market structure and LCR composition.

*Source: IMF staff calculations and Bank of Albania data as presented in the chapter.*

### 1. Exchange Rate, Foreign Reserves, and Foreign Exchange Intervention ___________ 3

### FOREIGN EXCHANGE INTERVENTION THROUGH THE LENS OF THE QUANTITATIVE INTEGRATED POLICY FRAMEWORK: THE CASE OF ALBANIA

### Context and key facts
- Albania is a small open emerging market with an inflation targeting regime and a sustained appreciation trend of the Albanian lek amid a tourism boom.
- Bank of Albania (BoA) foreign exchange intervention (FXI) during the first 9 months of 2024 amounted to 2.6 percent of 2023 nominal GDP, almost triple the amount of the same period in 2023.
- Albania’s OTC FX market turnover is estimated at 11 billion USD yearly, compared with an emerging market average of 52 billion USD.
- Pass-through evidence: cumulative impact of about 0.2 percent (for a 1-percent appreciation) on headline CPI over four quarters (empirical analysis following Gopinath (2015)).

### Measures of market liquidity and balance-sheet exposures
- FX market shallowness
  - High volatility in UIP premia (especially since Q1 2023) and low turnover indicate shallow FX markets.
  - Shallow markets imply financial intermediaries have limited capacity to absorb excess currency demand/supply, increasing exchange rate deviations from fundamentals.
- Currency mismatches and hedging
  - Share of unhedged FX loans declined from 50 percent of total loans in 2014 to 25 percent in 2024.
  - Two-thirds of unhedged FX loans in 2024 are accounted for by the real estate sector.
- Inflation expectations
  - Inflation expectations are described as well-anchored in the medium and long run.
  - There is room for monetary policy to support a sustainable return of inflation to target without FXI.

### QIPF (Quantitative Integrated Policy Framework) model — specification and estimation
- Model features
  - A New Keynesian open-economy QIPF model adapted for Albania, linearized following Chen et al. (2023) and extended to include trend productivity growth and a domestic financial sector.
  - Assumes FX intermediaries have limited risk-bearing capacity (as in Gabaix and Maggiori, 2015), which generates realistic exchange rate volatility and implies sterilized FX interventions have real effects through UIP premia.
- Estimation details
  - Estimated on a dozen quarterly series for Albania over 2009–2024 using Bayesian maximum likelihood.
  - Euro area block estimated separately; euro area variables included to capture foreign influences.
- Key estimated properties
  - Pass-through of nominal exchange rate movements to consumer prices is gradual due to local currency pricing-to-market behavior.
  - Interest rate policy gradually adjusts to deviations of expected one-year ahead core inflation from target and output deviations from trend.
  - Foreign exchange market is estimated to be often shallow, implying FX interventions can have sizeable macroeconomic effects when domestic policy rate is passive.

### Scenario analysis — Tourism-driven fundamental shock
- Scenario design
  - Stylized tourism boom driven by fundamentals: productivity (catch-up trend and temporary), relative export prices, and domestic demand shocks phased in gradually as MIT shocks.
  - Central bank modeled as gradually learning; does not have perfect foresight and initially understates the rise in potential output.
- Baseline vs alternative monetary policy responses
  - Baseline: monetary policy follows estimated historical rule; monetary policy eases because appreciation puts downward pressure on import prices and productivity lowers unit costs, causing inflation to fall below target.
  - Alternative (discretionary) policy: additional policy rate cuts assumed to offset the appreciation in the lek, narrowing the output gap and bringing inflation closer to target while avoiding volatile interest rate movements.
  - Illustrative welfare metric: the alternative policy delivers a 25 percent improvement over the historical policy response measured by a standard central bank quadratic objective in output gap and inflation (note: still leaves output gap negative and inflation below target, implying scope for even more stimulative policy).

- Policy implication from the tourism scenario
  - Monetary policy is a suitable tool to address fundamental appreciation pressures from a tourism boom; FXI is not necessary under this fundamental shock given monetary policy retains significant traction over exchange rates.

