## 1. Profitability ## Source details **Canonical URL:** [1. Profitability](https://www.imf.org/-/media/files/publications/wp/2018/wp1899.pdf) ## Other formats - [Markdown version](/-/media/files/publications/wp/2018/wp1899.pdf.md) - [Structured JSON version](/-/media/files/publications/wp/2018/wp1899.pdf.json) --- ### Abstract — key findings - Banks that were more successful at protecting their profits: - had a less pronounced deterioration in loan quality; - had a larger improvement in cost efficiency; - downsized their assets more aggressively during the crisis; - reduced reliance on wholesale funding more markedly post-crisis. - Net interest margins remained broadly stable over the financial cycle, including post-crisis. - There is no clear evidence that aspects of bank business model, such as higher reliance on fees and commission income, were associated with better profitability post-crisis. - JEL Classification Numbers: E30, G01, G21. - Keywords: Financial Cycle, Bank Profitability, Financial Crisis. - Authors: Enrica Detragiache, Thierry Tressel, Rima Turk-Ariss. - WP/18/99. ### Context and systemic drivers - EU banking system experience over the sample: - rapid expansion, deep crisis, tepid recovery over ~15 years. - Post-crisis headwinds listed: - crisis legacies; - adaptation to stricter financial regulation; - increased competition from nonbank intermediaries; - disruption from new technologies; - weak economic growth and lackluster business investment reducing credit demand; - negative policy interest rates and a flat yield curve pressuring interest margins. - Result: bank profitability plunged during the crisis and remained low post-crisis. ### Empirical emphasis and data scope - Data: bank-level balance sheet and income statement data for 114 EU banks over 2000–2016 (Fitch database). - Focus: determinants of profitability performance across banks and over time, including loan quality, cost efficiency, asset growth/downsizing, funding structure, net interest margins, and revenue mix. - Subperiods: 2000–04 (pre-boom), 2005–07 (boom), 2008–12 (crisis), 2013–16 (post-crisis). ### I. Main findings and summary - Containing rise in NPLs associated with better profit preservation. - Cost discipline (reductions in operating costs) translated into better profit performance. - Banks that reduced assets more aggressively during the crisis experienced smaller post-crisis profit declines. - Larger declines in the share of wholesale funding correlated with better post-crisis profit performance. - Net interest margins (NIM) broadly stable over the cycle. - No evidence that greater reliance on fees and commission income was systematically associated with better profit performance post-crisis. ### II. Sample construction and subperiods - Sample composition: - 114 EU banks reporting consolidated accounts over 2000–2016. - Country groups: Eastern EU (Hungary, Poland, Romania, Slovenia), Nordics (Denmark, Finland, Sweden), Other EU (Austria, Belgium, France, Germany, Luxembourg, Netherlands, UK), EC-6 (Cyprus, Greece, Ireland, Italy, Portugal, Spain). - Includes G-SIBs for international comparisons; U.S. assets adjusted for accounting differences. - Exclusions: unconsolidated accounts, subsidiaries, banks that closed during the sample. ### III. Stylized facts: profitability, asset quality, dividends, assets, leverage, funding - Profitability patterns - ROA and ROE: rose in the boom, declined dramatically in the crisis, remained depressed post-crisis. - Nordics post-crisis ROE: 9.1 percent (down from 13.8 percent in the pre-boom period). - EC-6 experienced the sharpest decline in ROE (partly due to provisioning for high NPLs). - Net interest margin (selected weighted averages for EU G-SIBs by period) - Net Interest Margin: Pre-boom 1.4, Boom 1.0, Crisis 1.1, Post-crisis 1.2. - ROA: Pre-boom 0.6, Boom 0.6, Crisis 0.2, Post-crisis 0.1. - ROE: Pre-boom 13.4, Boom 16.3, Crisis 2.7, Post-crisis 2.4. - NPLs / Gross loans (selected weighted averages) - EC-6: Pre-boom 2.4, Boom 34.1, Crisis 98.9, Post-crisis 17.4. - Eastern EU: Pre-boom 20.6, Boom 37.0, Crisis 48.7, Post-crisis 29.6. - Dividends - Eastern EU and EC-6 cut dividend payouts post-crisis. - Other EU and Nordics increased payout ratios to close to 70 percent post-crisis (from around 50 percent pre-crisis). - For largest banks, post-crisis dividend payout ratios averaged close to 75 percent of earnings. - US G-SIB dividend payout ratio dropped to 18 percent post-crisis. - Asset growth (annual rates, weighted averages by region) - EC-6: Pre-boom 13.8, Boom 21.4, Crisis 1.6, Post-crisis -6.2. - Eastern EU: Pre-boom 29.3, Boom 28.7, Crisis 5.4, Post-crisis -0.6. - Other EU: Pre-boom 16.1, Boom 23.0, Crisis -0.6, Post-crisis -7.3. - Nordics: Pre-boom 21.7, Boom 20.8, Crisis 4.1, Post-crisis -4.7. - Overall sample: mean asset growth ~7.7 percent, median 1.7 percent. - Leverage / capitalization (Tangible equity / Tangible assets, TE ratio; weighted averages) - EC-6: Pre-boom 4.5, Boom 4.4, Crisis 4.2, Post-crisis 5.2. - Eastern EU: Pre-boom 9.5, Boom 8.0, Crisis 9.1, Post-crisis 10.5. - Other EU: Pre-boom 3.2, Boom 2.8, Crisis 2.9, Post-crisis 4.4. - Nordics: Pre-boom 4.2, Boom 3.8, Crisis 3.6, Post-crisis 4.6. - Large banks had lower TE than small banks in pre-boom (3.4 percent vs 6.0 percent); TE fell to 2.8 percent for largest banks in boom. - Funding (wholesale funding / total funding, weighted averages) - EC-6: Pre-boom 30.3, Boom 38.7, Crisis 32.9, Post-crisis 26.3. - Eastern EU: Pre-boom 