## APPENDIX 1. THE EVOLUTION OF SENIOR DEBT-DEPOSIT SPREAD FOLLOWING THE INTRODUCTION OF DEPOSITOR PREFERENCE IN THE EUROPEAN UNION

## Source details

**Canonical URL:** [APPENDIX 1. THE EVOLUTION OF SENIOR DEBT-DEPOSIT SPREAD FOLLOWING THE INTRODUCTION OF DEPOSITOR PREFERENCE IN THE EUROPEAN UNION](https://www.imf.org/-/media/files/publications/tnm/2020/tnmea2020002.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/tnm/2020/tnmea2020002.pdf.md)
- [Structured JSON version](/-/media/files/publications/tnm/2020/tnmea2020002.pdf.json)

---

### Introduction and context
- BRRD introduced two-tiered depositor preference across the European Union via Article 108 of the BRRD:
  - Higher priority ranking among unsecured claims for deposits up to the coverage level under the Deposit Guarantee Scheme (DGS).
  - Second priority ranking for eligible deposits exceeding the coverage level held by natural persons, and micro- and SMEs.
- DGS Directive: DGS subrogated to the rights and obligations of covered depositors.
- Large corporate deposits typically rank below eligible deposits, pari passu with other general unsecured creditors.
- BRRD timeline:
  - Officially adopted May 2014; entered into force July 2014.
  - Deadline to implement BRRD provisions into national law as of January 1, 2015 (bail-in provisions by January 1, 2016).
- Country focus where depositor preference/DGS preference did not exist pre-BRRD: Germany, the Netherlands, Spain, and the United Kingdom.
- National implementation dates for full BRRD provisions:
  - Germany and the United Kingdom: January 2015.
  - Netherlands: November 2015.
  - Spain: January 2016.

### Data and measurement
- Sample period: January 2013 to January 2018 (country-specific start dates applied).
- Deposit rates:
  - Average system-wide monthly bank deposit rates from central banks.
  - Household deposit rates only; use the ‘two years or more’ deposit rates for comparability to two- and five-year senior bond maturities.
- CDS data:
  - Use month-end CDS spreads (basis points) for top five banks by total assets (as of January 2016) with outstanding two- and five-year senior bonds between 2012 and 2018.
  - CDS-deposit spread = weighted average CDS (top five banks) − deposit rate; weights = sum of total assets of top five banks.
- Country-specific CDS series start dates:
  - United Kingdom: January 2013 through January 2018.
  - Germany: January 2013 through January 2018.
  - Netherlands: April 2013 through January 2018 (weighted CDS series start in May 2013).
  - Spain: June 2014 through January 2018 (weighted CDS series start in June 2014).

### Methodology
- Monthly CDS-deposit spread series split into before and after based on month of BRRD transposition into national law.
- Hypothesis testing:
  - Two-sample t-test assuming equal variance.
  - Null hypothesis (H0): difference between before and after spread is greater than or equal to 0.
  - Alternative hypothesis (HA): difference between before and after spread is less than 0.
  - Rejection of H0 implies after spread is greater than before spread.
- Event-study analyses:
  - Mean-adjusted approach (expected value = mean of pre-BRRD values).
  - Market-adjusted approaches:
    - One-factor model: sovereign bond yield.
    - Two-factor model: sovereign bond yield + European VSTOXX.
  - Abnormal = observed − expected (fitted) values; statistical significance at the 5 percent level.
  - Proxied monetary policy indicator = sovereign bond yield for each country.

### High-level empirical findings
- CDS-deposit spread increased for all sample countries following national transposition of BRRD into law.
- The increase in CDS-deposit spread is primarily driven by declining deposit rates rather than rising CDS.
- Using two-year and five-year CDS, the null hypothesis is strongly rejected for all countries—implying mean spread in the after period is higher than in the before period.
- Two-year CDS results: change in CDS is small; slight increase in CDS for German banks.
- Five-year CDS results: change in CDS is negative for all countries.
- Weighted CDS increased slightly and returned to pre-BRRD levels toward end-2017.
- Deposit-to-liabilities ratio of banking sector showed no significant variation over the sample period—no evidence banks adjusted the ratio of deposit funding to total liabilities after depositor preference introduction.