### Scenario analysis — Non-fundamental (risk-on) capital flow shocks and trade-offs
- Role of non-fundamental shocks
  - Non-fundamental (“risk-on”) shocks can appreciably affect the real exchange rate in the presence of shallow FX markets by widening the UIP “wedge.”
  - Estimated contributions to long-run RER variance:
    - Around 5 percent if FXI is reacting to exchange rate movements.
    - Around 15 percent assuming no FX interventions are used to lean against exchange rate movements.
  - Cumulative contribution since 2021: non-fundamental shocks appear to have appreciated the lek by around 11 percent (subject to estimation and specification uncertainty).
- Policy trade-offs
  - Risk-driven exchange rate volatility can create a trade-off between output and inflation: an appreciating exchange rate from risk-on inflows can lower inflation while producing a persistent rise in output.
  - Interest rate policy primarily moves inflation and output in the same direction via aggregate demand and thus is not ideally suited to mitigate shocks that imply opposite movements in inflation and output.
  - In such cases, FXI may be useful to lean against temporary, non-fundamental capital in- and outflows, especially given market shallowness and limited intermediary capacity.

### Overall findings and policy recommendations
- Primary IPF friction in Albania: shallow FX markets that can justify FXI to moderate excessive exchange rate movements arising from temporary non-fundamental shocks.
- FX mismatches and de-anchored inflation expectations are broadly contained; remaining FX mismatch risks (notably in real estate) are better addressed by preemptive and targeted macroprudential measures (MPMs) rather than FXI.
- For fundamental shocks (e.g., tourism-driven productivity and demand changes), monetary policy is the appropriate tool; FXI is not warranted.
- For non-fundamental, sentiment-driven capital flow shocks in the context of shallow FX markets, FXI can be a useful complementary tool to lean against excessive exchange rate movements, given the limitations of interest rate policy to address the output–inflation trade-off these shocks create.
- Policymakers face real-time challenges in recognizing shifts in potential output and distinguishing fundamental from non-fundamental shocks; the QIPF model can help quantify trade-offs and guide the appropriate policy mix.

*Prepared by David Bartolini, Jakree Koosakul, Rebecca Huang, Jesper Linde, and Roland Meeks (IMF staff).*

### 14. Adverse impacts from risk shocks can be addressed using FXI. A better tool to address

### 14. Adverse impacts from risk shocks can be addressed using FXI. A better tool to address

### FX intervention (FXI): modeling and key findings
- Sterilized FX intervention (FXI) is modeled as an endogenous reaction to changes in the nominal exchange rate combined with gradual mean-reversion to restore the desired ratio of reserves to GDP.
- Under an inflow shock that makes the euro temporarily cheap, the FX rule calls for euro purchases, which:
  - Greatly reduces the lek exchange rate appreciation;
  - Eliminates the output-inflation trade-off observed without FXI;
  - Suggests that an easing of interest rate policy coordinated with FXI could realize greater stabilization benefits.
- Sensitivity: results depend on the strength of the FX policy response to the exchange rate and on which historical intervention data are used in the QIPF model; across alternatives FXI remains an effective tool for addressing risk shocks, though quantitative responses vary.
- Figure 6 (IMF staff estimates) contrasts central bank responses “With FXI” and “Without FXI” for:
  - Policy Rate (APR)
  - Output (% trend development)
  - Real Exchange Rate (Consumption based)
  - CPI Inflation (APR)

### Conclusion and policy considerations on exchange rate role
- The exchange rate should continue to play its role as a shock absorber with FXI reserved to moderate the impact of non-fundamental shocks.
- Empirical application of the quantitative IPF model to Albania indicates:
  - Appreciation of the lek has historically been driven largely by fundamentals;
  - Non-fundamental factors have played a modest role in some episodes.
- Policy recommendations:
  - Consider letting the exchange rate adjust more flexibly and rely on interest rate policy as the primary tool for price stability.
  - Use FX interventions in cases of non-fundamental shocks to lower output and inflation volatility.
  - Internalize potential adverse consequences of FXI and further reserve accumulation, including on risks to the central bank balance sheet, interest rate transmission, and financial market development.

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

### Banking sector context and resilience (overview)
- Structural and performance indicators:
  - NPLs declined to just under 5 percent of total loans in Q1 2024.
  - Foreign ownership share of assets fell from 90 percent in 2013 to 65 percent in Q1 2024.
  - Return-on-assets (ROA) averaged around 1 percent between 2019 and 2022, rising to an average of 1.7 percent in 2023 and 2024.
  - Average capital adequacy ratio of 19.3 percent in July 2024 (minimum regulatory requirement: 12 percent).
  - The banking sector is predominantly bank-based: 11 commercial banks own over 90 percent of financial sector assets.
  - Four banks identified as D-SIBs, owning 65 percent of banks’ assets and 66 percent of banks’ capital.
  - Herfindahl-Hirschman Index (HHI) of 1428 based on Q1:2024 asset data (market considered competitive).

