## 1. Evolution of Total Bank Deposits in Select Euro Area Countries, 2008–12

## Source details

**Canonical URL:** [1. Evolution of Total Bank Deposits in Select Euro Area Countries, 2008–12](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2016/_wp16145.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2016/_wp16145.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2016/_wp16145.pdf.json)

---

### I. Introduction and motivation
- Post-2008 crisis bank responses:
  - Asset-side deleveraging varied across countries; liability-side response was uniform: reduced use of wholesale, short-term, and cross-border funding in favor of more stable domestic retail funding.
- Deposit outflows example:
  - A total of €326 billion was drawn from banks in Ireland, Greece, Portugal, and Spain in a 12-month period ending in July 2012.
  - Only about a quarter of this decline was due to core European banks’ reduction of exposure to the Eurozone “periphery”; the remainder was a decline in domestic customer deposits.
- Deposit-cost examples:
  - Average deposit rates in Italy jumped 50 percent to 3.1 percent in July 2013 from a year earlier; German average deposit cost in July 2013 was 1.5 percent; two years prior both Italian and German deposit rates were 1.3 percent.
- Sovereign risk framing:
  - Sovereign risk split into solvency and liquidity components.
  - Conventional measures focus narrowly on default probability; a wider construct would include core fiscal variables, balance sheet developments, debt portfolio structure, investor base and cross-border linkages, and financial assets.
- Data note:
  - Figure 1 (IMF, International Financial Statistics) shows Evolution of Total Bank Deposits in select Euro Area countries, 2008–12. Spain series includes deposits created by securitizations and affiliates issuing preference shares and other financial instruments.

### II. Public sector borrowing and contingent liabilities
- Aggregate borrowing and debt:
  - For 2003–08, average net borrowing in the United States, Euro area, and the United Kingdom was around 2.5–3.3 percent of GDP.
  - Net borrowing rose to 6-15 percent of GDP in 2009 and 2010.
  - Public debt forecast to reach 110 percent of GDP on average by 2015—nearly 40 percentage points above pre-crisis levels.
- Contingent liabilities:
  - Explicit contingent liabilities (guarantees on financial institution debt and implicit signals) increased sharply.
  - Table 1 (Outstanding Stock of Government Guarantees, in percent of GDP) — selected entries:
    - Ireland: 2010 96.0; 2011 79.8; 2012 66.1
    - Austria: 2010 53.4; 2011 44.5; 2012 38.9
    - United Kingdom: 2010 27.8; 2011 15.4; 2012 10.3
    - Germany: 2010 20.3; 2011 18.7; 2012 17.9
    - EU Average: 2010 13.7; 2011 13.3; 2012 12.7
  - Stylized fact: countries with highest debt-to-GDP ratios tend to have the highest proportion of government debt held by banks, exacerbating the sovereign–bank nexus.

### III. Sovereign risk and depositor confidence
- Transmission channels from sovereign risk to banks:
  - Asset-side: valuation losses on bank holdings of sovereign securities (direct sovereign risk).
  - Liability-side: increase in bank funding costs due to repricing and credit rating downgrades (indirect sovereign risk).
- Feedback loop risk: deteriorating bank balance sheets can increase sovereign debt burden via required assistance.
- Evidence of perceived credibility limits:
  - Deposit insurance may be viewed as less credible by savers under elevated sovereign risk, prompting deposits to be reduced or reallocated.
- ECB reporting:
  - ECB July 2012 questionnaire: 18 percent of European banks attributed deterioration in funding conditions to the sovereign debt crisis (versus 4 percent in Q1 2012).

### IV. Empirical estimation

H3 Modeling and measurement
- Reduced-form estimating equation:
  - Deposits_change_{tji} = β1 Performance_{ji,t-1} + β2 SovereignRisk_{ji,t-1} + β3 Controls_{ji,t-1} + ε_{tji}
  - Deposits change measured at bank i in country j between t and t-1; Performance is a composite bank performance index; SovereignRisk is bank’s exposure to sovereign risk; Controls include market share, unemployment, inflation.
- Sovereign risk proxy (innovative):
  - Factiva news-count: number of articles where “sovereign risk” or “government debt” appear with the bank name for each country-year.
  - Demeaned country-adjusted Factiva used:
    - SovereignRisk_{tji} = Factiva_{tji} - Factiva_{tj}
  - Non-English translations of search terms applied where needed.
- Bank-level data:
  - Annual financial data from Bankscope covering 2006–11.
  - Summary sample statistics (No. of observations = 957):
    - Change in deposits, million euros: Mean 1.84; Standard Deviation 15.53; Minimum -62.58; Maximum 87.39
    - Performance (DEA), a [0, 1] index: Mean 0.60; Standard Deviation 0.15; Minimum 0.28; Maximum 0.89
    - Inflation, percent: Mean 2.95; Standard Deviation 2.29; Minimum -1.09; Maximum 12.35
    - Sovereign risk (Factiva, demeaned) count: Mean 11.57; Standard Deviation 75.46; Minimum -71.00; Maximum 483.31
    - Market share, ratio: Mean 0.11; Standard Deviation 0.14; Minimum 0.0001; Maximum 0.65
    - Unemployment, percent: Mean 8.79; Standard Deviation 4.13; Minimum 3.20; Maximum 21.60
- Performance measure:
  - Data Envelopment Analysis (DEA) produced composite [0,1] bank efficiency/performance indices relative to peers.
  - DEA scores corrected for efficiency bias following Simar and Wilson (1998, 2000).

