A note on the adequacy of the EU scheme for bank recovery, resolution and deposit insurance in Spain

Transcription

1 A note on the adequacy of the EU scheme for bank recovery, resolution and deposit insurance in Spain Pilar Gómez-Fernández-Aguado is a Senior Lecturer at the Department of Financial Economics and Accounting, University of Jaén (Spain). Her research interests are focused on banking and finance, with a particular emphasis on the analysis of deposit insurance schemes. She has acted as a consultant to several banking institutions in Spain. She has also published many academic research articles. Antonio Partal-Ureña is a Professor of Banking and Finance at the Department of Financial Economics and Accounting, University of Jaén (Spain). His research interests are focused on the study of the role of credit cooperatives in the banking system. He has acted as a consultant to several banking institutions and he has also published many academic research papers. Antonio Trujillo-Ponce is currently the Director of the Banking and Entrepeneurial Finance Research Group (BANEF) at the Pablo de Olavide University of Seville (Spain). He is a specialist in the area of banking and finance regulation, with a particular focus on Basel Capital Accords. He has published over 20 academic research articles in various journals, such as the Journal of Banking and Finance, Accounting and Finance, Journal of Economic Policy Reform and Applied Economics. Correspondence: Antonio Trujillo-Ponce, Department of Financial Economics and Accounting, Universidad Pablo de Olavide, Ctra. de Utrera, Km. 1, ES Seville, Spain. atrujillo@upo.es ABSTRACT This paper analyzes whether the funds set by the recent EU directives on bank resolution and deposit insurance to create a safer and sounder financial sector (i.e., 1% and 0.8% of covered deposits, respectively) are adequate to cover unexpected losses for the Spanish banking system. By applying a framework based on the foundation internal ratings-based (FIRB) approach established in Basel Capital Accords, we find that the fixed target levels proposed by the EU bank rules would not ensure the highest credit quality for a resolution and deposit insurance Fund (BRDIF) in Spain. Nevertheless, these resources would be sufficient to ensure a good solvency level for the Fund, equivalent to an AA S&P rating in Keywords: deposit insurance system; resolution funds; capital adequacy; internal ratings-based approach JEL classification: G21; G22; G28. INTRODUCTION As the financial crisis progressed and turned into the Eurozone debt crisis in , it became clear that greater interdependency for those countries sharing the euro required a deeper integration of the banking system. Consequently, in 2012, EU authorities agreed to the creation of an integrated financial framework (the so-called banking union) to restore confidence in banks and the euro. The banking union relies on common rules that all financial institutions in the EU must comply with. These rules include the establishment of 1 Electronic copy available at:

2 more adequate capital requirements (the EU Capital Requirements Directive was approved in 2013), better protection for all EU depositors (the EU Directive on Deposit Guarantee Schemes was adopted in 2014), and common tools to effectively address failing credit institutions (the EU Directive on Bank Recovery and Resolution was published in 2014 and entered into force as of 1 January 2015). The Bank Recovery and Resolution Directive (BRRD) sets out the rules for the resolution of banks and large investment firms in all EU Member States (MS). 1 It was adopted to provide authorities with comprehensive and effective arrangements to deal with failing banks at national level, as well as cooperation arrangements to tackle cross-border banking failures. An effective resolution regime should minimize the costs of the resolution of a failing institution assumed by the taxpayers as well as it should ensure that systemic institutions can be resolved without jeopardizing financial stability. The resolution tools referred to in BRRD include the following: (a) the sale of business; (b) the bridge institution; (c) the asset separation; and (d) the bail-in tool. 2 The bail-in tool gives shareholders and creditors of institutions a stronger incentive to monitor the health of an institution as they may suffer losses and bear an appropriate part of the costs arising from the failure of the bank. To ensure the effective application of the resolution tools and powers, MS shall constitute national resolution funds. The contribution of each institution to the national resolution fund must be adjusted in proportion to its risk profile, in accordance with the criteria specified in article Moreover, the banking union tries to ensure an orderly resolution of failing banks with minimal costs to taxpayers and real economy by applying a Single Resolution Mechanism (SRM) to banks covered by the Single Supervisory Mechanism (SSM), which places the European Central Bank (ECB) as the central prudential supervisor of financial institutions in the euro area. While under the BRRD the target level of the resolution funds is set at national level, under the SRM the target level of the Single Resolution Fund (SRF) is the sum of the covered deposits of all institutions of MS participating in the banking union. However, to mitigate any abrupt increase in fees for banks in some MS when switching from a national to a European target level, the implementing regulation provides for an eight-year mutualization phase (from 2016 to 2023) during which national compartments in the SRF will be gradually merged. Therefore, while in the first year 60% of banks' contributions will still be calculated in accordance with national target levels, this share will decrease annually. By 2023, all banks' contributions will be calculated on the basis of the SRM target level. As the protection of depositors is one of the most important objectives of resolution, deposits covered by deposit guarantee schemes (DGS) should not bear any losses in the resolution process. Therefore, those deposits that are protected under Directive 2014/49/EU 2 Electronic copy available at:

3 should not be subject to the exercise of the bail-in tool. 4 The DGS should, however, contribute to funding the resolution process by absorbing losses to the extent of the net losses that it would have had to suffer after compensating depositors in normal insolvency proceedings. As the DGS fund and the resolution fund are not necessarily the same fund, it means that a MS may want to change its funding and financing arrangements to create a unique fund making it easier to raise and manage resources. 5 Funds for bank resolution and deposit insurance should reach (after a transition period) a target level of at least 1% and 0.8%, respectively, of the amount of covered deposits of all the institutions authorized in their territory. Therefore, the EU has decided that resources needed for a bank resolution and deposit insurance Fund (BRDIF) will depend only on a fixed percentage of members exposure, regardless of the risk of the Fund s portfolio, which should consider parameters such as the probability of default of its bank members or the estimated loss in the event of a member s default. By ignoring these parameters the EU does not contemplate that the solvency of the Fund may diverge between countries, which may generate problems in building a single EU Fund. Similarly, ignoring the evolution of those risk factors over time may cause, during recessions, the Fund s resources to be insufficient to ensure an adequate solvency, leading to possible bailouts. Previously, a growing body of literature has examined the adequacy of deposit insurance schemes (DIS). Episcopos attempts to link risk-based premiums with guarantee fund reserves in a partial equilibrium setting, by employing a methodology based on options with credit risk. 6 Kuritzkes et al. argue that the risk management problem faced by the Federal Deposit Insurance Corporation (FDIC) is similar to that of a bank managing a loan portfolio, only in the FDIC s case, the risk arises from the potential for loss of the individual banks in its portfolio. 7 De Lisa et al. propose an estimation of the loss distribution of a DIS based on the Basel II regulatory framework. 8 The application of their model to 2007 data for a sample of Italian banks shows that the target size of the Italian DIS covers up to 98.96% of its potential losses. In Cariboni et al. the DIS is treated as a portfolio of banks whose default probabilities are estimated from credit default swap (CDS) spreads and losses are simulated using the Gaussian one-factor model. 9 Ho et al. examine the effectiveness of financial reforms carried out in Taiwan recently and measures the adequacy of the DIS. 10 The authors have improved on the methodology of Episcopos and report estimates of the cost of deposit insurance and implied reserves for each bank or financial holding company. They conclude that the fixed target ratio for the Taiwanese DIS may not be appropriate. Finally, Lee et al., based on the Merton put option framework, develop a deposit insurance pricing model that incorporates asset correlations, a measurement for the systematic risk of a bank, to account for the risk of joint bank failures. 11 3

4 To the best of our knowledge, we are the first to analyze whether the fixed target levels set by the recent EU directives on bank resolution and deposit insurance (i.e., 1% and 0.8% of covered deposits, respectively) are adequate to cover unexpected losses for the Spanish banking system. To do so, we estimate capital requirements needed for a BRDIF in Spain by applying a framework based on the foundation internal ratings-based (FIRB) approach established in Basel Capital Accords. We also contemplate some stress scenarios. SAMPLE AND METHODOLOGY Our sample consists of 45 Spanish banking groups (18 commercial banks, 13 savings banks and 14 credit cooperatives) with information available on all of the variables considered in The sample represents about 100% of the deposits that in theory should be covered by the Spanish BRDIF. By applying the IRB approach, we need to estimate the following variables that determine the risk of the BRDIF: - Exposure at default (EAD i ), which is determined by the amount of deposits covered by the member institutions. This variable is not directly observable; thus, we estimate the EAD by multiplying the amount of deposits of each credit institution by the percentage of deposits theoretically covered by the BRDIF (i.e., 66.57% in 2013) Severity or loss caused by the intervention of BRDIF (LGD). We initially consider the normalized value for the FIRB approach (i.e., 45%). - Probability of intervention of BRDIF, which is linked to the probability of default of the bank members (PD i ). We follow to De Lisa et al. by modeling the PD of a bank as the probability that its asset portfolio losses (L) exceed the sum of the expected losses (EL) and the total actual capital (CAP) (given by the sum of regulatory capital requirements and any excess capital). To calculate this PD, we first estimate a proxy of the average default probability of the asset portfolio for bank i (PD! ) by utilizing the capital requirement (K) formula for a single exposure in the Basel IRB approach. 13 As we know K, we can obtain PD! by solving the following equation: K = LGD N 1 1 R N!! PD! + R 1 R N!! PD! LGD (1 1.5 b)!! 1 + M 2.5 b 1.06 [1] 4