### Detailed results — Table 1 (basis points)
- Panel A. Using two-year CDS
  - Germany:
    - MEAN OF WEIGHTED-CDS BEFORE: 47
    - MEAN OF WEIGHTED-CDS AFTER: 68
    - Δ (T-STAT): 21 (-2.9)
    - MEAN OF DEPOSIT RATE BEFORE: 205
    - MEAN OF DEPOSIT RATE AFTER: 156
    - Δ (T-STAT): -49 (13.3)
    - MEAN OF SPREAD BEFORE: -15
    - MEAN OF SPREAD AFTER: 8
    - Δ (T-STAT): 23 (-9)
  - United Kingdom:
    - MEAN OF WEIGHTED-CDS BEFORE: 46
    - MEAN OF WEIGHTED-CDS AFTER: 42
    - Δ (T-STAT): -4 (0.8)
    - MEAN OF DEPOSIT RATE BEFORE: 232
    - MEAN OF DEPOSIT RATE AFTER: 171
    - Δ (T-STAT): -60 (9)
    - MEAN OF SPREAD BEFORE: -18
    - MEAN OF SPREAD AFTER: 5
    - Δ (T-STAT): 23 (-7)
  - Netherlands:
    - MEAN OF WEIGHTED-CDS BEFORE: 41
    - MEAN OF WEIGHTED-CDS AFTER: 32
    - Δ (T-STAT): -9 (2.4)
    - MEAN OF DEPOSIT RATE BEFORE: 245
    - MEAN OF DEPOSIT RATE AFTER: 173
    - Δ (T-STAT): -72 (10.3)
    - MEAN OF SPREAD BEFORE: -20
    - MEAN OF SPREAD AFTER: -14
    - Δ (T-STAT): 6 (-11)
  - Spain:
    - MEAN OF WEIGHTED-CDS BEFORE: 58
    - MEAN OF WEIGHTED-CDS AFTER: 61
    - Δ (T-STAT): 3 (-0.5)
    - MEAN OF DEPOSIT RATE BEFORE: 69
    - MEAN OF DEPOSIT RATE AFTER: 13
    - Δ (T-STAT): -57 (9.1)
    - MEAN OF SPREAD BEFORE: -20
    - MEAN OF SPREAD AFTER: 22
    - Δ (T-STAT): 42 (-7)
- Panel B. Using five-year CDS
  - Germany:
    - MEAN OF WEIGHTED-CDS BEFORE: 108
    - MEAN OF WEIGHTED-CDS AFTER: 92
    - Δ (T-STAT): -16 (2)
    - MEAN OF DEPOSIT RATE BEFORE: 215
    - MEAN OF DEPOSIT RATE AFTER: 151
    - Δ (T-STAT): -65 (16.1)
    - MEAN OF SPREAD BEFORE: -92
    - MEAN OF SPREAD AFTER: -59
    - Δ (T-STAT): 33 (-5.1)
  - United Kingdom:
    - MEAN OF WEIGHTED-CDS BEFORE: 161
    - MEAN OF WEIGHTED-CDS AFTER: 112
    - Δ (T-STAT): -50 (6)
    - MEAN OF DEPOSIT RATE BEFORE: 264
    - MEAN OF DEPOSIT RATE AFTER: 172
    - Δ (T-STAT): -92 (10.3)
    - MEAN OF SPREAD BEFORE: -103
    - MEAN OF SPREAD AFTER: -60
    - Δ (T-STAT): 43 (-6.8)
  - Netherlands:
    - MEAN OF WEIGHTED-CDS BEFORE: 80
    - MEAN OF WEIGHTED-CDS AFTER: 54
    - Δ (T-STAT): -26 (4.8)
    - MEAN OF DEPOSIT RATE BEFORE: 245
    - MEAN OF DEPOSIT RATE AFTER: 166
    - Δ (T-STAT): -79 (12.1)
    - MEAN OF SPREAD BEFORE: -165
    - MEAN OF SPREAD AFTER: -112
    - Δ (T-STAT): 53 (-12.6)
  - Spain:
    - MEAN OF WEIGHTED-CDS BEFORE: 115
    - MEAN OF WEIGHTED-CDS AFTER: 112
    - Δ (T-STAT): -3 (0.3)
    - MEAN OF DEPOSIT RATE BEFORE: 69
    - MEAN OF DEPOSIT RATE AFTER: 12
    - Δ (T-STAT): -57 (10.5)
    - MEAN OF SPREAD BEFORE: 46
    - MEAN OF SPREAD AFTER: 95
    - Δ (T-STAT): 49 (-4)

- Note: Change (Δ) = difference between before and after mean weighted CDS; mean spread = weighted-CDS − over-two-years household deposit rate. Null hypothesis rejected when | t-statistic | < t-critical (less than 1.68), implying after spread greater than before spread.