- Remaining vulnerabilities highlighted:
  - High sovereign securities holdings and sovereign-bank nexus risks.
  - Local-currency government bond and money markets remain illiquid.
  - High euroization of the economy; two-thirds of unhedged FX loans concentrated in the real estate sector, which saw rapid credit growth in 2024.

### Stress testing approach (capital and liquidity)
- Methodology:
  - Bottom-up balance-sheet approach per Cihak (2014) using bank-by-bank balance sheet data for Q1:2024.
  - Horizon: one year.
  - Banks considered undercapitalized only if CAR falls below 12 percent.
- Risks covered:
  - Credit risks: NPL-related shocks and large-exposure shocks.
  - Market risks: interest rate and exchange rate channels.
  - Liquidity resilience assessed via an exercise varying haircuts on liquid assets under the LCR standard.

### Credit risk stress test scenarios and assumptions
- General credit risk (NPL shock):
  - Three scenarios:
    1. Case A: 100-percent increase in NPLs (baseline, in line with 2023 Article IV Report assumptions).
    2. Case B: 200-percent increase in NPLs.
    3. Case C: 440-percent increase in NPLs (would bring NPL ratio back to 2014 historical high).
  - Provisioning rate assumed: 65 percent.
  - NPL increase distributed proportionally to current bank-level NPL shares.

- Large exposure risk:
  - Two scenarios:
    1. Failure of the largest borrower (Top 1).
    2. Failure of the top two largest borrowers (Top 2).

### Credit risk stress test results (capital adequacy)
- General credit risk results (capital adequacy ratio, percent):
  - Current weighted average CAR: 19.07
  - Case A (100% increase in NPLs): weighted avg CAR: 17.08
  - Case B (200% increase in NPLs): weighted avg CAR: 15.06
  - Case C (440% increase in NPLs): weighted avg CAR: 9.79
- Capital shortfall and injections:
  - Under Case A (100% NPL increase): all banks remain above minimum; average CAR around 17 percent.
  - Under Case B (200%): average CAR remains above 12 percent; three banks (including one D-SIB) experience capital shortfalls; required capital injection minimal at 0.2 percent of GDP.
  - Under Case C (440%): multiple banks, including three D-SIBs, experience significant capital shortfalls; required capital injection around 1.5 percent of GDP.

- Large exposure results:
  - System-wide CAR falls from around 19 percent to 15 percent if the largest borrower defaults.
  - If the two largest borrowers default, number of banks with capital shortfalls increases to six (including one D-SIB), with system-wide CAR falling almost below 12 percent.
  - Capital injection needed under the Top 2 scenario: 0.5 percent of GDP.

### Market risk stress test: scenarios and channels
- Interest rate risk:
  - Channels:
    - Flow channel: changes in interest income and interest expenses from the gap of interest-sensitive assets and liabilities.
    - Stock channel: impact of interest changes on marked-to-market (MtM) bond values.
  - Assumption: 25 percent of bonds in banks’ portfolios are MtM, rest held-to-maturity (HtM).
  - Scenarios:
    1. Baseline: 150-basis point (bps) increase in interest rate.
    2. Extreme: 250-bps increase.

- Exchange rate risk:
  - Channels:
    - Direct: changes in local currency value of banks’ net FX positions.
    - Indirect: exchange rate movements affecting corporates’ and households’ creditworthiness and NPLs.
  - Scenarios:
    1. Baseline: 25-percent depreciation of the lek vis-à-vis the euro (would bring the exchange rate roughly to the pandemic-era level).
    2. Extreme: 40-percent depreciation (would bring the exchange rate to its pre-pandemic level). 

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

### 14.      Banks are resilient to interest rate shocks, although vulnerability emerges if the

### 14.      Banks are resilient to interest rate shocks, although vulnerability emerges if the baseline MtM assumption is relaxed.

### Interest rate risk: main findings
- Under both baseline and more extreme scenarios, banks’ capital ratio is found to be little affected by interest rate shocks.
- Two channels operate in opposite directions:
  - Income (repricing) channel: banks generally have more interest-sensitive assets than interest-sensitive liabilities and therefore appear to benefit from an interest rate increase.
  - Repricing (mark-to-market) channel: an interest rate rise leads to a fall in the MtM value of bonds in banks’ portfolios, reducing their capital ratio.
- Under the baseline assumption that only 25 percent of banks’ bond portfolio is MtM, the quantitative effects of the two channels almost completely offset each other.
- If all bonds in banks’ portfolios are instead assumed to be MtM, the repricing channel produces a sizable negative effect such that the overall impact of an interest rate rise is negative and sizable.
- Under the baseline scenario of a 150-bps increase:
  - All banks remain capitalized above the minimum requirement.
  - Two banks (including one D-SIB) are found to experience a mild shortfall despite the system-wide CAR remaining at 15.4 percent.
- Note: Stress test scenarios include 1) a 150 bps interest rate increase, and 2) a 250 bps interest rate increase.