5 where: - LGD is the loss given default (considered to be 45% in the FIRB approach); - R is the correlation factor, computed as 0.12 (1 EXP(-50 PD! )) / (1 EXP(-50)) [1 (1 EXP(-50 PD! )) / (1 EXP(-50))]; - M is the time to maturity (considered to be 2.5 years in the FIRB approach); - b is the maturity adjustment, computed as ( ln(pd! )) 2. We now generate with a Monte Carlo simulation a sample of the bank s asset losses. This sample is generated employing the same distribution for the banks losses as utilized by the Basel Capital Accords in the IRB approach. If we denote by L ij the j th realization of the maximum loss for bank i given a certain random value of probability α ij and we assume that banks share the same value of asset correlation, we obtain L ij after 500,000 simulations of the following equation: 14 L!" = LGD N!!!! N!! PD! +!!!! N!!!" [2] Finally, as stated previously, we assume that a bank i defaults when L exceeds the sum of EL and CAP: L i > EL i + CAP i [3] Therefore, once a sample of banks asset losses has been generated, we check which banks fail according to the condition established by equation [3]. The PD of the i member institution of BRDIF (PD i ) is determined as the number of times bank fails on total simulations. Table 1 shows some descriptive statistics for the sample. The average PD for Spanish banks in 2013 was 0.05%, with a maximum value of 0.92%. The mean value for EAD was 17,596 million euros, with a maximum value of 257,361 million. Table 1. Descriptive statistics for the sample PD EAD Mean % 17,596 M Standard Deviation % 43,496 M Coefficient of Variation 263% 247% Minimum % 35 M Maximum % 257,361 M 5

6 After computing PD i, we can estimate the capital requirements needed by the BRDIF to cover unexpected losses at a given level of confidence or solvency level (K BRDIF ). This is determined by the sum of the capital requirements estimated for each bank member of the Fund, using the IRB formula for bank exposures: 15! K!"#$% =!!! K! EAD! [4] where, as stated before, EAD i is determined by the amount of deposits covered by the member institutions, and K (as a percentage of EAD) is obtained as follows: K! = LGD N 1 1 R N!! PD! + R 1 R N!! CL PD! LGD (1 1.5 b)!! 1 + M 2.5 b 1.06 [5] where, again, - LGD is the loss given default (considered to be 45%); - R is the correlation factor, computed as 0.12 (1 EXP(-50 PD i )) / (1 EXP(-50)) [1 (1 EXP(-50 PD i )) / (1 EXP(-50))]; - M is the time to maturity; - b is the maturity adjustment, computed as ( ln(pd i )) 2 ; - CL is the confidence level. Because the risk associated with the coverage of deposits does not depend on their maturity, we eliminate this effect setting the variable M equal to 1 in equation [5]. To examine different levels of solvency of the BRDIF, we consider that confidence levels are linked to the global corporate average cumulative 1-year default rates by Standard & Poor's. 16 RESULTS Table 2 reports the results obtained by considering different confidence levels linked to Standard & Poor's ratings. The capital requirements to covered deposits ratio (R1 = K BRDIF /Covered deposits) determines the resources needed in terms of covered deposits to assure a determined level of solvency of the Fund. The R2 ratio (1.8% covered deposits/ K BRDIF ) indicates whether the proposed EU fixed target levels are sufficient to ensure a 6

7 determined level of solvency of the Fund. Ratios greater than 100% suggest that the target levels set by the EU are adequate whereas ratios under 100% indicate insufficient resources to ensure a specific credit rating. Table 2. Capital requirements needed for a BRDIF in Spain Rating CL K BRDIF (Million euros) R1 R2 AAA/AA+ 100% 26, % 53% AA 99.98% 12, % 119% AA % 10, % 135% A % 8, % 173% A 99.93% 7, % 183% A % 7, % 192% BBB % 5, % 240% BBB 99.80% 5, % 280% BBB % 4, % 347% BB % 3, % 400% BB 99.32% 2, % 509% BB % 2, % 685% B % 1, % 1,114% B 95.27% % 2,067% B % % 3,762% The resources needed by the BRDIF to cover unexpected losses are greater as the desired rating increases. We find that the BRDIF would have required resources close to 3.4% of covered deposits to have obtained an equivalent AAA/AA+ S&P rating for the Spanish banking system in Therefore, the fixed target levels proposed by the EU bank resolution and deposit scheme directives would not ensure the highest credit quality of the BRDIF in Spain. Nevertheless, these resources would be sufficient to assure a good solvency level for the Fund, equivalent to an AA S&P rating in Finally, Table 3 considers different stress scenarios for a BRDIF with resources equal to 1.8% of covered deposits (i.e., the EU joint target level). We find that, as expected, when either LGD or PD increases, the rating of the BRDIF decreases. Table 3. Estimated rating for a Spanish BRDIF with resources equal to EU target levels Rating Scenario base AA LGD = 70% A+ LGD = 90% BBB+ 7