3.1, Boom 12.4, Crisis 12.0, Post-crisis 6.4. - Other EU: Pre-boom 23.1, Boom 39.2, Crisis 33.7, Post-crisis 27.4. - Nordics: Pre-boom 39.2, Boom 44.1, Crisis 48.1, Post-crisis 49.0. - Average wholesale funding / total funding: Pre-boom 25.9, Boom 37.8, Crisis 34.3, Post-crisis 29.3. ### IV. Decomposing the decline in ROA (components and regional patterns) - Accounting identity used: change in ROA = ∆Non-interest income – ∆Non-interest expense – ∆Provisions + ∆NIM (full formula in source). - Regional drivers of boom → post-crisis decline: - Eastern EU: large decline driven by sharp fall in NIM and non-interest income, and increased loan-loss provisions; partially offset by contraction in operating expenses. - EC-6: decline mainly driven by provisioning for NPLs; operational cost savings did not offset provisioning. - Other EU: decline mainly reflected fall in revenue from non-traditional banking activities (other non-interest income). - Nordics: reduction in income from non-traditional activities and some increased provisioning contributed to lower ROA, but operational cost savings limited deterioration. - NIM patterns: - Only Eastern EU showed sizable decline in NIM among regions; Other EU experienced improvement. - By size: small banks saw worsening intermediation spreads; largest banks registered slight increase in NIM. - G-SIBs: - Intermediation spreads rose for EU G-SIBs but declined for non-EU peers. - All G-SIBs except Swiss and Chinese saw declines in earnings from non-traditional sources. - EU G-SIBs faced greater increases in operating and provisioning costs relative to other G-SIBs; U.S. G-SIBs compensated with sharp cost cutting. ### V. Empirical approach and estimation - Dependent variable: change in ROA between consecutive financial-cycle phases (pre-boom → boom, boom → crisis, crisis → post-crisis). - Main regression (panel): ΔROA_ijt = X_{ijt-1} β + ΔZ_ijt γ + ΔM_jt δ + α_t + ε_ijt. - X: bank initial characteristics averaged over previous period. - ΔZ: contemporaneous changes in bank variables (percentage points). - ΔM: changes in country-specific macro variables (decimals) in some specifications. - Period fixed effects and some specifications include regional fixed effects. - Estimation: OLS with robust standard errors clustered at bank level; panel has three period changes; winsorization at 5 percent used in robustness checks. ### VI. Bank-level explanatory variables - Initial conditions (averaged over previous period): - Bank size (log of assets). - Past expansion (asset growth rate). - Funding model (share of wholesale funding). - Loan portfolio quality (NPLs / net loans). - Share of interest income in total income. - Tangible equity / tangible assets (TCE/TA). - Initial profitability (for mean reversion). - Contemporaneous changes: - Change in wholesale funding share. - Change in operating costs (non-interest expenses / TA). - Change in reliance on fees and commission income. - Change in NPL ratio. ### VII. Macroeconomic controls - Country-specific macro proxies experimented with: - Real GDP growth, output gap, unemployment rate. - Policy rate proxies, term premium, aggregate credit growth. - Stock price growth, housing price growth. - Sovereign stress: spreads relative to bunds. - Difference between average system-wide deposit interest rates and policy rate. - Note: cross-border operations of EU banks may limit relevance of domestic macro variables. ### VIII. Key empirical facts, summary statistics, and regression results - Sample summary statistics (selected) - ROA: overall sample mean 0.24 percent; median 0.31 percent. - Asset growth: overall sample mean 7.68; median 1.67. - Wholesale funding / total funding: mean 30.90; median 27.89. - NPLs: average 6.4 percent of gross loans; median 3.9 percent. - Interest income share: on average 80 percent of total income (source text reports “on average 80 percent”). - Tangible common equity: on average about 5 percent of total tangible assets (source text). - ROA changes and correlations - ROA declined on average by 0.21 percentage points from one period to the next; median decline -0.06 percentage points. - Share of fees and commission income approximately stable on average; large heterogeneity across banks. - Average decline in wholesale funding share: -1.72 percentage points; median -0.87 percentage points. - Ratio of non-interest expenses to total assets declined (widespread cost cutting). - NPL ratio increased on average by 3 percentage points. - Pairwise correlations (selected) - Level correlations: ROA : Asset growth 0.6842; ROA : Tangible common equity / total equity 0.4609; ROA : Non performing loans / gross loans -0.4132. - Change correlations: ROA : Wholesale funding / total funding -0.5918; ROA : ROE -0.7173; Non interest expenses / total assets : Non performing loans / gross loans -0.6152. - Panel regression key coefficients and magnitudes (selected) - Mean reversion: A 100-basis points higher initial ROA associated with an additional decline in profits of some 80 basis points in the subsequent phase. - Asset growth: A ten percentage-point higher asset growth in the previous phase associated with a subsequent decline in profitability of an additional 12 basis points. Unweighted sample average decline in ROA is 21 basis points. - Capitalization: An increase in tangible capital ratio of 1 percentage point raises ROA by slightly over 4 basis points (significant at 10 percent). - Period effects: Crisis period associated with decline in ROA of some 50 basis