H3 Baseline regression results
- Estimation: fixed effects regressions for 2006–11 and subsamples 2006–09 and 2010–11.
- Core empirical findings (2006–11, sample: 29 countries, 957 observations):
  - Perceived sovereign risk (Factiva, demeaned) has a negative and significant impact on growth of deposits in almost all specifications.
  - Bank efficiency (DEA) positively affects consumer deposit growth.
  - Macroeconomic controls (contemporaneous inflation and unemployment) do not have a direct impact on bank deposit growth in baseline regressions.
  - Market share not statistically significant in DEA-based specification; limited evidence of “too big to fail” in DEA specifications.
- Time-split differences:
  - 2006–09 (pre-crisis):
    - Performance positive and significantly related to deposit growth.
    - Sovereign risk not important.
    - Inflation positively and significantly related to deposit growth.
  - 2010–11 (during crisis):
    - Performance effect vanishes.
    - Perception of sovereign exposure becomes an important negative predictor of deposit behavior.
    - Inflation no longer significant.
- Non-linearity and high-exposure effects:
  - Dummy75pc = 1 if Sovereign Risk is in the upper quartile for the country-year.
  - Interaction shows banks strongly exposed to sovereign-related news may face higher risk of losing deposits.
  - During crisis, effect is negative for fourth quantile: coefficient equals -0.137 (statistically significant).
- Selected regression coefficients (Fixed Effects, dependent variable: Change in Nominal Deposits; 2006–11):
  - Performance (DEA): 13.699***  [3.052]
  - Sovereign Risk (Factiva): -0.031**  [-2.413]
  - Sovereign Risk (Factiva) x Dummy75pc: -0.153***  [-3.355]
  - Inflation (2006–09): 1.689***  [2.800]
  - Constant (2006–11): -8.021***  [-2.830]
  - Observations: 957; Number of countries: 29
  - R-squared examples: 0.117; 0.162; 0.149
  - Significance: *** p<0.01, ** p<0.05, * p<0.1

H3 Robustness tests
- Alternative performance measure:
  - Replaced DEA with Return on Equity (ROE); results largely consistent.
  - With ROE, market share became positive and statistically significant—suggesting a “too large to fail” effect in ROE specifications.
- Alternative Factiva construction:
  - Reintroduced the country component (not demeaned); results did not materially change.
- Additional controls:
  - Included BIS data on home banking sector exposure to country groups (United States, core Eurozone, peripheral Eurozone, emerging Europe, non-Eurozone Advanced Europe).
- Robustness table highlights (selected coefficients and examples):
  - Performance (DEA) coefficient: 13.555***  [3.038]
  - Alternate Performance (ROE) coefficients: 0.102**  [2.326]; 0.055**  [2.392]; 0.085***  [3.473]
  - Sovereign Risk (demeaned Factiva) coefficients: -0.031**  [-2.409]; -0.030***  [-2.845]; -0.027**  [-2.150]; -0.037***  [-2.825]
  - Sovereign Risk (level of Factiva) examples: -0.003  [-0.080]; -0.033***  [-3.035]; -0.033***  [-2.593]
  - Country-group exposures (nominal home banking sector exposure to sovereigns):
    - United States: -2.08E+05  [-0.908]
    - Emerging Europe: -3.9e+05**  [-2.320]
    - Non-EZ Advanced Europe: 2.05E+05  [0.765]
    - Core EZ: -8.24E+03  [-0.0883]
    - EZ “Periphery”: 1.71E+05  [0.489]
  - Observations across columns: 957; 575; 382; 957; 575; 382; 957; 533
  - R-squared examples: 0.154; 0.156; 0.119; 0.162; 0.153; 0.121; 0.167

### V. Key empirical conclusions
- Core result:
  - Exposure of the banking sector to sovereign risk negatively affected growth of consumer deposits over and above macroeconomic conditions (sample: 29 European countries, 2006–11).
- Timing and nonlinearity:
  - Holdings of sovereign bonds benefited banks during tranquil times.
  - Banks frequently mentioned in sovereign-related news during crisis were penalized by depositors.
  - In tranquil times depositors reward financial performance; in crisis times perceived sovereign risk dominates depositor behavior.
- Country-group perceptions:
  - Coefficients on exposures to sovereigns in EZ “Periphery,” Emerging Europe, and United States are negative; only Emerging Europe coefficient is statistically significant—suggesting Emerging Europe may be perceived as riskiest.
- Too-big-to-fail:
  - With one exception, no consistent evidence of “too big to fail” from depositors’ perspective.
- Macro-financial implications:
  - Excessive sovereign exposure may have contributed to deposit outflows in Europe after the crisis.
  - The negative link between sovereign risk and deposit collection can limit banks’ lending capacity and amplify procyclicality in fiscal shock transmission.
  - Recommendation to include banks explicitly in macro models to capture bank-mediated pass-through of fiscal shocks.

