- Email: [email protected]
How do European banks cope with macroprudential capital requirements
Journal Pre-proof
How do European banks cope with macroprudential capital requirements
Sergio Mayordomo Conceptualization;Methodology;Software;Formal analysis; Writing- Original Draft;Writing Revie Mar´ıa Rodr´ıguez-Moreno Conceptualization;Methodology;Software;Formal analysis;Writing- Original Draft;Writing PII: DOI: Reference:
S1544-6123(19)31022-0 https://doi.org/10.1016/j.frl.2020.101459 FRL 101459
To appear in:
Finance Research Letters
Received date: Revised date: Accepted date:
19 September 2019 4 December 2019 11 February 2020
Please cite this article as: Sergio Mayordomo Conceptualization;Methodology;Software;Formal analysis; Writing- O Mar´ıa Rodr´ıguez-Moreno Conceptualization;Methodology;Software;Formal analysis;Writing- Original Draft;Writing How do European banks cope with macroprudential capital requirements, Finance Research Letters (2020), doi: https://doi.org/10.1016/j.frl.2020.101459
This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. © 2020 Published by Elsevier Inc.
Highlights Both Systemically Important Institutions (SII) and Non-SII have improved their capital ratios since the approval of the new macroprudential framework.
The sensitivity to macroprudential buffers varies between SII and Non-SII. Non-SII build up capital buffers to a higher extent than SII and tend to overreact to macroprudential requirements. Non-SII improve their capital position through a de-risking portfolio strategy, achieved by rebalancing portfolios towards safer assets.
How do European banks cope with macroprudential capital requirements?* Sergio Mayordomo Banco de España María Rodríguez-Moreno** Banco de España Draft: December 2019 * Mayordomo:
Banco de España, ADG Economics and Research, C/ Alcalá 48, 28014, Madrid, Spain. Email: [email protected]. Rodriguez-Moreno: Banco de España, Financial Stability and Macroprudential Policy Department, C/ Alcalá 48, 28014, Madrid, Spain. Email: [email protected]. We are very grateful for comments from Luis Gutiérrez de Rozas, Germán López-Espinosa, David Marqués-Ibáñez, Javier Mencía, Fernando Restoy, Rhiannon Sowerbutts and conference/seminar audiences at 2nd Annual Workshop of ESCB Research Cluster 3 (Frankfurt) and the Financial Stability Department (Banco de España). The views expressed in this paper are those of the authors and do not necessarily coincide with those of the Banco de España or the Eurosystem. Declarations of interest: none. ** Corresponding author.
Abstract This paper studies the effect of macroprudential requirements on capital ratios for a sample of euro area banks. We first document that banks’ capital ratios are typically above minimum regulatory levels. The banks in our sample differ in their degree of systemic importance and once we split the banks according to this criterion, we find that non-systemically important banks build up capital buffers to a higher extent than systemic banks and in excess of minimum requirements. The main channel through which these banks enhance their capital ratios is the optimization of risk-weighted assets, particularly by rebalancing portfolios towards safer assets.
JEL classification: G21, G28. Keywords: European banks; macroprudential requirements; capital ratios; risk weighted assets
1
1.
Introduction
The 2008 financial crisis brought about an unprecedented regulatory reform which has resulted in an increase in the quantity and quality of prudential requirements applicable to banks worldwide. In this new step-up, capital becomes a key cornerstone of the macroprudential regulatory framework whose primary objective is to enhance the resilience of the financial system and limit the build-up of vulnerabilities. At European level, the new capital requirements are defined by the Capital Requirements Regulation and Directive (CRR/CRD IV). These buffers include: i) countercyclical capital buffer (CCyB); ii) capital conservation buffer (CCoB); iii) systemic risk buffer (SyRB); and iv) capital buffers for global and other systemically important institutions (SIIB).1 Our paper contributes to the existing literature in three ways. We first study to what extent euro area (EA) banks boost their capital ratios when complying with these new macroprudential requirements. Although it has been widely documented that the consequence of stricter microprudential capital requirements is the adjustment of capital ratios (see Bridges et al., 2014; Gropp et al., 2019, among others), little is known about the effect of macroprudential measures on capital ratios, which contrary to microprudential requirements are publicly communicated and apply to all banks and not just to specific banks. Secondly, we analyze whether the enhancement of the capital ratio is achieved by increasing capital or by optimizing risk-weighted assets (RWA). Finally, we contribute to the literature that studies whether higher capital requirements hamper credit supply (i.e., Hyun and Rhee (2011), or Aiyar, Calomiris, and Wieladek (2014, 2016), among others) with the analysis of the effect of macroprudential requirements on bank’s credit supply and investments.
