## wpiea2024080-print-pdf ## Source details **Canonical URL:** [wpiea2024080-print-pdf](https://www.imf.org/-/media/files/publications/wp/2024/english/wpiea2024080-print-pdf.pdf) ## Other formats - [Markdown version](/-/media/files/publications/wp/2024/english/wpiea2024080-print-pdf.pdf.md) - [Structured JSON version](/-/media/files/publications/wp/2024/english/wpiea2024080-print-pdf.pdf.json) --- ### Major findings and framing - March 2023 bank failures (Silicon Valley Bank, Signature Bank) and loss of market confidence in Credit Suisse exposed interactions between tighter monetary/financial conditions and built-up vulnerabilities after a decade of ample liquidity. - Markets expect policy rates to remain elevated for a prolonged period; the entire OIS curve has moved up substantially since December 2021. - A “higher-for-longer” interest rate environment could: - Challenge banks’ risk management in securities portfolios and loan exposures. - Slow economic momentum if central banks keep rates higher due to persistent inflation. - Increase corporate bankruptcies after pandemic-era forbearance. - Concerns for MENA: - Vulnerabilities may be hidden by accounting/regulatory treatments. - Concentration in certain asset classes (e.g., government bonds) and reliance on foreign funding are key vulnerabilities. - Interest-rate-sensitive borrowers may face debt servicing challenges, affecting profitability, provisioning, credit provision, and growth. ### Research questions and stress-test design - Three core questions: 1. Impact of a higher-for-longer interest rate environment on corporate sectors (debt servicing, interest coverage ratios, zombie credit). - Corporate stress scenario: a 100-basis point (bps) increase in interest rates for two subsequent years and a negative profitability shock affecting more than 1,000 firms in MENA. - Outcome inference: increase in NPLs inferred from deterioration in distribution of interest coverage ratios (ICRs). 2. Resilience of banks under higher-for-longer rates. - Banking stress test quantifies losses under liquidity and credit quality shocks and their interaction. - Estimates impact of rising interest rates on banks’ net interest margins following Schnabl and others (2021). - Recognizes offsetting potential from increasing net interest margins for banks whose assets reprice faster than liabilities. 3. Macro effects of banking sector stress. - Uses local projections (Jordà 2005) to estimate link between bank capital and credit provision. - Uses a small-scale macro-econometric model to estimate elasticity of output to changes in credit provision and simulate counterfactual output paths under banking stress. ### Key quantitative results (summary) - Corporate sector: - Zombie-held debt among publicly listed firms rises from 12 percent to 30 percent in a higher-for-longer scenario. - Share of zombie firms increases from about 12 percent in 2022 to almost 30 percent by end-2024 in the adverse scenario. - Sectors relatively more affected: transportation, capital goods, food and beverage. - Banking sector: - In the joint stress scenario (higher rates, corporate stress, liquidity pressures): - Tier-1 capital ratios decline by 2.3 percentage points in the GCC. - Tier-1 capital ratios decline by 4-percentage points in non-GCC MENA. - While few banks become undercapitalized, lower bank capital could limit lending and add downside risks to output. - Macro outcomes: - Average output declines in the adverse banking sector stress scenario: - GCC: 0.4 percent contraction. - Non-GCC MENA: 0.7 percent contraction. ### Data, sample, and representativeness - Bank-level data: - Source: Fitch Connect (consolidated financial statements). - Final sample: 108 banks from 10 countries (Bahrain, Egypt, Jordan, Kuwait, Morocco, Oman, Pakistan, Qatar, Saudi Arabia, United Arab Emirates). - Coverage: at least 78 percent of banking assets in each country. - Regional averages (Table 1): - Total assets (US$, million): GCC 49,545.66; MENA (excl. GCC) and Pakistan 17,143.61 - Tier-1 capital ratio (%): GCC 17.5; MENA (excl. GCC) and Pakistan 14.78 - Return on assets (RoA) (%): GCC 1.82; MENA (excl. GCC) and Pakistan 1.23 - Loan share: GCC 0.61; MENA (excl. GCC) and Pakistan 0.48 - Government bond share: GCC 0.12; MENA (excl. GCC) and Pakistan 0.29 - Cash share: GCC 0.08; MENA (excl. GCC) and Pakistan 0.07 - Deposit share: GCC 0.74; MENA (excl. GCC) and Pakistan 0.75 - NPL ratio (in %): GCC 5.24; MENA (excl. GCC) and Pakistan 5.8 - Provisioning ratio (in %): GCC 1.33; MENA (excl. GCC) and Pakistan 1.3 - N (number of banks): GCC 61; MENA (excl. GCC) and Pakistan 47 - Stylized comparisons: - GCC banks average assets $49.5 billion versus $17.1 billion for remaining MENA and Pakistan. - GCC Tier-1 capital ratios 17.5 percent versus 14.8 percent in non-GCC; RoA 1.82 percent versus 1.23 percent. - GCC loan share 61 percent vs. non-GCC loan share 48 percent; non-GCC higher sovereign bond holdings (29 percent). - Average bank provisioning ratio ~130 percent across regions. - Firm-level data: - Source: S&P’s Capital IQ. - Coverage: more than 1,000 publicly listed firms from 15 MENA countries and Pakistan. - 2022: ~400 firms from GCC; ~700 firms from rest of MENA and Pakistan. - Pakistan, Saudi Arabia, Egypt: each have over 100 firms. - Jordan, Kuwait, Oman, Morocco, Tunisia: 50–100 firms each. - Median firm (2022): - Total assets: about $130 million overall; median GCC firm ~7 times larger than median non-GCC firm. - EBITDA: median GCC firm about $23 million; median non-GCC firm $5 million. - Corporate sector historical trends: - Median return on assets fell from about 7.5 percent in 2006 to 5.5 in 2019. - Median leverage (total liabilities to assets) around 20 percent; one-quarter of firms have leverage exceeding 40 percent. - Median maturity of long-term corporate debt stable between three and four years; one-quarter of firms have debt duration < two years. ### Corporate sector: interest rates, buffers, and zombification - Interest-rate path and assumptions: - Median effective interest rate historically near 6 percent; peaked at 7.5 percent in 2009; bottomed near 5 percent in 2017. - Adverse scenario: effective interest rate increase of 100 bps per year for each firm over 2023–24. - Effective interest rates increase from about 6 percent at end-2022 to more than 8 