points; post-crisis associated with further decline of some 50 basis points. - Regional effects: “Other EU” dummy estimated 18 basis points lower change in profits than Nordic banks (Nordics omitted). - Region × period interactions: - EC-6 * crisis coefficient: -0.584***. - Eastern EU * crisis coefficient: -0.781***. - Eastern EU * post crisis coefficient: -1.307**. - Macroeconomic controls (example coefficients): - Stock price growth coefficient 1.544*** (positive for bank profits). - Spread to Bunds coefficient -0.109*** (negative for profits). - Deposit rate minus short-term rate coefficient -0.237* (widening associated with declining profitability in one specification). - Contemporaneous bank changes: - Wholesale funding change coefficient ~ -0.0132*** (decline in wholesale funding associated with higher ΔROA): a 5-percentage point decline → increase in ROA of 8 basis points. - Operating cost efficiency: lowering non-interest expenses / TA by 100 basis points → increase in ROA by some 30–40 basis points. - Asset quality: a 5 percentage points increase in NPL ratio → additional decline in ROA by more than 30 basis points. - Fees and commission income: changes not systematically related to ΔROA in most specifications. ### Probit results — characteristics of highly profitable post-crisis banks - Definition: top quintile of ROE distribution during post-crisis period; ROE ≥ 9 percent. - Country-group effects (selected coefficients) - EC-6 dummy: -0.949* (negative and significant in one specification). - Other EU dummy: -1.414*** (negative and significant). - Eastern EU coefficient reported as 0.543 in one specification (not consistently significant). - Macro variables (post-crisis averages): - Short-term interest rate (ST rate) positive and significant: ST rate coefficient 0.841*** in one specification. - Deposit rate minus ST rate coefficients reported as -0.395 and -0.154 (not significant across all). - Bank-specific characteristics associated with higher probability of high post-crisis profitability: - Smaller size: log(assets) negative and significant (e.g., -0.193**). - Better loan quality: NPL/GL negative and highly significant (e.g., -0.102***). - Some evidence that lower reliance on interest income relates to higher profitability (interest income share coefficients include -0.677, -4.224***, -4.093*** across specifications). - Wholesale funding share not robustly associated with being highly profitable. - Asset growth marginally significant negative in some specifications (e.g., -0.00189*). ### Conclusions and policy-relevant implications - 2000–2007: rapid balance sheet expansion financed by rising wholesale funding share; leverage rose without commensurate high-quality capital; ROE mechanically boosted. - From 2008: wholesale funding withdrew, banks downsized and delevered; deteriorating loan quality in some countries; state liquidity and capital injections often required. - Post-crisis profitability remained low owing to lower leverage, difficulty cutting costs while downsizing, and persistent high NPLs for some banks. - Net interest margins remained broadly stable over the financial cycle in the sample; interest earnings and expenses fell in lockstep during downturns. - Policy and supervisory implications (empirically supported): - Strengthen asset-quality monitoring and loss-recognition practices to limit deterioration in loan quality and support profitability. - Policies and supervisory measures that promote cost efficiency can materially affect profitability outcomes; lowering non-interest expenses yields substantial ROA gains. - Incentives or frameworks that encourage active balance-sheet management (downsizing non-core assets during stress) can be associated with better profit protection, subject to macroprudential trade-offs. - Monitoring and, where appropriate, reducing reliance on wholesale funding can improve resilience and post-crisis profitability; the quality and orderliness of the funding shift matters. - Stability of NIMs suggests measures aimed solely at restoring margins may have limited short-term payoff; broader structural and demand-side recovery and a more favorable interest-rate environment matter for bank returns (post-crisis higher short-term rates increase probability of ROE ≥ 9 percent). *Source: IMF Working Paper WP/18/99 (wp1899) — extracted sections on profitability, sample, stylized facts, empirical methodology, regression and probit results, and policy implications.* ### 1. Profitability ### 1. Profitability ### Abstract — key findings - Banks that were more successful at protecting their profits: - had a less pronounced deterioration in loan quality; - had a larger improvement in cost efficiency; - downsized their assets more aggressively during the crisis; - reduced reliance on wholesale funding more markedly post-crisis. - Net interest margins remained broadly stable over the financial cycle, including post-crisis. - There is no clear evidence that aspects of bank business model, such as higher reliance on fees and commission income, were associated with better profitability post-crisis. - JEL Classification Numbers: E30, G01, G21. - Keywords: Financial Cycle, Bank Profitability, Financial Crisis. - Authors: Enrica Detragiache, Thierry Tressel, Rima Turk-Ariss. - WP/18/99. ### Context and systemic drivers - The EU banking system experienced a rapid expansion, a deep crisis, and a tepid recovery over the last 15 years or so. - The post-crisis recovery was hampered by: - crisis legacies; - the