### Event-study and robustness conclusions
- Mean-adjusted event-study results equivalent to differences-in-means tests; show increase in CDS-deposit spread post-transposition.
- Market-adjusted results:
  - One-factor (sovereign yield) and two-factor (sovereign yield + VSTOXX) models show abnormal spread tends to follow abnormal deposit rates; abnormal CDS generally statistically insignificant except for Germany.
  - Plotting negative of abnormal deposit rates aligns with abnormal spread movements.
- Conclusion: Increase in CDS-deposit spread driven by declining deposit rates, not by increasing CDS, after accounting for monetary policy and stock market volatility.
- Deposit-to-liabilities ratios for MFIs in sample countries show no significant variation during 2013–2017 (end-of-period, seasonally unadjusted, excl. Eurosystem).

### Data and methodology limitations
- Unavailability of bank-specific deposit rates and uninsured deposit volumes (e.g., > EUR 100,000) prevents analysis of changes in uninsured deposit volumes and bank-specific spreads.
- Cannot account for heterogeneity in substitutability between senior bonds and uninsured deposits across banks and countries.
- Controlled for country-level monetary policy and European stock market volatility, but cannot account for bank-level factors affecting deposit rates and bond prices, such as:
  - Asset quality
  - Returns
  - Volatility
  - Liquidity
  - Capital
  - Changes in credit ratings
- Other legislative changes before and after BRRD transposition are not controlled for.
- Analysis uses largest banks only; too-big-to-fail premia not accounted for.

*Source: APPENDIX 1. THE EVOLUTION OF SENIOR DEBT-DEPOSIT SPREAD FOLLOWING THE INTRODUCTION OF DEPOSITOR PREFERENCE IN THE EUROPEAN UNION (tnmea2020002).*

### Appendix 1. The Evolution of Senior Debt-Deposit Spread Following the

### Appendix 1. The Evolution of Senior Debt-Deposit Spread Following the

### I. Introduction
- Purpose:
  - Take stock of different types of depositor preference, offer a framework to assess their relative advantages and disadvantages, and make policy recommendations.
- Structure:
  - Section II: Context and principles for depositor preference.
  - Section III: Overview of different forms of depositor preference.
  - Section IV: Relative advantages and disadvantages.
  - Section V: Cross-jurisdictional analysis.
  - Final section: Conclusions.

### II. Rationale for depositor preference
- Key policy motivations:
  - Increase recoveries to depositors in bank failure by ranking deposit liabilities more senior than other senior unsecured claims.
  - Reduce costs of protecting deposits via deposit insurance payouts or government support in resolution, thereby enhancing the financial safety net.
- Legal and economic justification:
  - Deposits often constitute bank-created money: bank acquires ownership title to the deposited money, can use the money, may pay little or no interest (e.g., demand deposits), and must fulfill repayment on demand or at maturity.
  - Retail depositors typically lack the capacity to assess bank risk or secure collateral; depositor preference and deposit insurance obviate this need.
- Public policy objectives linked to depositor preference:
  - Facilitate effective resolution by aligning loss allocation with insolvency ranking and preserving the "no creditor worse off than in liquidation" (NCWOL) safeguard.
  - Protect payment systems and economic activity by facilitating access to deposits during resolution and liquidation.
  - Reduce the cost to Deposit Insurance Schemes (DIS) or taxpayers when deposits must be protected in liquidation or resolution.

### III. General principles for creditor hierarchies (Box 1)
- Equality:
  - Creditors should be treated pari passu unless fundamental legal/economic differences or compelling public interest justify differentiation.
  - Excessive creditor classes can undermine efficiency of insolvency/resolution.
- Necessity:
  - Depositor ranking should best serve public policy objectives including enhancing financial stability and reducing resolution/DIS costs.
- Transparency:
  - Creditor hierarchy should be clear and transparent in bank liquidation or resolution law; modify general insolvency law with bank-specific provisions if needed.
- Consistency:
  - Order of loss allocation in resolution versus liquidation should not differ materially to avoid legal challenges and compensation claims.
  - IADI’s 2014 revisions recommend resolution procedures follow a defined creditor hierarchy protecting insured deposits from sharing losses.