percent by end-2024. - Corporate liquidity and buffers: - Median firm holds about 4 percent of total assets in cash and cash equivalents. - Some firms hold liquid buffers closer to 10 percent of total assets. - Zombie definition and metrics: - Zombies: firms with average ICR below 2.5 over two years that simultaneously receive “subsidized” lending (effective interest rate below that of top-rated firms). - Share of zombie firms in 2022: about 12 percent. - Zombie firms’ median leverage at end-2022: 40 percent versus 20 percent for other firms. - At end-2022, zombie firms held about 12 percent of corporate debt. - Stress-test results (corporate): - Firm profitability could decline, on average, to about 3 percent in 2024 (from 5 percent pre-pandemic). - Median ICR estimated to decline from 3.5 to 1.5 (reported in conclusions). - Share of debt at risk of default rises from about 12 percent of total debt in 2022 to almost 30 percent by 2024. - Sectoral vulnerabilities: transportation, capital goods, food and beverage; food and beverages sector’s median share of zombie firm debt increases particularly sharply by 2024. - Alternative stress scenario calibrated to 2014–15 oil price shock: zombie-held debt increases to 25 percent of total debt by 2024; interquartile range 18-58. ### Banking sector stress test: scope, methodology, and calibration - Three-step stress-test approach: 1. Estimate sensitivity of bank profitability to interest rate changes (RoA betas). 2. Simulate a liquidity shock through sudden withdrawal of deposits and non-deposit funding. 3. Map corporate sector stress scenarios into increasing NPLs and simulate impact on bank balance sheets. - RoA beta estimation: - RoA beta defined as βiRoA = ∑훽i,휏RoA (휏=0 to 3); uses average RoA over last four quarters. - Mean RoA beta: 0.097; standard deviation: 0.37. - About one-third of banks have a negative RoA beta. - Banks with lower NPL ratios, more held-to-maturity securities, a lower share of wholesale funding, or higher capital ratios have higher RoA betas on average. - For banks with <40 quarterly observations, betas set equal to a weighted average of country peers. - Liquidity stress design: - Withdrawal assumptions: - 20 percent withdrawal of resident deposits. - 30 percent withdrawal of foreign deposits and wholesale funding. - Withdrawal per bank formula: - Withdrawal_b = .2 LC Deposits_b + .3 (FX Deposits_b + Wholesale Funding_b) - Asset liquidation pecking order: pre-existing cash → government bonds marked-to-market → government bonds not yet market-to-market → other securities → illiquid assets like loans. - Loss assumptions: - Additional 30 percent discount if loans or other illiquid assets sold to meet withdrawals. - Fire-sale discounts for securities simulated via 1,000 draws calibrated to increase in sovereign spreads in 2020; medians reported. - Credit stress mapping: - Deterioration in corporate ICRs mapped into default probabilities using Damodaran (2023). - Implied increase in NPLs: 2.1 percent of net loans for 2023 and 2.7 percent for 2024. - Banks provision 100 percent against the increase in NPLs (Kuwait proviso: increase in provisions is the minimum of the increase in NPLs and amount needed to maintain provisioning ratio of at least 150 percent). ### Scenarios analyzed - Four scenarios: 1. Liquidity stress: 20 percent deposit withdrawal with 30 percent withdrawal of foreign deposits and wholesale funding. 2. Liquidity stress amid higher interest rates: scenario 1 amid 200 basis points higher interest rates causing capital losses on fixed income assets. 3. Credit stress: rise in corporate sector NPLs consistent with corporate stress test, with increasing NPLs 100 percent provisioned. 4. Combined liquidity and credit stress amid higher interest rates: combination of scenarios 2 and 3. ### Banking sector results and heterogeneity - Regional and subregional results: - Many banks benefit from higher rates, but about one-third are adversely affected (negative RoA betas). - Banking systems in most countries could withstand isolated liquidity or corporate-sector stress amid higher interest rates. - Combined scenario (higher rates, corporate-sector stress, liquidity stress) could materially stress banking systems, particularly non-GCC countries. - Quantitative outcomes (examples quoted in source): - GCC losses: range from 0.3 percent of capital in isolated liquidity stress to 13.4 percent of Tier-1 capital in combined stress. - For GCC, weighted average Tier-1 capital ratio would drop from 17.1 percent to 14.8 percent (text fragments indicate decline; full end-point given elsewhere as 14.8 percent, significantly above Basel-III minima of 8.5 percent). - For non-GCC MENA and Pakistan: - Liquidity shock under higher interest rates: 12.8 percent of Tier-1 capital. - Isolated corporate sector stress: 6 percent of Tier-1 capital. - Combined liquidity and corporate sector stress amid higher interest rates: 29 percent of Tier-1 capital. - Undercapitalization shares (asset-weighted): - GCC: 1.2 percent become undercapitalized under most severe scenario using Basel-III minima; 1.9 percent using country-specific minima. - MENA (excluding GCC) and Pakistan: 20.2 percent undercapitalized using Basel-III minima; 23.1 percent using country-specific minima. - Recapitalization needs (most severe scenario using country-specific minima): - GCC: $0.9 billion. - Non-GCC MENA: $4.8 billion. - Heterogeneity and drivers: - State-owned banks defined as ≥50 percent government ownership. - Weighted average losses: - GCC privately owned banks: 10.3 percent of capital. - GCC state-owned banks: 18.1 percent of capital. - MENA (excluding GCC) and Pakistan privately owned banks: 22.9 percent of capital. - MENA (excluding GCC) and Pakistan state-owned banks: 26.5 percent of capital. - Median-split vulnerabilities: - Below-median capitalization: 18 percent average losses of capital; above-median capitalization: 6 percent. - Banks with above-median duration risk: average losses 32 percent of capital; below-median duration risk: 9 percent. - Larger banks (above-median market share) average losses: 18 percent of capital. - Losses concentrated in banks with below-median liquid assets, lower ex-ante provisioning, and weak profitability. ### Macro outcomes and linkages - Bank capital → lending: - A 1-percentage-point lower