### VI. Policy-relevant analysis and recommendations
- Measures to weaken sovereign risk–deposit contraction link:
  - Fiscal tightening
  - Liability management exercises (aimed at reducing short-run gross borrowing needs)
  - Loss-sharing mechanisms
- Regulatory caution:
  - New regulations that could encourage banks to hold more government debt (e.g., for liquidity purposes) should be introduced with care and mitigated where possible.
- Strengthening resilience:
  - Improve bank efficiency to boost depositor confidence.
  - Build sufficient buffers:
    - For banks: more capital and liquidity.
    - For sovereigns: lower levels of debt and quasifiscal activity.
- Modeling recommendation:
  - Consider more explicit inclusion of banks in traditional macro models to account for bank-mediated pass-through of fiscal shocks.

### VII. DEA methodology (Appendix I) and selected Appendix II indices
- DEA summary:
  - Data Envelopment Analysis (DEA) produces efficiency score θ in [0, 1]; θ = 1 implies on the frontier (efficient).
  - Inputs: Labor; Fixed assets; Interest expenditures.
  - Outputs: Revenues (sum of interest and non-interest income); Net loans (loans net of loss provisions); Liquid assets (cash, balances with monetary authorities, holdings of treasury bills).
  - DEA primal and dual linear programming formulations reported in the source; efficiency indexes decomposed into pure technical and scale efficiency.
- Selected Average DEA indices by country (2006–2011 averages and selected annual values):
  - Austria: 2006 0.769; 2007 0.601; 2008 0.516; 2009 0.578; 2010 0.608; 2011 0.596; Average 0.611
  - Belgium: 2006 0.559; 2007 0.657; 2008 0.600; 2009 0.609; 2010 0.760; 2011 0.844; Average 0.672
  - Cyprus: 2006 0.443; 2007 0.481; 2008 0.466; 2009 0.626; 2010 0.598; 2011 0.517; Average 0.522
  - Sweden: 2006 0.848; 2007 0.827; 2008 0.850; 2009 0.807; 2010 0.792; 2011 0.698; Average 0.804
  - United Kingdom: 2006 0.626; 2007 0.628; 2008 0.627; 2009 0.693; 2010 0.677; 2011 0.752; Average 0.667
  - (Appendix II lists all 29 countries’ annual DEA indices for 2006–2011 and averages; selected values shown above.)

### VIII. Appendix III — country classifications (selected groupings)
- EU 27: Austria; Belgium; Bulgaria; Cyprus; Czech Republic; Denmark; Estonia; Finland; France; Germany; Greece; Hungary; Ireland; Italy; Latvia; Lithuania; Luxembourg; Malta; Netherlands; Poland; Portugal; Romania; Slovakia; Slovenia; Slovenia; Spain; Sweden; United Kingdom
- EZ 17: Austria; Belgium; Cyprus; Czech Republic; Estonia; Finland; France; Germany; Greece; Ireland; Italy; Luxembourg; Malta; Netherlands; Portugal; Slovakia; Slovenia; Spain
- EZ “Periphery”: Austria; Belgium; Cyprus; Finland; France; Germany; Greece; Ireland; Italy; Luxembourg; Netherlands; Portugal; Slovakia; Slovenia; Spain
- Core EZ: Austria; Belgium; France; Germany
- Emerging Europe: Bulgaria; Czech Republic; Hungary; Latvia; Lithuania; Poland; Romania; Slovakia; Slovenia
- Non-EZ Advanced Europe: Denmark; Estonia; Finland; Ireland; Latvia; Lithuania; Sweden; United Kingdom; Norway; Switzerland

*Content extracted from _wp16145 - References (PDF), IMF.*

### 1. Evolution of Total Bank Deposits in Select Euro Area Countries, 2008–12 .........................5

### 1. Evolution of Total Bank Deposits in Select Euro Area Countries, 2008–12

### Major sections in this content unit
- 1. Evolution of Total Bank Deposits in Select Euro Area Countries, 2008–12
- 2. Total Net Borrowing Needs of the Public Sector as of 2010

### Tables included
- 1. Outstanding Stock of Government Guarantees
- 2. Summary Statistics
- 3. Fixed Effects Regressions, 2006–11
- 4. Robustness Tests

### Appendices included
- I. Description of DEA Methodology and Choice of Inputs/Outputs
- II. Average Data Envelopment Analysis Indices by Country and Year
- III. Country Classifications