### IV. Litigation and compensation risk illustrated (Figure 1 example)
- Balance sheet and scenario data (as presented):
  - Panel A: Balance Sheet of Failed Bank
    - $8 equity
    - $2 subdebt
    - $50 other senior creditors
    - $40 small retail deposits
    - $30 loses on ‘bad’ assets
    - $70 remaining value of assets
  - Panel B: Recoveries in liquidation (percentages shown in figure)
    - 0%
    - 100%
    - 60%
    - 100%
    - 78%
    - 0%
    - 0%
    - 0%
    - 78%
    - 64%
  - Panel C (P&A) and Panel D (Bail-in) outcomes:
    - P&A: purchaser accepts US$2 more of liabilities than assets (premium for deposits). Wholesale creditors receive US$32 instead of US$39 after P&A (an extra loss of US$7 compared to liquidation).
    - Bail-in: equity and subordinated debt written down to zero; portion of wholesale creditors written down or converted to equity without imposing losses on depositors. Wholesale creditors incur an extra loss of US$9 compared to liquidation.
  - Illustrative point:
    - Protecting retail deposits in resolution while deposits rank pari passu with other senior creditors in insolvency can make other senior creditors worse off than in liquidation, creating grounds for litigation and compensation claims.

### V. Forms of depositor preference (Box 2)
- Insured depositor preference:
  - Preferential treatment for insured deposits (eligible and within deposit insurance limit) over other ordinary senior unsecured creditors.
  - DIS normally subrogated to insured depositor claims to the extent of deposit insurance payouts.
  - In absence of a DIS, preference can be a single-tier priority up to a certain limit; deposits exceeding this limit treated as unsecured senior claims.
  - Examples: Switzerland, Vietnam.
- Tiered depositor preference (two tiers):
  - Preference for insured deposits (and the DIS by subrogation) over uninsured eligible deposits; both preferred over other senior unsecured creditors.
  - In jurisdictions without a DIS, priority can be created by deposit-size thresholds.
  - Examples: EU, Georgia, Mexico.
- General depositor preference:
  - All deposit liabilities of a deposit-taking institution are preferred over other senior unsecured creditors, regardless of DIS eligibility or coverage.
  - DIS is subrogated for insured deposits and bears losses to same degree as all deposits.

### VI. Recoveries to DIS and role of asset encumbrance
- Recoveries by the DIS depend on:
  - Degree of depositor preference (insured, tiered, or general).
  - Asset encumbrance: secured creditors must be satisfied from proceeds of encumbered assets, reducing asset pool available to satisfy lower-ranking creditors.
- Illustrative findings:
  - Insured and tiered depositor preference provide the most protection to the DIS or preferred depositors.
  - Tiered preference also provides some protection to uninsured eligible deposits (retail deposits above the deposit insurance limit) at the expense of other senior unsecured creditors.
  - Recoveries to the DIS decline significantly with higher asset encumbrance, which typically increases in stress when banks rely more on secured funding.

*Technical Notes and Manuals 2020/002 — Appendix 1.*

### BOX 3. Creditor Recoveries Under Different Forms of Depositor Preference

### BOX 3. Creditor Recoveries Under Different Forms of Depositor Preference

### Illustrative balance sheets and recovery tables
- The tables show two balance sheets, Bank A and Bank B, entering liquidation after losses have eroded bank capital to zero, with remaining creditor claims totaling US$100 and recoveries from the liquidation of assets of only US$80 after losses in liquidation of US$20.
- Raw recovery lines as presented in the source:
  - BANK A LIABILABILITIES RECOVERIES (AS % OF CLAIM) WITHOUT INSURED TIERED GENERAL
  - (a) Secured creditors20100100100100
  - (b) Insured deposits (DIS subrogated)5075100100100
  - (c) Uninsured deposits107533100100
  - (d) Other senior unsecured creditors20753300
  - Total creditor claims100
  - BANK B LIABILABILITIES RECOVERIES (AS % OF CLAIM) WITHOUT INSURED TIERED GENERAL
  - (a) Secured creditors50100100100100
  - (b) Insured deposits (DIS subrogated)306010010075
  - (c) Uninsured deposits10600075
  - (d) Other senior unsecured creditors1060000
  - Total creditor claims100
- Bank B is characterized as having higher asset encumbrance (i.e., higher secured liabilities) and lower senior unsecured liabilities relative to Bank A.

### DIS recovery simulation (scenario and key parameters)
- Loss distribution modeled as normal with:
  - mean loss of US$50
  - standard deviation of US$15
  - 1,000 simulations
- Presentation of simulated results:
  - Boxes show the lower quartile (25 percent) and upper quartile (75 percent) of each distribution.
  - Lines (whiskers) show the minimum and maximum recoveries.
  - Data points show the median recovery rates for the DIS.
- Key descriptive findings from the simulation:
  - The results are identical for insured and tiered depositor preference because insured depositors (and the DIS subrogated) are ranked alone in second place in both hierarchies, with differences arising further down the hierarchy.
  - Insured and tiered depositor preference provide the most protection to the DIS.
  - The tiered form also provides some protection to uninsured eligible deposits (i.e., deposits above the deposit insurance limit) to the detriment of other senior unsecured creditors (e.g., bond holders).
  - Comparing Banks A and B shows that recoveries to the DIS decline significantly as asset encumbrance increases.