Tier-1 capital ratio is associated with 1.5 percentage points lower real lending after 10–12 quarters (statistically significant from five quarters onward). - Caution: effect is not strictly causal due to potential simultaneity. - Output implications (small-scale macroeconometric model calibrated with micro estimates): - Output contraction under combined banking sector stress scenario: - MENA: 0.7 percent contraction. - GCC: 0.4 percent contraction. - Country ranges: - MENA (excluding GCC) and Pakistan: potential output losses range from 0.2 to 1.2 percent. - GCC: potential output contraction range between 0.05 and 1.2 percent. - Note: estimates likely understate actual output losses because several general equilibrium channels and fiscal impacts from recapitalization are not directly modeled. ### Policy implications and recommendations - Policymakers should be attentive to financial stability risks in a higher-for-longer interest rate environment. - Macroprudential measures: - Consider broader use of macroprudential tools; countercyclical capital buffer (CCyB) use remains subdued in MENA. - Preserve bank capital where banking systems face elevated interest rate risk to maintain loss absorption capacity. - Conduct bank stress tests with explicit focus on sovereign-bank nexus. - Over the medium term, develop alternative funding sources for domestic sovereign bonds to diversify investor base and reduce sovereign-bank nexus. - Bank resolution and NPLs: - Avoid loan evergreening; resolution regimes should facilitate swift resolution of NPLs to prevent build-up of impaired legacy assets and emergence of persistent zombie firms. - Liquidity and foreign funding: - Address banks’ foreign currency liquidity risks where relevant; consider measures to reduce reliance on nonresident deposits and wholesale funding. ### Methodological contributions and annex highlights - Corporate stress tests draw firm profitability shocks for 2023–2024 from sector-specific distributions calibrated to the Global Financial Crisis and, as robustness, the 2015 oil shock; standard deviations capped at 50 percent. - Annex II: Distribution of estimated RoA betas; about 1/3 of banks have negative RoA betas; bank characteristics correlated with higher RoA betas include lower NPL ratios, higher HTM share, lower wholesale funding, higher Tier-1 ratios. - Annex III: Liquidity stress-test mechanics, valuation proxies, fire-sale discount calibration (based on sovereign spread increases Feb–Jul/Sep 2020), sale sequencing (AFS first, then HTM), and excess loss accounting. - Annex IV: Two-equation macro model linking credit and output: - Credit equation: c_t = ρ_1 c_{t−1} + ρ_2 y_{t−1} + ε^c_t - Output equation: y_t = α_1 y_{t−1} + α_2 ε^c_t + ε^y_t - Quarterly data from Haver Analytics 2001Q4–2023Q1; trends calibrated with HP filter (1600). *Source: wpiea2024080-print-pdf; https://www.imf.org/-/media/files/publications/wp/2024/english/wpiea2024080-print-pdf.pdf* ### References ............................................................................................................. ### wpiea2024080-print-pdf - References ............................................................................................................. ### Major findings and framing (Introduction) - March 2023 bank failures (Silicon Valley Bank, Signature Bank) and loss of market confidence in Credit Suisse exposed interactions between tighter monetary/financial conditions and built-up vulnerabilities after a decade of ample liquidity. - Markets expect policy rates to remain elevated for a prolonged period and at much higher levels than expected two years prior; the entire OIS curve has moved up substantially since December 2021 (Figure 1: US Federal Funds Overnight Indexed Swaps Curve). - A “higher-for-longer” interest rate environment could: - Challenge banks’ risk management in securities portfolios and loan exposures. - Slow economic momentum if central banks keep rates higher due to persistent inflation. - Increase corporate bankruptcies after pandemic-era forbearance. - Concerns for MENA: - Vulnerabilities may be hidden by accounting/regulatory treatments. - Concentration in certain asset classes (e.g., government bonds) and reliance on foreign funding are key vulnerabilities. - Interest-rate-sensitive borrowers may face debt servicing challenges, potentially affecting profitability, provisioning, credit provision, and growth. ### Research questions and stress test design - The paper performs cross-country corporate and banking sector stress tests for MENA using publicly available microdata, common methodologies, and scenarios to facilitate cross-country comparisons. - Three core questions: 1. Impact of a higher-for-longer interest rate environment on corporate sectors (debt servicing, interest coverage ratios, zombie credit). - Corporate stress scenario: a 100-basis point (bps) increase in interest rates for two subsequent years and a negative profitability shock affecting more than 1,000 firms in MENA. - Outcome inference: increase in NPLs inferred from deterioration in distribution of interest coverage ratios. 2. Resilience of banks under higher-for-longer rates. - Banking stress test quantifies losses under liquidity and credit quality shocks and their interaction. - Estimates impact of rising interest rates on banks’ net interest margins following Schnabl and others (2021). - Recognizes offsetting potential from increasing net interest margins for banks whose assets reprice faster than liabilities. 