*Source: _wp16145 - 1. Evolution of Total Bank Deposits in Select Euro Area Countries, 2008–12 (PDF chapter/section).*

### References................................................................................................18

### _wp16145 - References................................................................................................18

### I. INTRODUCTION AND MOTIVATION

- Post-2008 crisis, European banks restructured and deleveraged; asset-side deleveraging varied across countries (see IMF, 2013; p. 23), while liability-side response was uniform: reduced use of wholesale, short-term, and cross-border funding in favor of more stable domestic retail funding.
- Some countries experienced massive deposit outflows: a total of €326 billion was drawn from banks in Ireland, Greece, Portugal, and Spain in a 12-month period ending in July 2012.
- Only about a quarter of this decline in deposits was due to core European banks’ reduction of exposure to the Eurozone “periphery”; the residual was accounted for by a decline in domestic customer deposits.
- Examples of deposit-cost consequences:
  - Average deposit rates in Italy jumped 50 percent to 3.1 percent in July 2013 from a year earlier; for comparison, average deposit cost at German banks in July was 1.5 percent and two years prior both Italian and German deposit rates were 1.3 percent.
- Crisis fiscal effects and sovereign risk:
  - Combined fiscal support and loss of tax revenues deteriorated fiscal balances, increased public debt, and elevated contingent liabilities.
  - Public debt composition, maturity, and holder profile altered due to investor risk aversion.
  - Sovereign risk concerns included contagion among sovereign, quasi-sovereign, and financial sectors and protracted-recovery-related debt sustainability problems.
- Definitions and considerations:
  - Sovereign risk split into solvency and liquidity components; solvency risk defined as risk that a country is not able to meet present value of obligations evaluated at interest rates commensurate with the debt stock and primary balance; liquidity risk defined as the risk that the government is unable to discharge obligations despite being solvent due to lack of liquidity.
  - Conventional sovereign risk measures focus narrowly on default probability; a wider construct (core fiscal variables plus balance sheet developments, debt portfolio structure, investor base and cross-border linkages, and financial assets of a country) would be more meaningful but is not fully integrated into standard tools.
- Figure 1: Evolution of Total Bank Deposits in select Euro Area countries, 2008–12 (IMF, International Financial Statistics). (Note: Spain series includes deposits created by securitizations and affiliates issuing preference shares and other financial instruments.)

### Public sector borrowing and contingent liabilities

- Total annual net sovereign borrowing in advanced countries increased sharply in 2009 and 2010 (IMF, 2010).
  - For 2003–08, average net borrowing in the United States, Euro area, and the United Kingdom was around 2.5–3.3 percent of GDP.
  - Net borrowing rose to 6-15 percent of GDP in 2009 and 2010, and was projected to decline only moderately in 2011.
  - Public debt forecast to reach 110 percent of GDP on average by 2015—nearly 40 percentage points above pre-crisis levels.
- Public contingent liabilities increased sharply (explicit guarantees on debt issued by financial institutions and implicit signals to support banking sectors).
  - As of mid-2010, explicit contingent liabilities represented a significant proportion of GDP for some issuing European countries (see Table 1).

### Sovereign risk and depositor confidence

- Given elevated sovereign risk, deposit insurance may not be regarded as fully credible by savers, leading to deposit reduction or reallocation to safer/better-performing banks.
- Deposit-sourcing channel: sovereign risk could transfer to banks via:
  - (i) Asset side: valuation losses on bank holdings of sovereign securities (direct sovereign risk).
  - (ii) Liability side: increase in bank funding costs due to repricing of risk and credit rating downgrades (indirect sovereign risk).
- Feedback loop: deteriorating bank balance sheets may require direct assistance, increasing sovereign debt burden.
- Eurozone context:
  - ECB July 2012 questionnaire: 18 percent of European banks attributed deterioration in funding conditions to the sovereign debt crisis (versus 4 percent in Q1 2012), via direct exposures to sovereign debt, reduced collateral value of government bonds, or other effects (ECB, 2012; Box 2).
- Research focus: link between sovereign risk and liability side of banks’ balance sheets in 29 European countries during 2006–11; main question—how sovereign risk affects depositors’ willingness to maintain deposits in a particular bank.

### Table 1. Outstanding Stock of Government Guarantees (in percent of GDP)
- 2010 / 2011 / 2012
  - Ireland: 96.0 79.8 66.1
  - Austria: 53.4 44.5 38.9
  - United Kingdom: 27.8 15.4 10.3
  - Germany: 20.3 18.7 17.9
  - EU Average: 13.7 13.3 12.7
- Source: Eurostat.

- Stylized fact: countries with highest debt-to-GDP ratios tend to have the highest proportion of government debt held by banks, exacerbating the sovereign-bank nexus and amounting to a form of financial repression (recently present in advanced economies as well).