- Representative percentile and point values extracted from charts (as shown in the source):
  - 0.0%, 85.7%, 13.6%, 28.1%, 42.6%
  - 0.0%, 100.0%, 19.1%, 39.3%, 59.6%
  - 0.0%, 100.0%, 15.9%, 32.8%, 49.7%
  - 0%, 20%, 40%, 60%, 80%, 100% (axis ticks shown)
  - Bank A: Distributions of DIS recoveries; Bank B: Distributions of DIS recoveries
  - 0.0%, 87.5%, 24.4%, 37.1%, 49.8%
  - 0.0%, 100.0%, 39.1%, 59.3%, 79.6%
  - 0.0%, 100.0%, 39.1%, 59.3%, 79.6%
  - 0.0%, 100.0%, 32.5%, 49.4%, 66.4%
  - 0.0%, 100.0%, 19.1%, 39.3%, 59.6%

### Pros and cons of depositor preference (summary of substantive points)
- Advantages (varied by type of depositor preference):
  - Protecting financial stability with reduced costs: may facilitate prompt resolution, protect the payment system, and reduce overall possible costs to the DIS and the public, including NCWOL compensation risks.
  - Maximizing market discipline: by increasing potential loss exposure for other unsecured creditors, depositor preference provides incentives for closer monitoring and pricing of bank risk by financial investors.
    - For depositor preference to affect creditor incentives, sufficient unsecured debt must remain and the risk that nonpreferred bank creditors incur losses must be credible.
  - Cross-border cooperation: harmonization of creditor hierarchies through depositor preference (non-discriminatory by nationality or location) would facilitate cooperative resolution of cross-border banks.
- Potential adverse impacts and behavioral responses:
  - Impact on bank wholesale funding: introducing depositor preference would, ceteris paribus, increase the risk and potential cost of accessing unsecured wholesale funding.
  - Increased reliance on secured funding and asset encumbrance: creditors may secure claims (e.g., covered bonds), increasing encumbrance, limiting resolution options, and potentially raising costs for unsecured creditors.
  - Regulatory arbitrage: banks and creditors may restructure liabilities or exploit exemptions (very short-term liabilities or derivative claims) to attenuate loss risk.
- Evidence and context dependence:
  - Hardy (2013) concludes introducing depositor preference in the US had “little “systemic effect” on overall bank funding costs.
  - Danisewicz et al. (2018) find staggered introduction across US states led to a decline in uninsured deposit rates and an increase in nondeposit rates, widening the spread and improving bank soundness.
  - Staff analysis of tiered depositor preference in the EU: spread between bank bonds and deposit rates widened by between 40 and 70 basis points for four EU member-countries studied; change driven primarily by a decline in deposit rates rather than higher bond rates.
  - Findings are sensitive to contemporaneous events (e.g., strengthening of capital and liquidity supervision, quantitative easing); results cannot be generalized.
- Potential policy responses to unintended effects:
  - Obligatory minima for unsecured debt over a certain maturity that could be bailed-in in resolution.
  - Limits on the encumbrance of bank balance sheets.
  - Authorities should weigh trade-offs between intermediation costs and financial stability benefits.

### Country experience and lessons
- Most FSB member countries have established depositor preference, though not universal.
  - Examples: Argentina, Australia, China, India, Indonesia, Mexico, Russia, Turkey, the United States, and the EU member states; also Hong Kong SAR, Singapore, and Switzerland.
  - Canada, Japan, and South Korea do not have depositor preference.
  - Reforms underway in some countries (e.g., South Africa).
- Deposit insurance and depositor preference do not have a clear one-to-one relationship:
  - Some countries with a DIS do not have depositor preference (e.g., Japan and South Korea).
  - Some jurisdictions have depositor preference without a DIS (e.g., Panama and Zambia).
- Coordination issues and special cases:
  - Jurisdictions that establish preference for DIS claims should also establish it for insured deposits; DIS subrogation to insured deposit claims is advisable where the DIS funds payouts in resolution.
  - Some jurisdictions set ceilings for depositor preference differently from deposit insurance coverage (examples noted: Argentina and India; also approach followed in Afghanistan, Mexico, and Moldova).
  - Related-party deposits are typically excluded from insurance and may be subordinated; IADI Core Principle 8 allows such exclusions if clearly specified and not slowing reimbursements.