3. Macro effects of banking sector stress. - Uses local projections (Jordà 2005) to estimate the link between bank capital and credit provision. - Uses a small-scale macro-econometric model to estimate elasticity of output to changes in credit provision and simulate counterfactual output paths under banking stress. ### Key quantitative results (summary) - Corporate sector: - Zombie-held debt among publicly listed firms rises from 12 percent to 30 percent in a higher-for-longer scenario. - Share of zombie firms increases from about 12 percent in 2022 to almost 30 percent by end-2024 in the adverse scenario. - Sectors relatively more affected: transportation, capital goods, food and beverage. - Banking sector: - In the joint stress scenario (higher rates, corporate stress, liquidity pressures): - Tier-1 capital ratios decline by 2.3 percentage points in the GCC. - Tier-1 capital ratios decline by 4-percentage points in non-GCC MENA. - While few banks become undercapitalized, lower bank capital could limit lending and add downside risks to output. - Macro outcomes: - Average output declines in the adverse banking sector stress scenario: - GCC: 0.4 percent contraction. - Non-GCC MENA: 0.7 percent contraction. ### Data, sample, and representativeness - Bank-level data: - Source: Fitch Connect (consolidated financial statements). - Final sample: 108 banks from 10 countries (Bahrain, Egypt, Jordan, Kuwait, Morocco, Oman, Pakistan, Qatar, Saudi Arabia, United Arab Emirates). - Coverage: at least 78 percent of banking assets in each country. - Table 1 (Bank Summary Statistics) — regional averages: - Total assets (US$, million): GCC 49,545.66; MENA (excl. GCC) and Pakistan 17,143.61 - Tier-1 capital ratio (%): GCC 17.5; MENA (excl. GCC) and Pakistan 14.78 - Return on assets (RoA) (%): GCC 1.82; MENA (excl. GCC) and Pakistan 1.23 - Loan share: GCC 0.61; MENA (excl. GCC) and Pakistan 0.48 - Government bond share: GCC 0.12; MENA (excl. GCC) and Pakistan 0.29 - Cash share: GCC 0.08; MENA (excl. GCC) and Pakistan 0.07 - Deposit share: GCC 0.74; MENA (excl. GCC) and Pakistan 0.75 - NPL ratio (in %): GCC 5.24; MENA (excl. GCC) and Pakistan 5.8 - Provisioning ratio (in %): GCC 1.33; MENA (excl. GCC) and Pakistan 1.3 - N (number of banks): GCC 61; MENA (excl. GCC) and Pakistan 47 - Stylized comparisons: - GCC banks are about three times larger than banks in remaining MENA and Pakistan (average assets $49.5 billion versus $17.1 billion). - GCC banks have higher Tier-1 capital ratios (17.5 percent versus 14.8 percent) and higher profitability (RoA 1.82 percent versus 1.23 percent). - Asset composition differences: GCC loan share 61 percent vs. non-GCC loan share 48 percent; non-GCC higher sovereign bond holdings (29 percent) indicating elevated bank-sovereign nexus. - Average bank provisioning ratio ~130 percent across regions. - Firm-level data: - Source: S&P’s Capital IQ. - Coverage: more than 1,000 publicly listed firms from 15 MENA countries and Pakistan. - 2022: ~400 firms from GCC; ~700 firms from rest of MENA and Pakistan. - Pakistan, Saudi Arabia, Egypt: each have over 100 firms. - Jordan, Kuwait, Oman, Morocco, Tunisia: 50–100 firms each. - Median firm (2022): - Total assets: about $130 million overall; median GCC firm ~7 times larger than median non-GCC firm. - EBITDA: median GCC firm about $23 million; median non-GCC firm $5 million. - Corporate sector historical trends: - Median return on assets fell from about 7.5 percent in 2006 to 5.5 in 2019. - Median leverage (total liabilities to assets) around 20 percent; one-quarter of firms have leverage exceeding 40 percent. - Median maturity of long-term corporate debt stable between three and four years; one-quarter of firms have debt duration < two years. ### Methodological contributions and literature positioning - Corporate zombie literature: - Confirms substantial deterioration in corporate financial health after major shocks (Global Financial Crisis, oil price slump). - Historical pattern: share of zombie firms more than doubles about 2–3 years after shocks (from 10–15 percent to 30–40 percent). - Banking sector literature: - Extends recent work on higher rates' impact on banks (primarily focused on advanced economies) by providing a banking sector stress test for 10 large MENA banking systems that jointly considers liquidity and credit risk and their interaction. - Explicitly incorporates corporate sector stress test results into banking stress tests. - Macrofinancial linkages: - Empirically quantifies output losses from reduced credit provision when bank capital declines in MENA, adding to regional literature on bank profitability and growth. ### Structure of the paper (sections) - Section 2: Data and stylized facts for corporate and banking sectors in MENA. - Section 3: Corporate sector stress test. - Section 4: Banking sector stress test and integration of corporate sector results. - Section 5: Banking sector stress test results and macro-financial linkages. - Section 6: Macroprudential policy status in the region. - Section 7: Conclusion. - Annexes: - I. The Design of Corporate Stress Tests - II. Estimation of RoA Betas - III. Details of Bank Stress Test - IV. Model of Macro-financial Linkages *Source: wpiea2024080-print-pdf; https://www.imf.org/-/media/files/publications/wp/2024/english/wpiea2024080-print-pdf.pdf* ### 1. Corporate Sector: Profitability (Return on assets, ### 1. Corporate Sector: Profitability (Return on assets, percent) ### Interest rates and corporate buffers - Historically, the median effective interest rate hovered near 6 percent. - Corporate rates peaked in 2009 when the median interest rate reached 7.5 percent; interest rates for the top 25 percent of riskiest firms were almost twice that amount. - Corporate rates bottomed out near the 5 percent threshold in 2017. - The effective interest rate is assumed to increase homogeneously for each firm by 100 bps per year over 2023–24 in the stress-test adverse scenario. - Effective interest rates are sequentially increased from about 6 percent at the end of 2022 by 100 basis points per year, reaching an average of more than 8 percent by the end of 2024, or 2 percentage points higher than pre-pandemic levels. - The median firm typically holds about 4 percent of total assets in cash and cash equivalents. - Some firms keep liquid buffers closer to 10 percent of total assets. - The top quarter most liquid firms significantly increased their cash buffers (by about 1 percentage point) before the pandemic. ### COVID-19 shock, recovery, and interest coverage - The COVID-19 shock adversely impacted the corporate sector—particularly contact-intensive industries—but strong policy support in ME&CA helped mitigate the overall impact. - The share of firms with an interest coverage ratio (ICR) below 1 has declined to about 10 percent of the total in the aftermath of the pandemic. - The portion of debt held by firms with an ICR between 1 and 5 has almost halved from pre-pandemic levels, representing about 25 percent of the total. - Many firms received temporary support during the pandemic, including loan relief and tax breaks; together with lower interest