need to adapt to stricter financial regulation; - increased competition from nonbank financial intermediaries; - disruption from new technologies; - weak economic growth and lackluster business investment reducing credit demand; - negative policy interest rates and a flat yield curve putting pressure on interest margins. - As a result, bank profitability plunged during the crisis and has remained low in the post-crisis period. ### Empirical emphasis and data scope - Analysis based on bank-level balance sheet and income statement data for EU banks over the recent financial cycle. - Focus on determinants of profitability performance across banks and over time, including loan quality, cost efficiency, asset growth/downsizing, funding structure, net interest margins, and revenue mix. ### Implications for policy and supervision - Strengthening asset-quality monitoring and loss-recognition practices can support profitability by limiting deterioration in loan quality. - Policies and supervisory measures that promote cost efficiency may materially affect profitability outcomes. - Incentives or frameworks that encourage the active management of balance-sheet size during stress episodes can be associated with better profit protection. - Monitoring and, where appropriate, reducing reliance on wholesale funding can improve resilience and post-crisis profitability. - Persistence of stable net interest margins suggests that measures aimed solely at restoring margins may have limited short-term payoff; broader structural and demand-side recovery remains important. *Source: IMF Working Paper WP/18/99, April 2018.* ### section II for the definition of subperiods). For listed banks, the extent of the current ### wp1899 - section II for the definition of subperiods). For listed banks, the extent of the current ### I. Main findings and summary - Banks that managed to contain the rise in NPLs were better able to preserve profitability. - Cost discipline mattered: reductions in operating costs translated into better profit performance. - Banks that reduced their assets more aggressively during the crisis experienced a smaller decline in profits in the post-crisis years. - Better profit performance post-crisis is associated with larger decline in the share of wholesale funding. - Net interest margins (NIM) remained broadly stable over the financial cycle. - No evidence that reliance on fees and commission income (diversification of income sources) was associated with better profit performance in the post-crisis environment. ### II. Sample construction and subperiods - Sample: 114 EU banks reporting consolidated balance sheet and income statement data over 2000-2016 (Fitch database). - Exclusions: banks reporting only unconsolidated accounts, subsidiaries (to avoid double-counting), and banks that closed during the sample (survivorship bias acknowledged). - Subperiods defined: 2000-04 (pre-boom), 2005-07 (boom), 2008-12 (crisis), 2013-16 (post-crisis). - Country groups: Eastern EU (Hungary, Poland, Romania, Slovenia), Nordics (Denmark, Finland, Sweden), Other EU (Austria, Belgium, France, Germany, Luxembourg, Netherlands, UK), EC-6 (Cyprus, Greece, Ireland, Italy, Portugal, Spain). - Sample also includes G-SIBs for international comparisons; total assets for U.S. banks adjusted for accounting differences. ### III. Stylized facts: profitability, asset quality, dividends, assets, leverage, funding - Profitability - ROA and ROE rose in the boom, declined dramatically in the crisis, and remained depressed post-crisis. - Nordic banks post-crisis ROE: 9.1 percent (down from 13.8 percent in the pre-boom period). - EC-6 experienced the sharpest decline in ROE (partly due to provisioning for high NPLs). - Net interest margins declined from pre-boom to boom but remained on average broadly stable since the boom. - Weighted averages by G-SIB country (selected figures): - EU G-SIBs: Net Interest Margin by period — Pre-boom 1.4, Boom 1.0, Crisis 1.1, Post-crisis 1.2; ROA — Pre-boom 0.6, Boom 0.6, Crisis 0.2, Post-crisis 0.1; ROE — Pre-boom 13.4, Boom 16.3, Crisis 2.7, Post-crisis 2.4. - China: Net Interest Margin — Pre-boom 2.1, Boom 2.4, Crisis 2.5, Post-crisis 2.4; ROA — Pre-boom 4.6, Boom 10.9, Crisis 18.7, Post-crisis 17.1; ROE — Pre-boom 0.4, Boom 0.7, Crisis 1.1, Post-crisis 1.2. - US: Net Interest Margin — Pre-boom 2.3, Boom 1.9, Crisis 1.9, Post-crisis 1.7; ROA — Pre-boom 15.0, Boom 15.0, Crisis 4.2, Post-crisis 8.4; ROE — Pre-boom 1.0, Boom 1.0, Crisis 0.3, Post-crisis 0.6. - Asset quality and dividends - Dividend behavior: Eastern EU and EC-6 cut dividend payouts post-crisis; Other EU and Nordics increased payout ratios to close to 70 percent post-crisis (from around 50 percent pre-crisis). - For largest banks, post-crisis dividend payout ratios averaged close to 75 percent of earnings. - US G-SIB dividend payout ratio dropped to 18 percent post-crisis. - NPLs / Gross loans (selected weighted averages by region and period): - EC-6: Pre-boom 2.4, Boom 34.1, Crisis 98.9, Post-crisis 17.4. - Eastern EU: Pre-boom 20.6, Boom 37.0, Crisis 48.7, Post-crisis 29.6. - Other EU and Nordics data reported in tables (see sample). - Asset growth - Assets grew at annual rates over 15 percent (pre-boom) and over 20 percent (boom) on average across the sample; contraction in post-crisis period. - Weighted averages by region (Assets annual growth): - EC-6: Pre-boom 13.8, Boom 21.4, Crisis 1.6, Post-crisis -6.2. - Eastern EU: Pre-boom 29.3, Boom 28.7, Crisis 5.4, Post-crisis -0.6. - Other EU: Pre-boom 16.1, Boom 23.0, Crisis -0.6, Post-crisis -7.3. - Nordics: Pre-boom 21.7, Boom 20.8, Crisis 4.1, Post-crisis -4.7. - Median asset growth differs from mean: overall sample mean asset growth ~7.7 percent, median 1.7 percent. - Leverage and capitalization - Tangible common equity / Tangible assets (TE ratio) used as capitalization measure. - TE ratio weighted averages by region (percent): - EC-6: Pre-boom 4.5, Boom 4.4, Crisis 4.2, Post-crisis 5.2. - Eastern EU: Pre-boom 9.5, Boom 8.0, Crisis 9.1, Post-crisis 10.5. - Other EU: Pre-boom 3.2, Boom 2.8, Crisis 2.9, Post-crisis 4.4. - Nordics: Pre-boom 4.2, Boom 3.8, Crisis 3.6, Post-crisis 4.6. - Large banks had lower TE ratio than small banks in pre-boom (3.4 percent vs 6.0 percent). - TE ratio fell to lowest level during the boom (2.8 percent) for the largest banks group. - Post-crisis, EU banks increased capitalization but remained below some non-EU G-SIB peers. - Funding - Wholesale funding rose pre-crisis and fell after market disruptions; customer deposits increased correspondingly (except Nordics). - Wholesale funding ratio defined as (total funding - deposits – equity)/(total funding - equity); includes central bank liquidity provision. - Wholesale funding / Total funding weighted averages by region (percent): - EC-6: Pre-boom 30.3, Boom 38.7, Crisis 32.9, Post-crisis 26.3. - Eastern EU: Pre-boom 3.1, Boom 12.4, Crisis 12.0, Post-crisis 6.4. - Other EU: Pre-boom 23.1, Boom 39.2, Crisis 33.7, Post-crisis 27.4. - Nordics: Pre-boom 39.2, Boom 44.1, Crisis 48.1, Post-crisis 49.0. - Average wholesale funding / Total funding: Pre-boom 25.9, Boom 37.8, Crisis 34.3, Post-crisis 29.3. ### IV. Decomposing the decline in ROA (components and regional patterns) - ROA decomposition identity (accounting-based): change in ROA = ∆Non-interest income – ∆Non-interest expense – ∆Provisions + ∆NIM (as expressed in text; full formula and components provided in source). - Regional patterns in change from boom to post-crisis: - Eastern EU: large decline driven by sharp fall in NIM and non-interest income, and an increase in loan-loss provisions; partially offset by contraction in operating expenses. - EC-6: decline mainly driven by provisioning for NPLs; operational cost savings did not offset provisioning. - Other EU: decline mainly reflected fall in revenue from non-traditional banking activities (other non-interest income category). - Nordics: reduction in income from non-traditional activities and some increased provisioning contributed to lower ROA, but operational cost savings limited deterioration. - NIM specifics: - Only Eastern EU showed a sizable decline in NIM among regions; Other EU experienced an improvement in NIM. - Across banks by size: small banks experienced worsening intermediation spreads, while the largest banks registered a slight increase in NIM. - G-SIBs decomposition: - Intermediation spreads rose for EU G-SIBs but declined for non-EU peers. - All G-SIBs except Swiss and Chinese saw a decline in earnings from non-traditional sources. - EU G-SIBs faced greater increases in operating and provisioning costs relative to other G-SIBs; U.S. G-SIBs compensated with sharp cost cutting. ### V. Empirical approach to explain changes in profitability - Dependent variable: change in ROA between consecutive financial-cycle phases (pre-boom → boom, boom → crisis, crisis → post-crisis). ROA favored over ROE to limit sensitivity to leverage changes. - Regression specification (panel): - ΔROA_ijt = X_{ijt-1} β + ΔZ_ijt γ + ΔM_jt δ + α_t + ε_ijt (notation summarized from source; OLS with robust standard errors clustered at bank level). - X: vector of bank initial characteristics averaged over previous period (pre-determined). - ΔZ: vector of contemporaneous changes in bank variables (in percentage points). - ΔM: changes in country-specific macroeconomic variables (in decimals) in some specifications. - Period fixed effects included; some specifications include regional fixed effects. - Estimation details and robustness: - Regressions estimated using OLS with robust standard errors clustered at the bank level. - Panel has three periods (four stages imply three changes). - Robustness: winsorization at 5 percent for some extreme changers in robustness checks; results robust to such winsorization. - Separate cross-sectional regressions run for each phase due to potential changing impacts over the cycle. ### VI. Bank-level explanatory variables used - Initial conditions (pre-determined, averaged over previous period): - Bank size (log of assets). - Past expansion (asset growth rate over previous subperiod). - Funding model (share of wholesale funding in total funding). - Loan portfolio quality (NPLs / net loans). - Share of interest income in total income (proxy for reliance on traditional intermediation). - Initial tangible equity / total tangible assets (buffer measure). - Initial profitability to control for mean reversion. - Contemporaneous changes (from t-1 to t): - Change in funding model (change in share of wholesale funding). - Change in operating costs (cost efficiency adjustments). - Change in reliance on fees and commission income (income diversification). - Change in NPL ratio (deterioration in loan quality). ### VII. Macroeconomic controls and considerations - Country-specific macroeconomic proxies experimented with include: - Real GDP growth, output gap, unemployment rate. - Policy rate proxies, term premium, aggregate credit growth. - Stock price growth, housing price