### Staff recommendations and policy guidance
- Adoption of depositor preference should be informed by assessment of potential impact in the context of the legal, judicial, and financial system structure.
- Timing considerations:
  - In the short term, introducing depositor preference during funding stress could be destabilizing; may be advisable to delay introduction until a more stable period.
- Preferred forms:
  - Staff normally recommends either tiered or general depositor preference.
  - Insured and tiered depositor preferences offer the most protection to the DIS or the state.
  - Insured depositor preference may be insufficient where deposit insurance coverage is low or SME deposits are ineligible.
  - General depositor preference offers advantages where all deposits—including those ineligible for deposit insurance coverage—may need protection and would facilitate greater DIS contribution to resolution.
- Complementarity with other measures:
  - Depositor preference is a complement to, not a substitute for, deposit insurance.
  - Recommended as part of a package including a DIS (when conditions are right), effective resolution mechanisms, and measures ensuring continuity of critical functions and quick reimbursement of insured deposits.
- Operational recommendations:
  - National depositor preference should be avoided to prevent discrimination and impediments to cross-border coordination.
  - In the absence of a DIS, advance payment rules for liquidators to make limited withdrawals can be critical to prevent loss of depositor confidence (example cited: Panama).

*Source: tnmea2020002 - BOX 3. Creditor Recoveries Under Different Forms of Depositor Preference*

### APPENDIX 1. THE EVOLUTION OF SENIOR DEBT-DEPOSIT

### APPENDIX 1. THE EVOLUTION OF SENIOR DEBT-DEPOSIT SPREAD FOLLOWING THE INTRODUCTION OF DEPOSITOR PREFERENCE IN THE EUROPEAN UNION

### Introduction
- The BRRD introduced two-tiered depositor preference across the European Union via Article 108 of the BRRD:
  - Higher priority ranking among unsecured claims for deposits up to the coverage level under the Deposit Guarantee Scheme (DGS).
  - Second priority ranking for eligible deposits exceeding the coverage level held by natural persons, and micro- and SMEs.
- The DGS Directive provides for the DGS being subrogated to the rights and obligations of covered depositors.
- Large corporate deposits rank below eligible deposits, typically pari passu with other claims of general unsecured creditors (such as senior bondholders).
- By changing the insolvency creditor hierarchy, depositor preference may alter relative cost of senior unsecured debt vis-à-vis bank deposits of individuals and SMEs—potentially increasing severity of loss for senior unsecured debt and lowering severity for preferred deposits.
- The BRRD was officially adopted in May 2014 and entered into force in July 2014, with a deadline to implement BRRD provisions into national law as of January 1, 2015 (bail-in provisions by January 1, 2016).
- Analysis focuses on four countries where depositor preference and DGS preference did not exist pre-BRRD: Germany, the Netherlands, Spain, and the United Kingdom.
- National legislation implementing full BRRD provisions were passed:
  - Germany and the United Kingdom: January 2015.
  - Netherlands: November 2015.
  - Spain: January 2016.

### Data
- Sample period: January 2013 to January 2018 (country-specific constraints applied; see notes).
- Deposit rates:
  - Use average system-wide monthly bank deposit rates published by central banks.
  - Only household deposit rates considered (deposits of large corporations were not preferred).
  - Use the ‘two years or more’ deposit rates for comparability to the two- and five-year senior bond maturities.
  - Deposit rates generally followed ECB policy rates; ECB deposit facility rate brought to zero in July 2012 and became negative in June 2014. Bank deposit rates did not become negative.
- Credit default swap (CDS) data:
  - Use CDS spreads instead of bond rates due to continuous measure and liquidity advantages.
  - Calculate CDS swap rates for the top five banks by total assets (as of January 2016) in each country with outstanding two- and five-year senior bonds between 2012 and 2018.
  - Month-end values of senior bond CDS spreads (reported in basis points) used to compute spreads between CDS and deposit rates.
- CDS-deposit spread defined as: weighted average CDS of top five banks minus the deposit rate for each country. Weights = sum of total assets of top five banks in each country.
- Country-specific CDS series start dates (due to data availability):
  - United Kingdom: January 2013 through January 2018.
  - Germany: January 2013 through January 2018.
  - Netherlands: April 2013 through January 2018 (weighted CDS series start in May 2013).
  - Spain: June 2014 through January 2018 (weighted CDS series start in June 2014).