rates in 2020–22, this support contributed to a rebound in firm profitability and a lengthening of debt maturities. ### The rise of “zombie” firms - Definition used (following Acharya and others (2022)): zombie firms are those with an average ICR below 2.5 over two years that simultaneously receive “subsidized” lending (their effective interest rate is below that of top-rated firms). - The share of zombie firms in the region stood at about 12 percent in 2022. - Zombie firms’ median leverage (total liabilities to total assets) was 40 percent at the end of 2022 versus 20 percent for other firms. - Zombie firms’ operating losses continued to deteriorate while non-zombie firms’ median profitability recovered after the pandemic. - At the end of 2022, zombie firms held about 12 percent of corporate debt. - About 10 percent of the zombie firms exited during the pandemic shock. - Before and after the pandemic, zombie firms are considerably less profitable, more leveraged, and have weaker cash buffers than healthy firms, creating potential resource misallocation. ### Stylized facts about MENA banking sectors relevant to corporate stress - Capital ratios across the region are well above regulatory minimums, even after removal of most pandemic-related forbearance measures. - Bank profitability has recovered from pandemic lows, but performance of state-owned banks remains well below pre-pandemic levels in non-GCC MENA economies and, to a lesser extent, in the GCC. - For the median bank in the GCC, 85 percent of non-equity funding is through deposits or long-term debt; in non-GCC MENA countries, this figure is 90 percent. - Nonresident deposits and other foreign liabilities increase vulnerability to sudden shifts in investor sentiment, particularly in GCC countries with larger dependence on external funding. - In the median country, depositors are insured up to three times GDP per capita; at the fifth percentile of countries, less than 50 percent of GDP per capita is covered. - Rapid interest rate hikes since the beginning of 2022 expose banks to interest rate risk as fixed-income securities decline in value. - Unrealized losses on available-for-sale securities exceed net income in some cases; however, the share of unrealized gains/losses as a fraction of capital appears manageable on average. - Elevated bank holdings of domestic sovereign debt imply a rising sovereign-bank nexus, with banking sector claims on the public sector ranging between 10 and 30 percent of banking sector assets in most MENA countries and exceeding 50 percent in some countries (Egypt, Pakistan). - Elevated bank-sovereign nexus can inflate reported risk-based capital ratios due to zero-risk weighting on domestic government bond holdings and may crowd out private-sector credit. ### Corporate sector stress test: methodology - Adverse scenario calibrated to a higher-for-longer interest rate environment combined with a sector-specific profitability shock to simulate a global slowdown. - Effective interest rate shock: increase of 100 bps per year per firm over 2023–24 (see above for level path from about 6 percent to more than 8 percent by end-2024). - Sector-specific EBIT shock: calibrated based on observed changes in profitability during and after the Global Financial Crisis (2009 and 2010) for each sector; most sectors experienced double-digit negative returns in that period. - For each firm i in sector s, change in profits is given by: - EBIT_{ist+1} = EBIT_{ist} * (1 + π_{ist+1}) - where π_{ist+1} is the change in firm profitability drawn from a sector-specific distribution. ### Corporate sector stress test: results (partial) - The corporate stress test results show that firm profitability could decline, on average, to about 3 percent in 2024, below the pre-pandemic level of 5 percent. - The report states the median ICR is estimated to decline from 3.5 to (text truncated; end of provided content). *Sources: S&P Capital IQ; and IMF staff calculations.* ### 1.5 at the end of 2024, and the share of debt at risk of default more than doubles, rising from about 12 percent ### wpiea2024080-print-pdf - 1.5 at the end of 2024, and the share of debt at risk of default more than doubles, rising from about 12 percent ### Corporate-sector zombification and debt at risk - The share of debt at risk of default rises from about 12 percent of total debt in 2022 to almost 30 percent by 2024 (Figure 8, panel 1). - Under the main simulation, the median share of zombie firm debt increases sharply in several sectors: - Transportation, capital goods, and food and beverage sectors are most vulnerable. - The median share of zombie firm debt in the food and beverages sector increases particularly sharply by 2024 (Figure 8, panel 2). - Alternative stress scenario (robustness check, calibrated following the 2014–15 oil price shock): - Under this scenario, the share of zombie-held debt increases to 25 percent of total debt by 2024. - Sectors more sensitive to lower oil prices (for example, energy) are associated with more negative returns while others benefit (for example, transportation). - Transportation and food and beverages remain the sectors most sensitive to the interest rate shock in the alternative scenario. - Indirect channels may further increase zombification beyond the direct impact of the interest rate increase (see Albuquerque and Mao (2023)). ### Banking sector stress test: scope and purpose - The stress test examines liquidity shocks and borrower distress in MENA banking sectors, with results differentiated by the level of the interest rate. - The stress test proceeds in three steps: 1. Estimate sensitivity of bank profitability to interest rate changes (RoA betas). 2. Simulate a liquidity shock through sudden withdrawal of deposits and non-deposit funding. 