growth. - Sovereign stress: spreads relative to bunds. - Difference between average system-wide deposit interest rates and policy rate (to capture deposit market rigidities or regulatory floors). - Acknowledgement: several EU banks have significant cross-border operations; domestic macro variables may not fully capture the macroeconomic environment relevant for these banks. ### VIII. Sample summary statistics (selected) - Profitability (ROA): overall sample mean 0.24 percent; median 0.31 percent. - Asset growth: overall sample mean ~7.7 percent; median 1.7 percent. - Wholesale funding: average about 30 percent of total funding (wide dispersion across banks). - NPLs: average 6.4 percent of gross loans; median 3.9 percent. - Interest income share: on average 80 percent of total income. - Tangible common equity: on average about 5 percent of total tangible assets. *Italic: Source — wp1899 (section II and surrounding sections) — IMF staff working paper content provided in the supplied PDF content.* ### 0.21 percentage points on average from one period to the next, with a median decline of - ### wp1899 - 0.21 percentage points on average from one period to the next, with a median decline of - ### Key empirical facts and aggregate changes - ROA declined on average by 0.21 percentage points from one period to the next, with a median decline of -0.06 percentage points, driven by a tail of banks with large declines in ROA. - The share of fees and commission income was approximately stable on average, but with significant heterogeneity across banks. - The share of wholesale funding: average decline of 1.72 percentage points; median decline of 0.87 percentage points; large dispersion across banks. - Ratio of non-interest expenses to total assets declined, indicating widespread cost cutting. - The ratio of non-performing loans (NPL) to gross loans increased on average by 3 percentage points. ### Pairwise correlations (summary) - Low correlation between ROA and ROE, implying substantial changes in bank leverage over the sample. - ROA positively correlated with asset growth and capitalization (tangible common equity to tangible asset). - ROA negatively correlated with level of NPLs and change in NPLs. - Change in ROA positively correlated with change in operating expenses. ### Panel regressions — determinants of changes in profitability - Initial profitability (previous period ROA) enters negatively: higher past profits predict larger subsequent profit declines (mean reversion). - A 100-basis points higher initial ROA is associated with an additional decline in profits of some 80 basis points in the subsequent phase of the financial cycle. - Asset growth negatively associated with subsequent change in ROA. - A ten percentage-point higher asset growth in the previous phase is associated with a subsequent decline in profitability of an additional 12 basis points. - Unweighted average decline in ROA in the sample is 21 basis points. - Capitalization (TCE/TA) positively associated with subsequent ROA but effect is small: - An increase in tangible capital ratio of 1 percentage point raises ROA by slightly over 4 basis points. - Coefficient significant only at the 10 percent level. - Period effects: - Crisis period associated with decline in ROA of some 50 basis points. - Post-crisis period associated with a further decline of some 50 basis points. - Regional effects: - “Other EU” dummy negative and significant: estimated 18 basis points lower change in profits than Nordic banks over the full cycle (Nordics omitted). - Interaction of region and period: - EC-6 and Eastern EU had higher profit growth than Nordics during the boom but this reversed during the crisis. - EC-6 * crisis coefficient: -0.584***. - Eastern EU * crisis coefficient: -0.781***. - Eastern EU * post crisis coefficient: -1.307**. - Macroeconomic variables (when included instead of region dummies): - Stock price growth positive for bank profits (coefficient 1.544*** reported). - Spread to Bunds negative for profits (coefficient -0.109*** reported). - Deposit rate minus short-term rate: widening associated with declining profitability (coefficient -0.237* in one specification). - Real GDP growth not significantly correlated with bank profitability changes in primary specifications. - Contemporaneous bank-level changes: - Changes in share of fee & commission income: not systematic in most specifications. - Reduction in wholesale funding correlated with stronger profit growth (or smaller decline): - Wholesale funding change coefficient: about -0.0132*** (i.e., decline in wholesale funding associated with higher ROA change). - Magnitude: a 5-percentage point decline in wholesale funding to assets → increase in ROA of 8 basis points. - Operating cost efficiency: - Lowering non-interest expenses / TA by 100 basis points → increase in ROA by some 30–40 basis points. - Asset quality: - A 5 percentage points increase in NPL ratio → additional decline in ROA by more than 30 basis points. ### Cross-sectional regressions — phases of the financial cycle - Mean reversion strongest in transitions: - Boom to crisis and crisis to post-crisis show pronounced mean reversion. - Mean reversion smaller in pre-boom to boom transition. - Asset growth negative relationship with subsequent profit change particularly pronounced post-crisis (significant in larger sample). - Cost cutting and asset quality protection mattered more during crisis and post-crisis than during boom. - Peak NPL effect on profitability occurred