### Methodology
- Monthly CDS-deposit spread series (January 2013 to January 2018) split into before and after periods based on month of BRRD transposition into national law.
- Start date for each country chosen to include at least two years before BRRD legislation deadline; end date chosen to exclude months after BRRD Amendment.
- Hypothesis testing:
  - Test for difference in means of spread before vs. after using t-test assuming equal variance.
  - Null hypothesis (H0): difference between the before spread and the after spread is greater than or equal to 0.
  - Alternative hypothesis (HA): difference between the before spread and the after spread is less than 0.
  - Spread = weighted two-year (or five-year) senior bond CDS of top five banks minus the two years or more deposit rate for households.
  - Rejection of H0 implies after spread is greater than before spread.
- Event-study analyses:
  - Compute abnormal spread, abnormal CDS, and abnormal deposit rate.
  - Mean-adjusted approach: expected value = mean of pre-BRRD values (equivalent to testing differences in means).
  - Market-adjusted approach: one-factor model (sovereign bond yield) and two-factor model (sovereign bond yield + European VSTOXX).
  - Abnormal = observed − expected (fitted) values; statistical significance at the 5 percent level.
  - Proxied monetary policy indicator = sovereign bond yield for each country.

### Results — High-level findings
- The CDS-deposit spread increased for all countries in the sample following national transposition of BRRD into law.
- The increase in CDS-deposit spread is primarily driven by declining deposit rates rather than rising CDS.
- Using two-year and five-year CDS, the null hypothesis is strongly rejected for all countries—implying mean spread in the after period is higher than in the before period.
- Two-year CDS results: change in CDS is small; slight increase in CDS for German banks.
- Five-year CDS results: change in CDS is negative for all countries.
- Most of the change in spread can be attributed to declining deposit rates; weighted CDS increased slightly and returned to pre-BRRD levels toward end-2017.
- Deposit-to-liabilities ratio of banking sector did not exhibit significant variation over sample period—no evidence banks adjusted ratio of deposit funding to total liabilities after depositor preference introduction.

### Results — Table 1 (Testing for difference in spread before and after BRRD legislation; all values in basis points)
- Panel A. Using two-year CDS
  - Germany:
    - MEAN OF WEIGHTED-CDS BEFORE: 47
    - MEAN OF WEIGHTED-CDS AFTER: 68
    - Δ (T-STAT): 21 (-2.9)
    - MEAN OF DEPOSIT RATE BEFORE: 205
    - MEAN OF DEPOSIT RATE AFTER: 156
    - Δ (T-STAT): -49 (13.3)
    - MEAN OF SPREAD BEFORE: -15
    - MEAN OF SPREAD AFTER: 8
    - Δ (T-STAT): 23 (-9)
  - United Kingdom:
    - MEAN OF WEIGHTED-CDS BEFORE: 46
    - MEAN OF WEIGHTED-CDS AFTER: 42
    - Δ (T-STAT): -4 (0.8)
    - MEAN OF DEPOSIT RATE BEFORE: 232
    - MEAN OF DEPOSIT RATE AFTER: 171
    - Δ (T-STAT): -60 (9)
    - MEAN OF SPREAD BEFORE: -18
    - MEAN OF SPREAD AFTER: 5
    - Δ (T-STAT): 23 (-7)
  - Netherlands:
    - MEAN OF WEIGHTED-CDS BEFORE: 41
    - MEAN OF WEIGHTED-CDS AFTER: 32
    - Δ (T-STAT): -9 (2.4)
    - MEAN OF DEPOSIT RATE BEFORE: 245
    - MEAN OF DEPOSIT RATE AFTER: 173
    - Δ (T-STAT): -72 (10.3)
    - MEAN OF SPREAD BEFORE: -20
    - MEAN OF SPREAD AFTER: -14
    - Δ (T-STAT): 6 (-11)
  - Spain:
    - MEAN OF WEIGHTED-CDS BEFORE: 58
    - MEAN OF WEIGHTED-CDS AFTER: 61
    - Δ (T-STAT): 3 (-0.5)
    - MEAN OF DEPOSIT RATE BEFORE: 69
    - MEAN OF DEPOSIT RATE AFTER: 13
    - Δ (T-STAT): -57 (9.1)
    - MEAN OF SPREAD BEFORE: -20
    - MEAN OF SPREAD AFTER: 22
    - Δ (T-STAT): 42 (-7)
- Panel B. Using five-year CDS
  - Germany:
    - MEAN OF WEIGHTED-CDS BEFORE: 108
    - MEAN OF WEIGHTED-CDS AFTER: 92
    - Δ (T-STAT): -16 (2)
    - MEAN OF DEPOSIT RATE BEFORE: 215
    - MEAN OF DEPOSIT RATE AFTER: 151
    - Δ (T-STAT): -65 (16.1)
    - MEAN OF SPREAD BEFORE: -92
    - MEAN OF SPREAD AFTER: -59
    - Δ (T-STAT): 33 (-5.1)
  - United Kingdom:
    - MEAN OF WEIGHTED-CDS BEFORE: 161
    - MEAN OF WEIGHTED-CDS AFTER: 112
    - Δ (T-STAT): -50 (6)
    - MEAN OF DEPOSIT RATE BEFORE: 264
    - MEAN OF DEPOSIT RATE AFTER: 172
    - Δ (T-STAT): -92 (10.3)
    - MEAN OF SPREAD BEFORE: -103
    - MEAN OF SPREAD AFTER: -60
    - Δ (T-STAT): 43 (-6.8)
  - Netherlands:
    - MEAN OF WEIGHTED-CDS BEFORE: 80
    - MEAN OF WEIGHTED-CDS AFTER: 54
    - Δ (T-STAT): -26 (4.8)
    - MEAN OF DEPOSIT RATE BEFORE: 245
    - MEAN OF DEPOSIT RATE AFTER: 166
    - Δ (T-STAT): -79 (12.1)
    - MEAN OF SPREAD BEFORE: -165
    - MEAN OF SPREAD AFTER: -112
    - Δ (T-STAT): 53 (-12.6)
  - Spain:
    - MEAN OF WEIGHTED-CDS BEFORE: 115
    - MEAN OF WEIGHTED-CDS AFTER: 112
    - Δ (T-STAT): -3 (0.3)
    - MEAN OF DEPOSIT RATE BEFORE: 69
    - MEAN OF DEPOSIT RATE AFTER: 12
    - Δ (T-STAT): -57 (10.5)
    - MEAN OF SPREAD BEFORE: 46
    - MEAN OF SPREAD AFTER: 95
    - Δ (T-STAT): 49 (-4)