3. Map corporate sector stress scenarios into increasing NPLs and simulate impact on bank balance sheets. - The main exercise includes joint liquidity and corporate sector stress amid higher interest rates; liquidity and corporate stress are also considered in isolation. ### Methodology: estimating interest-rate sensitivity and liquidity stress - Estimation of bank-level income, expense, and RoA betas follows Drechsler and others (2021) adapted to a sample of 10 MENA countries. - RoA betas measure sensitivity of return on assets (RoA = quarterly net income divided by total assets) to policy rate increases. - RoA beta defined as βiRoA = ∑훽i,휏RoA (휏=0 to 3), using average RoA over the last four quarters to account for cyclicality. - Income betas: first difference of total interest income divided by quarterly assets. - Expense betas: analogous ratio with total interest expenses in numerator. - Banks included in estimation if they have at least 40 quarterly observations of the outcome variable. - Estimation results and correlations with bank characteristics (Annex 2): - Many banks tend to benefit on average from higher rates, but about one-third of banks have a negative RoA beta. - Mean RoA beta is 0.097, with a standard deviation of 0.37. - Banks with lower NPL ratios, more held-to-maturity securities, a lower share of wholesale funding, or higher capital ratios have, on average, higher RoA betas. - For banks with insufficient data (<40 quarters), betas are assumed equal to a weighted average of betas estimated for other banks within the same country. - Liquidity stress-test design: - Deposit withdrawal size: a 20 percent withdrawal of deposits; foreign deposits and wholesale funding assumed to have a 30 percent outflow rate. - Withdrawal per bank formula: Withdrawal_b = .2 LC Deposits_b + .3 (FX Deposits_b + Wholesale Funding_b) - Asset liquidation pecking order to meet withdrawals: 1. Pre-existing cash buffers. 2. Government bonds already marked-to-market. 3. Government bonds not yet market-to-market. 4. Other securities. 5. Illiquid assets like loans. - Assumptions on losses: - Banks incur an additional 30 percent discount if they need to sell loans or another illiquid asset to meet deposit withdrawals. - Fire-sale discounts for securities are simulated via 1,000 draws calibrated to the increase in sovereign spreads in 2020; reporting focuses on the median draw. - Credit stress mapping: - Deterioration in corporate-sector ICRs mapped into default probabilities using Damodaran (2023). - Implied increase in NPLs: 2.1 percent of net loans for 2023 and 2.7 percent for 2024. - Banks provision 100 percent against the increase in NPLs (with a Kuwait-specific proviso: for Kuwait, increase in provisions is the minimum of the increase in NPLs and the amount needed to maintain provisioning ratio of at least 150 percent). ### Scenarios considered - Four scenarios are analyzed: 1. Liquidity stress: 20 percent deposit withdrawal with 30 percent withdrawal of foreign deposits and wholesale funding. 2. Liquidity stress amid higher interest rates: scenario 1 amid 200 basis points higher interest rates that lead to capital losses on fixed income assets. 3. Credit stress: rise in corporate sector NPLs consistent with the corporate sector stress test, with increasing NPLs 100 percent provisioned for. 4. Combined liquidity and credit stress amid higher interest rates: combination of scenarios 2 and 3. ### Results: heterogeneity and regional outcomes - Heterogeneity across banks is substantial; bank-level microdata is critical to reveal individual vulnerabilities. - Many banks benefit from higher rates, but about one-third of banks are adversely affected due to negative RoA betas. - Regional summary: - Banking systems in most countries could withstand isolated liquidity or corporate-sector stress amid higher interest rates. - A combined scenario of higher rates, corporate-sector stress, and liquidity stress could materially stress banking systems, particularly non-GCC countries. - Loss drivers: lower profitability, lower capital ratios, and a large bank-sovereign nexus in non-GCC MENA countries (Egypt, Jordan, Morocco, Pakistan). - Quantitative results for the GCC: - Losses in the GCC range from 0.3 percent of capital in the isolated liquidity stress scenario to 13.4 percent of Tier-1 capital in the combined stress scenario. - In the combined shock scenario, the weighted average Tier-1 capital ratio in the GCC would drop from 17.1 percent to (text truncated in source). *IMF WORKING PAPERS Financial Stability in a Higher-for-Longer Interest Rate Environment — INTERNATIONAL MONETARY FUND* ### 14.8 percent, significantly above Basel-III regulatory minima of 8.5 percent. ### wpiea2024080-print-pdf - 14.8 percent, significantly above Basel-III regulatory minima of 8.5 percent. ### Bank stress test results - Current reported capital ratios: - "14.8 percent, significantly above Basel-III regulatory minima of 8.5 percent." - For non-GCC MENA countries, capital ratios would decline from their current level of "14 percent to 10 percent." - Stress scenarios and key outcomes: - Losses (as a fraction of Tier-1 capital) are reported across subregions and scenarios (liquidity stress; liquidity + 200bps; corporate stress; liquidity + corporate stress). - Losses in the combined scenario exceed the sum of losses under individual scenarios because "bank buffers can only be deployed once." - Example magnitude of losses in MENA (excluding the GCC) and Pakistan: - Liquidity shock under higher interest rates: "12.8 percent of Tier-1 capital." - Isolated corporate sector stress: "6 percent" of Tier-1 capital. - Combined liquidity and corporate sector stress amid higher interest rates: "29 percent" of Tier-1 capital. - Undercapitalization shares (weighted by assets): - GCC banks: "1.2 percent" would become undercapitalized under the most severe scenario using Basel-III minima; rises to "1.9 percent" when using country-specific minima. - MENA (excluding the GCC) and Pakistan: "20.2 percent" would become undercapitalized using Basel-III minima; "23.1 percent" when using country-specific minima. - Recapitalization needs if undercapitalized banks are recapitalized to meet minimum Tier-1 requirements: - GCC: total recapitalization needs amount to "$0.9 billion" in the most severe scenario using country-specific Tier-1 minima. - Non-GCC MENA: bank recapitalization would require "$4.8 billion." ### Capital requirement context and assumptions - Basel-III Tier-1 minimum used in analysis: "8.5 percent" (CET-1 regulatory minima "4.5 percent" + additional tier-1 capital "1.5 percent" + typical CCB "2.5 percent"). - Note on country variation: some countries temporarily lowered the CCB during the pandemic; example: the CCB in Oman "still stands at 1.25 percent, below its 2.5 percent normal level." - Country-specific Tier-1 capital requirements are