### II. LITERATURE REVIEW (selected findings)

- Market discipline theory: stakeholders (depositors, bondholders, shareholders) influence bank behavior; monitoring can signal bank riskiness.
- Empirical evidence on depositor response to bank risk:
  - Park and Peristiani (1998): riskier U.S. thrifts (1987–1991) faced declining deposits and paid higher interest.
  - Goldberg and Hudgins (2002): similar results over a larger timeframe.
  - Martinez Peria and Schmukler (2001): Argentina, Chile, Mexico—depositors punished high-risk banks with higher interest demands or reduced exposure.
- Bondholder discipline:
  - Flannery (1998), Morgan and Stiroh (2001), Jagtiani and Lemieux (2001), DeYoung et al. (2001): higher risk associated with higher bond spreads; subordinated debt provides informative signals.
  - Evanoff, Jagtiani, and Nakata (2011): mandatory subordinated debt programs may improve market discipline.
- Deposit insurance effects:
  - Demirgüç-Kunt and Huizinga (2004): deposit insurance reduces sensitivity of deposit rates to bank risk.
  - Nier and Baumann (2006): implicit government guarantees weaken market discipline; uninsured funds and informational disclosure strengthen it.
  - Current study does not control for deposit insurance because of (1) relatively minor differences in coverage across European countries and (2) relative ineffectiveness during systemic crises.
- Sovereign debt exposure and bank performance:
  - Angeloni and Wolff (2012): banks’ stock performance affected by exposure to Greek, Italian, Irish sovereign debt in late 2011; Spanish sovereign debt had less impact.
  - Acharya et al. (2014): banks dependent on stressed euro-area economies reduced lending, negatively affecting firm employment, investment, and sales growth.
  - Popov and van Horen (2013): holdings of impaired foreign sovereign debt impacted European banks’ lending.

### III. EMPIRICAL ESTIMATION

#### A. Modeling and Measurement

- Reduced-form estimating equation (consistent with market discipline literature):
  - Deposits_change_{tji} = β1 Performance_{ji,t-1} + β2 SovereignRisk_{ji,t-1} + β3 Controls_{ji,t-1} + ε_{tji}
  - Notation in text: tji Deposits , , Δ is change in deposits in bank i in country j between periods t and t-1; ji ePerformanc , is composite bank performance; ji iskSovereignR , is bank’s exposure to sovereign risk; ji Controls , is vector of bank- and country-specific controls (market share, unemployment, inflation).
- Limitations of conventional sovereign risk proxies:
  - Holdings of sovereign treasury securities measure net rather than gross effect and lack disaggregated by-country holdings data.
  - Most depositors lack access to bank-specific sovereign holdings data; mainstream media reports are more accessible.
- Innovative proxy for perceived sovereign risk:
  - Factiva news-count measure: count of articles where terms “sovereign risk” or “government debt” appear in the same article with the bank name for each country-year.
  - Notation example: Factiva_{2010,5,2} = 478 means 478 articles in 2010 from country 5 mentioning “sovereign risk” or “government debt” together with bank No. 2.
  - Demeaned country-adjusted Factiva measure used:
    - SovereignRisk_{tji} = Factiva_{tji} - Factiva_{tj} (Equation 2).
  - Translations of terms used for non-English languages.
- Data:
  - Annual bank-level financial data cover 2006–11 from Bankscope.
  - Summary sample statistics (No. of observations = 957):
    - Change in deposits, million euros: Mean 1.84, Standard Deviation 15.53, Minimum -62.58, Maximum 87.39
    - Performance (DEA), a [0, 1] index: Mean 0.60, Standard Deviation 0.15, Minimum 0.28, Maximum 0.89
    - Inflation, percent: Mean 2.95, Standard Deviation 2.29, Minimum -1.09, Maximum 12.35
    - Sovereign risk (Factiva, demeaned) count: Mean 11.57, Standard Deviation 75.46, Minimum -71.00, Maximum 483.31
    - Market share, ratio: Mean 0.11, Standard Deviation 0.14, Minimum 0.0001, Maximum 0.65
    - Unemployment, percent: Mean 8.79, Standard Deviation 4.13, Minimum 3.20, Maximum 21.60

- Performance measure:
  - Data Envelopment Analysis (DEA) used to produce composite [0,1] bank efficiency/performance indices relative to peers.
  - DEA scores corrected for efficiency bias following Simar and Wilson (1998, 2000); Appendix II reports corrected scores across countries and time.