during the crisis period. - Benefits of switching away from wholesale funding mainly visible in post-crisis period; small/insignificant during the crisis (possible fire-sale dynamics during crisis). ### Probit regressions — characteristics of highly profitable post-crisis banks - Definition: highly profitable banks = top quintile of ROE distribution during post-crisis period; ROE ≥ 9 percent. - Country-group effects: - EC-6 dummy significant and negative (coefficient -0.949* in one specification). - Other EU dummy significant and negative (coefficient -1.414*** in one specification). - Eastern EU coefficient reported as 0.543 in one specification (not statistically significant across all). - Macro variables (post-crisis averages): - Short-term interest rate positive and significant (ST rate coefficient 0.841*** in one specification), implying higher short-term rates increase probability of high profitability. - Deposit rate minus ST rate coefficients reported as -0.395 and -0.154 in specifications (not significant across all). - Bank-specific characteristics associated with higher probability of post-crisis high profitability: - Smaller size (log(assets) negative and significant: e.g., -0.193**). - Better loan quality (NPL/GL negative and highly significant: e.g., -0.102***). - Some evidence that lower reliance on interest income relates to higher profitability, though this is less robust across specifications (interest income share coefficients include -0.677, -4.224***, -4.093***). - Reliance on wholesale funding by itself is not robustly associated with being highly profitable (share of wholesale funding coefficients not significant across specifications). - Asset growth marginally significant with negative coefficient in some specifications (e.g., -0.00189*). ### Conclusions and interpretive findings - 2000–2007: rapid balance sheet expansion financed by rising share of wholesale funding; leverage rose without a commensurate increase in high-quality capital; returns on equity mechanically boosted. - From 2008: reversal—wholesale funding withdrew, banks downsized and delevered; deteriorating loan quality in some countries; state liquidity and capital injections were often necessary. - Post-crisis: profitability remained low due to lower leverage, difficulty cutting costs while downsizing, and persistent high NPLs for some banks. - Net interest margins remained broadly stable over the financial cycle in the sample: interest earnings and expenses fell in lockstep during downturns, leaving margins little changed. - Low interest rate environment not yet visible in the data as large adverse effect on profitability, but prolonged low rates pose a challenge going forward. ### Policy-relevant implications and behavioral findings - Containing NPLs materially supports profitability; differences across banks suggest credit risk management and institutional settings matter. - Cost discipline (reducing non-interest expenses) significantly protects profits; consolidation, branch reduction, technology adoption, and compensation policies are plausible channels. - Rapid disposal of non-performing or non-core assets during crisis may improve post-crisis profitability, but could cause macroprudential trade-offs (credit crunch, corporate insolvency). - Reorienting funding away from wholesale markets and toward deposits associated with better post-crisis profit outcomes; the quality and orderliness of the funding shift matters (fire sales during crisis may negate benefits). - In the post-crisis environment, higher short-term interest rates increase likelihood of banks earning ROE ≥ 9 percent, suggesting interest-rate environment matters for bank capital returns. *Italicized source: wp1899 — IMF working paper content (extracts from Tables 8–10 and accompanying text).* ### References ### References ### Bibliographic references - Acharya, V., and S. Steffen, 2014, “Falling Short of Expectations? Stress Testing the European Banking System,” CEPS Policy Brief No. 315. - Alessandri, P., and B. Nelson, 2014, “Simple Banking: Profitability and the Yield Curve,” Journal of Money, Credit and Banking 47, 143–175. - Bonaccorsi di Patti, Emilia, and Anil Kashyap, 2017, Which Banks Recover from Large Adverse Shocks, NBER Working Paper No. 23654. - Borio, C., L. Gambacorta, and B. Hofmann, 2015, “The Influence of Monetary Policy on Bank Profitability,” BIS Working Papers, No. 514 (October). - Borio, C, 2014, “The Financial Cycle and Macroeconomics: What Have we Learnt?”, Journal of Banking and Finance 45, 182-198. - Busch, R., and C. Memmel, 2015, “Bank’s Net Interest Margin and the Level of Interest Rates,” Deutsche Bundesbank Discussion Paper, No. 16/2015. - Calomiris, C. and D. Nissim, “Crisis-Related Shifts in the Market Valuation of Banking Activities,” Journal of Financial Intermediation 23, 400–435. - Claessens, S. N. Coleman, and M. Donnelly, 2016, “Low-for-long" Interest Rates and Net Interest Margins of Banks in Advanced Foreign Economies. Federal Reserve Board of Governors International Finance Discussion Papers Notes, April 11, 2016. - Constancio, Vitor, 2017, Challenges Faced by the European Banking Sector, speech at the Italian Banking Association, Rome, June 14. - Demirgüç-Kunt, A., E. Detragiache, and O. Merrouche, 2013, “Bank Capital: Lessons from the Financial Crisis,” Journal of Money, Credit, and Banking, Vol. 45, No. 6, 1147–1164. 