higher than Basel-III minima in several sample countries; domestic systemically important banks’ surcharges and other buffers are treated as buffers, not as capital requirements. ### Heterogeneity analysis — drivers of vulnerability - Role of state-owned banks (SOBs): - State-owned banks defined as banks with at least "50 percent" government ownership. - Weighted average losses: - GCC privately owned banks: "10.3 percent" of capital. - GCC state-owned banks: "18.1 percent" of capital. - MENA (excluding GCC) and Pakistan privately owned banks: "22.9 percent" of capital. - MENA (excluding GCC) and Pakistan state-owned banks: "26.5 percent" of capital. - State-owned banks are more vulnerable due to slower recovery and greater exposure to interest-rate-sensitive fixed-income securities, particularly held-to-maturity government bonds. - Role of ex-ante vulnerabilities (median splits): - Banks with below-median capitalization (Tier-1 Capital Ratio) average losses: "18 percent" of capital; above-median capitalization: "6 percent." - Banks in countries with low duration risk: average losses "9 percent"; with above-median duration: "32 percent." - Larger banks (above-median market share) have on average losses of "18 percent" of capital. - Losses concentrated in banks with below-median holdings of liquid assets, lower ex-ante provisioning, and weak profitability. ### Impact on credit and output - Bank-level relationship between capital and lending (local projections): - A "1-percentage-point" lower Tier-1 capital ratio is associated with "1.5 percentage points" lower real lending after 10–12 quarters. - The effect is statistically significant from five quarters onward. - Caution: capital structure and lending decisions may be simultaneous; the effect should not be interpreted as causal. - Macro impact (small-scale macroeconometric model calibrated with micro estimates): - Output contraction under the combined banking sector stress scenario: - MENA: "0.7 percent" contraction. - GCC: "0.4 percent" contraction. - Across countries: - MENA (excluding GCC) and Pakistan potential output losses range "from 0.2 to 1.2 percent." - GCC potential output contraction range "between 0.05 and 1.2 percent." - Note: estimates likely underestimate actual output losses since several general equilibrium channels are not directly modeled; banking sector stress that requires recapitalization would have fiscal impacts and could trigger fire sale externalities. ### Macroprudential frameworks in MENA - Use of macroprudential policies: - Overall, the use of macroprudential policies in the MENAP region has been "slow and subdued compared to other emerging markets." - Credit and capital-related tools dominate in the GCC and MENAP, followed by liquidity-based tools. - Specific observations: - Almost all countries have broad-based macroprudential tools regarding capital buffers (e.g., CCyB), but many countries have "never activated the CCyB." - Most countries implemented borrower-based tools for the household sector (e.g., caps on debt-service-to-income ratios). - Tools to guard against risks in the nonfinancial corporate sector have been "less utilized." - Some GCC countries have not employed measures to reduce banks’ foreign currency liquidity risks. - Some MENA countries have taken fewer actions to reduce risks from systemically important domestic financial institutions. - Cross-country association: - Countries that more frequently use macroprudential tools, other things being equal, tend to experience stronger and less volatile GDP growth, with effects influenced by openness and financial development. *IMF WORKING PAPERS Financial Stability in a Higher-for-Longer Interest Rate Environment — INTERNATIONAL MONETARY FUND* ### 7. Conclusion ### 7. Conclusion ### Summary of the stress-testing exercise - First corporate sector and banking sector stress tests for the MENA region using granular firm-level and bank-level data and common shock scenarios across ten countries. - Exercises rely on publicly available data, designed to be easy to update, customize, replicate, and extend to other countries. ### Corporate sector findings - Corporate sector return on assets (RoA) declines from 5 percent to 3 percent in the stress scenario. - Median interest coverage ratio (ICR) falls from 3.5 to 1.5. - The share of zombie-held debt could increase from 12 to 30 percent of total corporate sector debt under the baseline stress scenario. - Robustness: - Standard deviations for drawn firm profitability shocks were capped at 50 percent. - Under a downside oil price shock scenario, the share of zombie-held debt could increase to 25 percent of total debt by 2024, with an interquartile range of 18-58. - Alternative zombie definitions produce consistently similar shares of corporate debt held by zombies except for one definition (definition 3) that shows a subdued hump during/after the oil price shock. ### Banking sector findings - Joint credit and liquidity stress (using corporate stress test results and liquidity shocks amid higher rates) suggests overall MENA banking sectors appear resilient in the simulated joint stress scenario. - Vulnerabilities are more pronounced in non-GCC countries: - Bank capital would decline by 2.3 percentage points in the GCC and by 4 percentage points in non-GCC MENA countries. - Cross-sectional bank vulnerabilities: - Banks with weak fundamentals (for example, low profitability, low holdings of liquid assets, high leverage) and state-owned banks are more vulnerable. - Endogenous net income effects: - About two-thirds of banks benefit from higher rates (counterfactual RoA exceeds actual RoA), while about one-third do not benefit; banks that do not benefit tend to have lower ex-ante profitability. - Net income is the first buffer to absorb provisioning and losses; more profitable banks can better absorb increased provisioning needs. ### Liquidity stress-test methodology and assumptions (high-level) - Methodology builds on Jiang and others (2023) and Copestake, Kirti, and Liu (2023), with adaptations for MENA. - Key steps: - Mark banks’ sovereign bond holdings and other securities to market. - Calibrate deposit withdrawal scenarios to determine asset sales required at current market values. - Scale realized losses by regulatory capital if they exceed ex-ante net income. - Market-value