8 PD! 2 PD! 5 BBB- B CONCLUSIONS Our findings demonstrate that a Spanish BRDIF would have required resources close to 3.4% of covered deposits to obtain an equivalent AAA/AA+ S&P rating in However, although the fixed target levels proposed by the EU bank rules (i.e., 1.8% of covered deposits) would not ensure the highest credit quality of a BRDIF in Spain, these resources would be sufficient to ensure a good solvency level for the Fund, equivalent to an AA S&P rating in the analyzed period. REFERENCE AND NOTES 1. Directive 2014/59/EU of the European Parliament and of the Council of 15 May 2014 establishes a framework for the recovery and resolution of credit institutions and investment firms. 2. See Article 37 (3) of the BRRD. 3. See Article 103 (7) of the BRRD. 4. Directive 2014/49/EU of the European Parliament and of the Council of 16 April 2014 on deposit guarantee schemes. 5. LaBrosse, J. R., Olivares-Caminal, R. and Singh, D. (2014) The EU bank recovery and resolution directive some observations on the financing arrangements. Journal of Banking Regulation 15 (3-4): Episcopos, A (2004) The implied reserves of the bank insurance fund. Journal of Banking and Finance 28: Kuritzkes, A., Schuermann, T. and Weiner, S.M. (2005) Deposit insurance and risk management of the U.S. banking system: What is the loss distribution faced by the FDIC? Journal of Financial Services Research 27: De Lisa, R., Zedda, S., Vallascas, F., Campolongo, F. and Marchesi, M. (2011) Modelling deposit insurance scheme losses in a Basel 2 framework. Journal of Financial Services Research 40: Cariboni, J., Maccaferri, S. and Schoutens, W. (2011) Applying credit risk techniques to design an effective deposit guarantee schemes funds. Paper presented at the 2011 International Risk Management Conference, Amsterdam, The Netherlands, June. 10. Ho, C.L., Lai, G.C. and Lee, J.P. (2014) Financial reform and the adequacy of deposit insurance fund: Lessons from Taiwanese experience. International Review of Economics and Finance 30: Lee, S-C, Lin, C-T, and Tsai, M-S (2015) The pricing of deposit insurance in the presence of systematic risk. Journal of Banking and Finance, 51: Data obtained from the 2013 annual report of the current Spanish deposit guarantee fund of credit institutions (the socalled FGDEC), available at Similar to De Lisa et al. (2011) we consider a simplified banking system where banks are exposed only to credit risk and are fully compliant with the FIRB minimum capital requirements. 14. Initially, for the sake of simplicity, we assume that the asset correlation for our sample is 24%, i.e. the upper limit considered in the FIRB approach. By considering the existence of a correlation between bank s assets, we are taking into account a source of systemic risk (De Lisa et al., 2011). 15. The IRB approach proposed by the Basel framework to compute the regulatory capital requirements of banks derives from a credit portfolio model called the Asymptotic Risk Factor (ASRF) model, which in turn is resulting from the Merton model. The ASRF approach is used by the Basel framework to compute the capital needed to prevent a bank from bankruptcy under a one-year period, with a probability of more than This formula has been derived making the assumption that the portfolio is sufficiently fine grained (the number of assets is greater than 1000) so that the idiosyncratic risk is diversified away and therefore only the systematic risk remains. Although creating an homogeneous and fine grained portfolio could sound quite unrealistic, the ASRF approach presents some advantages in a regulatory point of view: it is relatively easy to implement, it yields the same capital charge when adding an asset independently on the structure of the initial portfolio, and it yields sufficiently good estimations of credit risk. Using multi-factor models would increase the accuracy of estimations but would be harder to implement because it requires a correlation matrix between each asset, where each element has to be estimated. Some commercial credit risk models (e.g., Moody s-kmv) propose a multi-factor approach to forecast credit risk with better accuracy. For a further discussion on this issue, please see Martin, L. (2013) Analysis of the IRB Asset Correlation Coefficient with an Application to a Credit Portfolio. Department of Mathematics, Uppsala University, Sweden. Project Report 2013: Standard and Poor s (2014) Default, Transition, and Recovery: 2013 annual global corporate default study and rating transitions, Mc Graw-Hill. 8