2013. - Ennis, H., H. Fesseden, and J. Walter, 2016, “Do Net Interest Margins and Interest Rates Move Together?” Economic Brief 16–05, Federal Reserve Bank of Richmond. - European Central Bank, “Financial Stability Review,” November 2016, Euro Area Financial Institutions. - International Monetary Fund (IMF), Global Financial Stability Report, 2016a, October, Washington DC. - International Monetary Fund (IMF), 2016b, France – Selected Macro-financial Issues, Country Report No. 16/228, July, Washington DC. - International Monetary Fund (IMF), 2017a, Global Financial Stability Report, April, Washington DC. - International Monetary Fund (IMF), 2017b, Spain Financial Sector Assessment Program – Technical Note: The Determinants of Bank Profitability, December, Washington DC. - Kok, Christoffer, Csaba Móré and Cosimo Pancaro, 2015, Bank Profitability Challenges in Euro Area Banks: The Role of Cyclical and Structural Factors, European Central Bank, Financial Stability Review, pp. 134-145. - McCauley, Robert N., Agustín S. Bénétrix, Patrick M. McGuire, and Goetz von Peter, 2017, Financial Deglobalisation in Banking?, BIS Working Papers No. 650, June. - Roengpita, Rungporn, Nikola Tarashev, and Kostas Tsatsaronis, 2014, Bank Business Models, BIS Quarterly Review, December Issue, pp. 55-65. - Sarin, N. and L. Summers, 2016, Have Big Banks Gotten Safer? Brookings Papers on Economic Activity. - Turk, Rima, 2016, Negative Interest Rates: How Big a Challenge for Large Danish and Swedish Banks? IMF WP/16/198, Washington DC. ### Data Appendix overview - The Data Appendix contains: - Table A1. Number of Banks by Country. - Table A2. List of G-SIBs. - Table A3. Initial Conditions and Changes over Time (summary statistics). - Table A4. Cross-correlations (variables in level and in change). ### Table A1. Number of Banks by Country (Total 114) - Austria 12 - Belgium 3 - Cyprus 2 - Denmark 5 - Finland 1 - France 7 - Germany 19 - Greece 4 - Hungary 3 - Ireland 3 - Italy 18 - Luxembourg 3 - Netherlands 5 - Poland 1 - Portugal 5 - Romania 1 - Slovenia 2 - Spain 7 - Sweden 4 - UK 9 - Total 114 ### Table A2. List of G-SIBs (Total 29) - China: Agricultural Bank of China - China: Bank of China - China: China Construction Bank - China: Industrial and Commercial Bank of China - France: BNP Paribas - France: Groupe Credit Agricole - France: Societe Generale - Germany: Deutsche Bank - Italy: UniCredit - Japan: Mitsubishi UFJ FG - Japan: Mizuho FG - Japan: Sumitomo Mitsui FG - Netherlands: ING Bank - Spain: Banco Santander - Sweden: Nordea Bank - Switzerland: Credit Suisse - Switzerland: UBS - United Kingdom: Barclays - United Kingdom: HSBC - United Kingdom: Standard Chartered - United Kingdom: Royal Bank of Scotland - United States: Bank of America - United States: Bank of New York Mellon - United States: Citigroup - United States: Goldman Sachs - United States: JP Morgan Chase - United States: Morgan Stanley - United States: State Street Corporation - United States: Wells Fargo - Total 29 ### Table A3. Initial Conditions and Changes over Time (summary statistics) - Variables in level - ROA 0.24 Median 0.31 Std. Dev. 0.80‐ Min -3.77 Max 2.86 - Log (assets) 11.42 Median 11.25 Std. Dev. 1.52 Min 7.18 Max 14.81 - Asset growth 7.68 Median 1.67 Std. Dev. 41.08 Min -34.99 Max 754.69 - Wholesale funding / total funding 30.90 Median 27.89 Std. Dev. 21.11 Min 0.60 Max 100.00 - Non performing loans / gross loans 0.00 Median 3.91 Std. Dev. 0.00 Min 0.00 Max 0.00 - Share of interest income 6.39 Median 0.81 Std. Dev. 7.46 Min 0.00 Max 51.26 - Tangible common equity / total equity 0.80 Median 4.63 Std. Dev. 0.27 Min -2.06 Max 4.28 - Variables in change - ROA -2.85 Median -0.06 Std. Dev. 125.10 Min -1490.81 Max 1470.31 - Fees and commission inc. / total income 0.01 Median 0.01 Std. Dev. 0.09 Min -0.68 Max 0.66 - Wholesale funding / total funding -1.83 Median -0.88 Std. Dev. 11.59 Min -83.51 Max 32.86 - Non interest expenses / total assets -0.21 Median -0.07 Std. Dev. 1.19 Min -19.70 Max 2.36 - Non performing loans / gross loans 2.98 Median 1.30 Std. Dev. 5.77 Min -7.99 Max 38.17 ### Table A4. Cross-correlations (selected entries; variables in level and in change) - Variables in level (pair : correlation) - ROA : ROE 0.1504 - ROA : Log (assets) -0.2084 - ROA : Asset growth 0.6842 - ROA : Wholesale funding / total funding -0.1221 - ROA : Non performing loans / gross loans -0.4132 - ROA : Share of interest income -0.196 - ROA : Tangible common equity / total equity 0.4609 - ROE : Log (assets) -0.0404 - Log (assets) : Asset growth -0.1704 - Wholesale funding / total funding : Log (assets) 0.2855 - Non performing loans / gross loans : ROE -0.0296 - Share of interest income : Wholesale funding / total funding 0.2674 - Tangible common equity / total equity : Asset growth 0.4082 - Variables in change (pair : correlation) - ROA : ROE -0.7173 - ROA : Log (assets) -0.1557 - ROA : Asset growth 0.1194 - ROA : Wholesale funding / total funding -0.5918 - ROA : Non performing loans / gross loans 0.1005 - ROA : Fees and commission inc. / total income 0.1328 - ROA : Non interest expenses / total assets 0.114 - ROE : Fees and commission inc. / total income -0.0677 - Fees and commission inc. / total income : Wholesale funding / total funding -0.0614 - Wholesale funding / total funding : Non interest expenses / total assets 0.1972 - Non interest expenses / total assets : Non performing loans / gross loans -0.6152 - Non performing loans / gross loans : ROE 0.0994 *Source: wp1899 - References (Data Appendix tables and reference list).* --- _Source: https://www.imf.org/-/media/files/publications/wp/2018/wp1899.pdf_