#### B. Baseline Regression Results

- Estimation approach:
  - Fixed effects regressions of Equation 1; table reported for 2006–11 and subsamples 2006–09 and 2010–11.
- Main findings:
  - Perceived sovereign risk (Factiva, demeaned) has a negative and significant impact on growth of deposits in almost all specifications—more media references to a bank’s exposure to sovereign risk prompt depositors to reduce deposits, ceteris paribus.
  - Bank efficiency positively affects consumer deposit growth—depositors reward more efficient banks with increased exposure.
  - Macroeconomic controls (contemporaneous inflation and unemployment) do not have a direct impact on bank deposit growth in baseline regressions.
  - Market share has the expected sign but is not statistically significant in DEA-based specification; no strong evidence of a “too big to fail” implicit guarantee effect on depositors amid heightened sovereign vulnerabilities in DEA-based regressions.
- Time-split comparisons:
  - 2006–09 (pre-crisis):
    - Performance was positive and significantly related to deposit growth.
    - Sovereign risk was not an important factor.
    - Inflation positively and significantly related to deposit growth.
  - 2010–11 (during crisis):
    - Impact of bank performance on deposits vanishes.
    - Perception of sovereign exposure/risk becomes an important negative predictor of deposit behavior.
    - Inflation no longer matters.
- Non-linearity and high-exposure effects:
  - Dummy75pc variable equals 1 if Sovereign Risk is in the upper quartile for a given country-year.
  - Interaction results indicate banks strongly exposed to sovereign-related news may face higher risk of losing deposits.
  - Before crisis: sovereign-related news associated with deposit accumulation across quantiles (potentially interpreted positively in tranquil times).
  - During crisis: effect disappears for first three quantiles and becomes negative for fourth quantile (coefficient equals -0.137, statistically significant).

- Selected regression coefficients from Table 3 (Fixed Effects, dependent variable: Change in Nominal Deposits):
  - Performance (DEA), 2006–11: 13.699*** (t-stat reported in brackets [3.052])
  - Sovereign Risk (Factiva), 2006–11: -0.031** ([-2.413])
  - Sovereign Risk (Factiva) x Dummy75pc, 2006–11: -0.153*** ([-3.355])
  - Inflation, 2006–09: 1.689*** ([2.800])
  - Constant, 2006–11: -8.021*** ([-2.830])
  - No. of observations: 957; Number of countries: 29.
  - R-squared values: examples—0.117 (column 1), 0.162 (column 3), 0.149 (column 4).
  - Significance notation: *** p<0.01, ** p<0.05, * p<0.1.

#### C. Robustness Tests

- Alternative performance measure:
  - Replaced DEA with Return on Equity (ROE); results largely consistent with baseline.
  - With ROE, market share became positive and statistically significant—suggesting presence of a “too large to fail” effect in ROE specifications.
- Alternative Factiva construction:
  - Reintroduced the country component in the Factiva variable (i.e., not demeaning per Equation 2); results did not materially change.
- Additional controls:
  - Used Bank for International Settlements data to control for countries’ banking sector exposure to various country groupings (United States, core Eurozone, peripheral Eurozone, emerging Europe, non-Eurozone Advanced Europe). (Further details not provided in the supplied excerpt.)

### Key empirical conclusions

- Perception of sovereign risk—measured via bank-level counts of media references to “sovereign risk” or “government debt” in articles mentioning the bank—negatively and significantly affects deposit growth, especially during crisis periods and for banks in the upper quartile of sovereign-related news exposure.
- Bank efficiency (DEA) is rewarded by depositors in normal times; this relationship weakens or disappears during crisis periods when sovereign concerns dominate depositor behavior.
- Macroeconomic contemporaneous indicators (inflation, unemployment) have limited direct explanatory power for deposit growth once sovereign perception and bank performance are controlled for, though inflation mattered pre-crisis.

### Policy-relevant analysis and implications (as emphasized in the text)

- Importance of assessing banking sector and sovereign vulnerabilities together to understand the full impact of large fiscal shocks and to design appropriate policy responses.
- Sovereign risk can affect bank funding through both asset- and liability-side channels; policy responses need to recognize and address both channels to prevent feedback loops that increase sovereign debt burdens.
- Perceived credibility of deposit insurance matters; during elevated sovereign risk deposit insurance may be viewed as less credible, reducing its effectiveness in stabilizing deposits.
- Public sector financial needs and contingent liabilities grew substantially in 2009–10; policy frameworks should incorporate broader balance sheet developments, debt portfolio structure, investor base, cross-border linkages, and financial assets when assessing sovereign risk.

*Italicized source attribution: Content extracted from _wp16145 - References (PDF), IMF.*