Notes from Table 1:
- Change (Δ) equals difference between before and after mean weighted CDS.
- Mean spread = weighted-CDS measure (two-year or five-year) minus over-two-years deposit rate for households.
- Weighted CDS = weighted average of two-year (or five-year) senior bond CDS of top five banks for each country.
- Null hypothesis rejected when | t-statistic | < t-critical (less than 1.68), implying after spread greater than before spread.

### Event-study and robustness findings
- Mean-adjusted event-study: results equivalent to testing differences in means; results align with Figure 4 (increase in CDS-deposit spread post-transposition).
- Market-adjusted approach:
  - One-factor model: sovereign bond yield for each country.
  - Two-factor model: sovereign bond yield + VSTOXX (European stock market volatility index).
  - Except for Germany, abnormal spread tends to follow abnormal deposit rates; abnormal CDS statistically insignificant.
  - Plotting negative of abnormal deposit rates shows alignment with abnormal spread movements.
- Conclusion from event-study: evidence of increase in CDS-deposit spread driven by declining deposit rates, not by increasing CDS, after accounting for monetary policy and stock market volatility.
- Deposit-to-liabilities ratios for MFIs in sample countries show no significant variation during 2013–2017 (end-of-period, seasonally unadjusted, excl. Eurosystem).

### Data and Methodology Limitations
- Unavailability of bank-specific deposit rates and data on uninsured deposits is a significant limitation:
  - Cannot study changes in volume of uninsured deposits (greater than EUR 100,000).
  - Cannot compute bank-specific spread.
- Dependence on wholesale vs. deposits varies across banks and countries; cannot account for substitutability differences between senior bonds and uninsured deposits.
- Controlled for country-level monetary policy changes and European stock market volatility, but cannot account for bank-level risk factors affecting deposit rates and bond prices, such as:
  - Asset quality
  - Returns
  - Volatility
  - Liquidity
  - Capital
  - Changes in credit ratings
- Other legislative changes before and after BRRD transposition are not accounted for.
- Use of largest banks only; too-big-to-fail premia not accounted for.

*Source: APPENDIX 1. THE EVOLUTION OF SENIOR DEBT-DEPOSIT SPREAD FOLLOWING THE INTRODUCTION OF DEPOSITOR PREFERENCE IN THE EUROPEAN UNION (tnmea2020002).*

---


_Source: https://www.imf.org/-/media/files/publications/tnm/2020/tnmea2020002.pdf_