and sale assumptions: - Use change in US yield curve since January 2022 at portfolio duration plus a country-specific spread as a proxy for valuation changes where local yield curves are unreliable. - Impose a fire sale discount based on the increase in sovereign spreads between February 2020 and the July–September 2020 average; draw 1,000 times per bank and country and report medians. - Banks sell available-for-sale (AFS) securities first, then held-to-maturity (HTM) securities if needed. - Withdrawal scenarios and parameters: - 20 percent withdrawal of resident deposits. - 30 percent withdrawal of nonresident deposits. - 30 percent withdrawal of wholesale funding and other short-term funding. - Banks initially meet withdrawals using available cash and equivalent liquid assets with no market value losses. - Asset-sale loss scaling: - Excess losses formula includes realized discounts on sovereign bond sales, other security sales, other asset sales (with a 1.3 multiplier applied to the applicable discount), net of net income buffers. ### Policy implications and recommendations - Policymakers should be attentive to financial stability risks for corporates and banks in a higher-for-longer interest rate environment. - Consider broad-based macroprudential tools: - Use of the countercyclical capital buffer remains subdued in MENA compared to other emerging markets. - Address the bank-sovereign nexus where it is a substantial vulnerability: - Preserving bank capital where banking systems face elevated interest rate risk preserves loss absorption capacity. - Conduct bank stress tests with explicit focus on the sovereign-bank nexus to understand severity and nature of risks. - Over the medium term, develop alternative funding sources for domestic sovereign bonds to diversify the investor base and reduce the nexus. - Avoid loan evergreening: - If borrower defaults rise, banks may be incentivized to delay recognition of loan losses by evergreening credit at advantageous rates, leading to long-lasting NPL buildups and the emergence of zombie firms that weigh on aggregate productivity growth. - Resolution regimes should facilitate swift resolution of NPLs to prevent banks from building up impaired legacy assets on their balance sheets. ### Annex highlights - Annex I (Corporate Stress Tests): - Firm profitability shocks for 2023–2024 drawn from sector-specific distributions calibrated to the GFC and, as robustness, the 2015 oil shock. - Standard deviations capped at 50 percent. - Annex II (Estimation of RoA Betas): - Distribution of estimated RoA betas across sample banks shows bulk near zero but sizable variation; about 1/3 of banks have negative RoA betas. - Bank characteristics correlated with higher RoA betas: lower NPL ratios, higher share of HTM securities, lower share of wholesale funding, higher Tier-1 capital ratios. - Annex III (Details of Bank Stress Test): - Liquidity stress-test mechanics, valuation proxies, fire sale discount calibration, sale sequencing (sovereign bonds first), and excess loss accounting are specified and applied in the simulations. *IMF Working Paper — Financial Stability in a Higher-for-Longer Interest Rate Environment (Section 7 and Annexes I–III).* ### Annex IV. Model of Macrofinancial Linkages ### Annex IV. Model of Macrofinancial Linkages ### Model - Two-equation macro model linking credit and business cycle fluctuations. - Credit equation (credit largely related to the business cycle): - c_t = ρ_1 c_{t−1} + ρ_2 y_{t−1} + ε^c_t - where c_t is the real credit gap, y_t is the real output gap, and ε^c_t is a stationary autocorrelated shock to real credit (ε^c_t = φ ε^c_{t−1} + ε^c_t). - Interpretation: banks set desired credit based on past economic activity; slow adjustment of credit reflected in ρ_1 c_{t−1}. - Note on inertia: banks cannot immediately adjust credit levels (footnote 31: For example, in Khwaja and Mian (2008), Pakistani banks take about one quarter before the impact of a bank liquidity shock is passed on into bank lending). - Output equation (output responds to autonomous credit shocks): - y_t = α_1 y_{t−1} + α_2 ε^c_t + ε^y_t - where ε^y_t is a stationary autocorrelated demand shock (ε^y_t = τ ε^y_{t−1} + ε^y_t) capturing demand shocks uncorrelated with domestic credit shocks including real interest and exchange rates, commodity prices, and global demand shocks. - All shocks (denoted ε^x_t for variable x_t) are assumed to be independently and identically distributed white noise processes. - Definitions of gaps: - y_t = Y_t − Ȳ_t, and c_t = C_t − C̄_t - where Y_t and C_t are (log) levels of real GDP and real credit, and Ȳ_t and C̄_t are their associated long-run trends. - Long-term trends follow I(2) processes: - ΔȲ_t = ΔȲ_{t−1} + ε^{Ȳ}_t, and ΔC̄_t = ΔC̄_{t−1} + ε^{C̄}_t. ### Data and Estimation - Estimation approach: - Parameters and trends estimated country by country using quarterly data and the Kalman filter and Bayesian methods (estimation details are available on request). - Data and sample: - Quarterly data from Haver Analytics ranging from 2001Q4 to 2023Q1. - Observable variables: real GDP and real credit. - Real credit defined as banking sector claims on the non-financial private sector; nominal credit is deflated with the CPI. - Trend calibration: - Long-term trends calibrated using the HP filter with the standard smoothing parameter for quarterly data (1600). ### Estimated Output Responses Following Past Credit Downturns - Two exercises reported in the main text: - Historical credit downturns: - Using historical data on credit downturns (ε̂^c_t < 0), plot the distribution of output losses defined as the two-year average output loss following historic credit downturns. Labeled “historical credit downturns” in Figure 11, panel 2. - Stress scenario mapping: - Declines in bank capital under the stress test scenario are mapped into a decline in credit using the estimates from equation (4). - The estimated macro-econometric model is used to compute implied country-by-country output losses, yielding the “stress scenario” results in Figure 11, panel 2. *Source: Annex IV. Model of Macrofinancial Linkages, Financial Stability in a Higher-for-Longer Interest Rate Environment, Working Paper No. WP/2024/080* --- _Source: https://www.imf.org/-/media/files/publications/wp/2024/english/wpiea2024080-print-pdf.pdf_