### Appendix III for the country classifications). The results (reported in final column of

### _wp16145 - Appendix III for the country classifications). The results (reported in final column of

### Robustness tests and key regression statistics (Table 4)
- Performance (DEA) coefficient: 13.555***  [3.038]
- Alternate Performance (ROE) coefficients: 0.102**  [2.326]; 0.055**  [2.392]; 0.085***  [3.473]
- Sovereign Risk (demeaned Factiva) coefficients: -0.031**  [-2.409]; 0.004  [0.0977]; -0.030***  [-2.845]; -0.027**  [-2.150]; -0.037***  [-2.825]
- Sovereign Risk (level of Factiva) coefficients shown: -0.003  [-0.080]; -0.033***  [-3.035]; -0.033***  [-2.593]
- Inflation coefficients (examples): -0.019  [-0.0591]; 1.568***  [2.694]; 0.288  [0.338]; -0.017  [-0.0565]; 1.688***  [2.802]; 0.438  [0.496]; 0.123  [0.383]; 2.878**  [2.048]
- Market Share coefficients (examples): -0.18  [-0.617]; 23.932**  [2.381]; 3.589  [0.372]; 16.129**  [2.214]; 15.122  [1.502]; 5.305  [0.549]; 11.822  [1.626]; 20.911  [1.614]
- Unemployment coefficients (examples): 11.501  [1.588]; 651.6  [1.309]; -26.57  [-0.0284]; -13.82  [-0.0456]; 0.457  [0.967]; -0.251  [-0.274]; -0.068  [-0.229]; -0.130  [-0.254]
- Country-group exposures (nominal home banking sector exposure to sovereigns):
  - United States: -2.08E+05  [-0.908]
  - Emerging Europe: -3.9e+05**  [-2.320]
  - Non-EZ Advanced Europe: 2.05E+05  [0.765]
  - Core EZ: -8.24E+03  [-0.0883]
  - EZ “Periphery”: 1.71E+05  [0.489]
- Model controls included: Yes (for listed specifications)
- Constant terms (examples): -7.084**  [-2.298]; -11.46***  [-2.621]; 1.850  [0.326]; -0.956  [-0.345]; -22.70***  [-4.860]; 3.268  [0.565]; -7.809**  [-2.521]; 32.100  [1.260]
- Observations: 957; 575; 382; 957; 575; 382; 957; 533 (as reported across columns)
- R-squared examples: 0.154; 0.156; 0.119; 0.162; 0.153; 0.121; 0.167
- Note: *** p<0.01, ** p<0.05, * p<0.1. With the exception of the last column (where number of countries is 14), the number of countries in all columns is 29.

### Main empirical findings and interpretation (Conclusion)
- Core result:
  - The exposure of the banking sector to sovereign risk negatively affected the growth of consumer deposits over and above the impact of macroeconomic conditions (sample: 29 European countries, 2006–11).
- Nonlinearity and timing:
  - Holdings of sovereign bonds benefited banks during tranquil times.
  - Banks that were in the news more often during crisis periods were penalized by depositors during the crisis.
  - During tranquil times depositors pay closer attention to banks’ financial performance; during crisis times perceived risks matter more than financial performance in depositors’ choices.
- Country-group ordering (depositors’ perception):
  - Coefficients on exposures to sovereigns in EZ “Periphery,” Emerging Europe, and United States are negative; only Emerging Europe coefficient is statistically significant, suggesting Emerging Europe may be perceived as riskiest while Core Eurozone and Non-Eurozone Advanced Europe seen as least risky.
- Too-big-to-fail:
  - With the exception of one specification, no evidence of “too big to fail” effects from depositors’ perspective.
- Macro-financial implications:
  - Excessive sovereign exposure may have contributed to deposit outflows in Europe after the crisis.
  - The negative link between sovereign risk and deposit collection can limit banks’ ability to lend and strengthen the procyclical role of banks in transmission of fiscal shocks.
  - This may call for more explicit inclusion of banks in macro models to capture full pass-through of shocks to the real economy and to improve macro projections.

### Policy implications and recommendations
- Measures to weaken the sovereign risk–deposit contraction link:
  - Fiscal tightening
  - Liability management exercises (aimed at reducing short-run gross borrowing needs)
  - Loss-sharing mechanisms
- Regulatory caution:
  - New regulations that could encourage banks to hold more government debt (e.g., for liquidity purposes) should be introduced with care and, if possible, their undesirable impacts mitigated.
- Strengthening resilience:
  - Improve bank efficiency (may boost depositor confidence and help stabilize deposits).
  - Build sufficient buffers:
    - For banks: more capital and liquidity.
    - For sovereigns: lower levels of debt and quasifiscal activity.
- Modeling recommendation:
  - Consider more explicit inclusion of banks in traditional macro models to account for bank-mediated pass-through of fiscal shocks.

### DEA methodology (Appendix I) — specification and inputs/outputs
- Method:
  - Data Envelopment Analysis (DEA): non-parametric, builds a frontier of most efficient banks to assess relative efficiency (Charnes, Cooper, and Rhodes (1978) approach; Banker, Charnes, and Cooper (1984) noted for variable returns to scale).
  - Efficiency score θ in [0, 1]; θ = 1 implies on frontier (efficient), θ < 1 implies inside frontier (inefficient).
- Inputs specified (following Grigorian and Manole (2006)):
  - Labor
  - Fixed assets
  - Interest expenditures
  - These capture: (1) personnel and management; (2) computer hardware and premises (branch network); (3) leveraged funds.
- Outputs defined:
  - Revenues (sum of interest and non-interest income)
  - Net loans (loans net of loss provisions)
  - Liquid assets (sum of cash, balances with monetary authorities, and holdings of treasury bills)
- Interpretation:
  - Holding outputs and two inputs constant, lesser use of the third input implies higher efficiency.
- Computational notes:
  - DEA linear programming primal and dual formulations reported (Equations (A-1) and (A-2) in the source).
  - Efficiency indexes can be decomposed into pure technical and scale efficiency indexes.