### Interest rate risk: illustrative statistics (from figures)
- Current CAR weighted average: 19.07
- CAR after 150 bps: 19.24
- CAR after 250 bps: 19.35
- Interest Rate Risk - CAR Impact Decomposition contrasts:
  - 25% MTM versus All MTM scenarios show shifts in contributions from the Income Channel and Repricing Channel (percent impacts shown in figure).

### Exchange rate risk: main findings
- There is some vulnerability to exchange rate shocks.
- Under the baseline scenario of a 25-percent lek depreciation against the euro:
  - The system-wide CAR drops by around 3 percentage points.
  - One bank falls short of the required capital ratio, although it remains well above the 12-percent requirement.
- Under a more extreme scenario of a 40-percent depreciation:
  - The system-wide CAR remains above the minimum requirement.
  - Three banks experience capital shortfalls; none of these banks is systemically important.
- Decomposition of the total impact indicates the direct channel is slightly more significant, responsible for around 60 percent of the reduction in the system-wide capital ratio.

### Exchange rate risk: illustrative statistics (from figures)
- Exchange Rate Risk - Capital Adequacy Ratio:
  - Current weighted average CAR: 19.07
  - After 25% depreciation weighted average CAR: 16.01
  - After 40% depreciation weighted average CAR: 14.43
- Exchange Rate Risk - Undercapitalized Banks (number):
  - Non-systemic and Systemic bank counts shown across Current, 25% depreciation, and 40% depreciation scenarios in figure.

### Liquidity resilience and LCR: context and methodology
- LCR was introduced under Basel III to ensure banks hold adequate HQLA to meet liquidity needs over a 30-calendar day stress scenario.
- HQLA must be convertible into cash easily and immediately with minimal price impact in secondary markets.
- The analysis focuses on LCR in lek; foreign-currency-denominated HQLAs (for example, euro bonds) are deemed appropriately liquid.
- Albanian government securities are treated with a 0 percent haircut and constitute the lion’s share of Albanian banks’ HQLA in lek due to large holdings amid a lack of other financial instruments.
- The analysis uses the March 2024 LCR dataset provided by the Bank of Albania (BoA).

### Liquidity stress scenarios examined
- Scenario A: Access to BoA’s liquidity facility with an assumed haircut of 20 percent on lek-denominated sovereign government bonds.
- Scenario B: No access to BoA’s liquidity facility with an assumed haircut of 90 percent on lek-denominated sovereign government bonds (bank would need to liquidate in secondary market subject to 90-percent haircut).
- Rationale: 90-percent haircut scenario represents absence of access to BoA facility (e.g., concurrent liquidity and solvency stress) and reflects the illiquid secondary bond market; the assumed liquidation amount remains well above actual 30-day turnover of the secondary market.

### Liquidity resilience: main findings
- Under the current LCR assumption of a zero haircut:
  - Almost all banks (except for one outlier) maintain a very comfortable LCR in lek well above 100 percent.
  - System-weighted average LCR: 302.82 percent.
- Under the 20-percent haircut assumption (access to BoA facility):
  - System-weighted average LCR falls marginally to 265.89 percent.
  - No additional bank experiences a liquidity shortfall.
- Under the 90-percent haircut assumption (no access to BoA facility):
  - System-weighted average LCR drops to 75.03 percent, below the minimum 100 percent requirement.
  - Eight banks experience a liquidity shortfall, including all four D-SIBs.
- Implication: Banks’ ability to meet liquidity needs under stress depends crucially on access to the BoA’s facility in the absence of a liquid secondary bond market.

### Policy implications and recommended actions
- Continued efforts to develop the secondary bond market and the secured money market are important to improve market-based resilience to liquidity risks.
- Ensuring reliable access to the BoA’s liquidity facility is critical for system-wide short-term liquidity resilience given the current market structure and LCR composition.

*Source: IMF staff calculations and Bank of Albania data as presented in the chapter.*

---


_Source: https://www.imf.org/-/media/files/publications/cr/2025/english/1albea2025002-print-pdf.pdf_