### Appendix II — Average DEA indices by country and year (selected values)
- Austria: 2006 0.769; 2007 0.601; 2008 0.516; 2009 0.578; 2010 0.608; 2011 0.596; Average 0.611
- Belgium: 2006 0.559; 2007 0.657; 2008 0.600; 2009 0.609; 2010 0.760; 2011 0.844; Average 0.672
- Bulgaria: 2006 0.566; 2007 0.613; 2008 0.567; 2009 0.593; 2010 0.628; 2011 0.645; Average 0.602
- Cyprus: 2006 0.443; 2007 0.481; 2008 0.466; 2009 0.626; 2010 0.598; 2011 0.517; Average 0.522
- Czech Republic: 2006 0.612; 2007 0.654; 2008 0.662; 2009 0.766; 2010 0.768; 2011 0.792; Average 0.709
- Finland: 2006 0.559; 2007 0.770; 2008 0.779; 2009 0.763; 2010 0.688; 2011 0.667; Average 0.704
- France: 2006 0.659; 2007 0.643; 2008 0.632; 2009 0.669; 2010 0.684; 2011 0.700; Average 0.665
- Germany: 2006 0.588; 2007 0.556; 2008 0.512; 2009 0.686; 2010 0.534; 2011 0.857; Average 0.622
- Greece: 2006 0.576; 2007 0.530; 2008 0.499; 2009 0.528; 2010 0.514; 2011 0.499; Average 0.524
- Ireland: 2006 0.643; 2007 0.660; 2008 0.696; 2009 0.677; 2010 0.684; 2011 0.702; Average 0.677
- Italy: 2007 0.557; 2008 0.584; 2009 0.502; 2010 0.629; 2011 0.657; Average 0.600
- Netherlands: 2006 0.465; 2007 0.568; 2008 0.561; 2009 0.612; 2010 0.602; 2011 0.599; Average 0.568
- Poland: 2006 0.610; 2007 0.597; 2008 0.539; 2009 0.586; 2010 0.638; 2011 0.638; Average 0.601
- Portugal: 2006 0.545; 2007 0.569; 2008 0.490; 2009 0.520; 2010 0.539; 2011 0.573; Average 0.540
- Spain: 2006 0.543; 2007 0.551; 2008 0.541; 2009 0.602; 2010 0.621; 2011 0.597; Average 0.576
- Sweden: 2006 0.848; 2007 0.827; 2008 0.850; 2009 0.807; 2010 0.792; 2011 0.698; Average 0.804
- Switzerland: 2007 0.672; 2009 0.710; 2010 0.555; 2011 0.777; 2011 0.796; Average 0.720
- United Kingdom: 2006 0.626; 2007 0.628; 2008 0.627; 2009 0.693; 2010 0.677; 2011 0.752; Average 0.667
- (Appendix II lists all 29 countries’ annual DEA indices for 2006–2011 and averages; selected country values shown above as reported.)

*Source: _wp16145 - Appendix III for the country classifications). The results (reported in final column of*

### Appendix III. Country Classifications

### Appendix III. Country Classifications

### EU 27
- Austria
- Belgium
- Bulgaria
- Cyprus
- Czech Republic
- Denmark
- Estonia
- Finland
- France
- Germany
- Greece
- Hungary
- Ireland
- Italy
- Latvia
- Lithuania
- Luxembourg
- Malta
- Netherlands
- Poland
- Portugal
- Romania
- Slovakia
- Slovenia
- Slovenia
- Spain
- Sweden
- United Kingdom

### EZ 17
- Austria
- Belgium
- Cyprus
- Czech Republic
- Estonia
- Finland
- France
- Germany
- Greece
- Ireland
- Italy
- Luxembourg
- Malta
- Netherlands
- Portugal
- Slovakia
- Slovenia
- Spain

### EZ “Periphery”
- Austria
- Belgium
- Cyprus
- Finland
- France
- Germany
- Greece
- Ireland
- Italy
- Luxembourg
- Netherlands
- Portugal
- Slovakia
- Slovenia
- Spain

### Core EZ
- Austria
- Belgium
- France
- Germany

### Emerging Europe
- Bulgaria
- Czech Republic
- Hungary
- Latvia
- Lithuania
- Poland
- Romania
- Slovakia
- Slovenia

### Non-EZ Advanced Europe
- Denmark
- Estonia
- Finland
- Ireland
- Latvia
- Lithuania
- Sweden
- United Kingdom
- Norway
- Switzerland

*Source: _wp16145 - Appendix III. Country Classifications*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2016/_wp16145.pdf_