2.
Data
Bank’s balance-sheet information is at consolidated level and comes from SNL Financial. Our sample consists of 144 banks domiciled in 17 euro area countries with available information at 2013-Q4 and 2017-Q1. Germany (27), France (19), Italy (18), and Spain (16) are the countries with the highest number of banks in the sample. The sample consists not only of parent institutions but also includes national and foreign subsidiaries and
1
See the ECB (2016), ESRB (2014a, 2014b) for a description of the macroprudential policy framework.
2
foreign branches operating in a given country, providing additional variability to the macroprudential capital requirements at country level. We classify banks as Systemically Important Institution (SII) whenever the competent authority designated the institution as either global or domestic SII at 2017Q1 while the remaining institutions are classified as Non-Systemically Important Institutions (Non-SII).2 According to this classification, the sample consists of 44 SII whereas the remaining 100 banks are classified in the category Non-SII. Based on the differences in terms of capital requirements faced by the different groups of institutions, which have been manually collected from the European Systemic Risk Board (ESRB), and the dispersion across countries; we depict in Figure 1 the changes in capital ratios and the changes in the numerator and denominator of these ratios for each group. The average increase in the CET1 capital ratio of SII (1.56 percentage points, pp) is significantly lower than that obtained for Non-SII (2.38 pp). [Insert Figure 1 here] Importantly, the average CET1 capital ratios in 2013-Q4 for SII and Non-SII are 14.4% and 13.2%, respectively; which are well above the requirements faced by the two groups of banks at the end of our sample period: 6.8% and 5.8%, respectively. It suggests that there was no urgency on the part of the banks in our sample to increase their CET1 capital ratios to comply with the macroprudential buffers.
3.
Results
3.1.
Macroprudential buffers and the CET1 capital ratio
To study the effect of macroprudential requirements on the banks capital ratios, we run a regression analysis in which the dependent variable (∆𝐶𝐸𝑇1𝑏𝑐 ) is the variation in each bank b CET1 ratio from 2013-Q4 to 2017-Q1 and it is regressed on the capital buffers applied to each specific bank during the same time period (𝑀𝑃𝐵𝑏 ). Given that the banks in our sample are exposed to different macroprudential buffers and have different
2
SII can be split into global and domestic (also so-called other-SII). However, from the capital requirements perspective, both types of SII have to hold the highest of the following buffers: SyRB and SIIB. Thus, the fundamental distinction arises between SII and Non-SII institutions and consequently we treat global and domestic SII as a single group.
3
business models; we interact 𝑀𝑃𝐵𝑏 with the dummy that indicate the bank systemic treatment: SII or Non-SII. We run the following regression: ∆𝐶𝐸𝑇1𝑏𝑐 = 𝛼𝑏−𝑆𝐼𝐼 + 𝛽1 𝑀𝑃𝐵𝑏 𝑥𝑆𝐼𝐼𝑏 + 𝛽2 𝑀𝑃𝐵𝑏 𝑥𝑁𝑜𝑛 − 𝑆𝐼𝐼𝑏 + 𝛾𝐵𝐶𝑜𝑛𝑡𝑟𝑜𝑙𝑉𝑎𝑟𝑏
(1)
+ 𝛿𝐶𝐶𝑜𝑛𝑡𝑟𝑜𝑙𝑉𝑎𝑟𝑐 +𝜀𝑏𝑐
We use a set of control variables at bank level (𝐶𝑜𝑛𝑡𝑟𝑜𝑙𝑉𝑎𝑟𝑏 ) that include: the leverage measured as Tier 1 capital over total assets, the profitability (ROA), and the ratio of non-performing loans over total loans and use two dummy variables indicating whether the bank is a SII or a Non-SII (𝛼𝑏−𝑆𝐼𝐼 ).3 In addition, we include two control variables at country level (𝐶𝐶𝑜𝑛𝑡𝑟𝑜𝑙𝑉𝑎𝑟𝑐 ): GDP growth and 5-year sovereign CDS spread. All the control variables are defined according to their values as of 2013-Q4. The standard errors are robust to heteroskedasticity. Results are reported in column (1) of Panel A in Table 1. We observe that within the group of Non-SII, the higher the macroprudentual buffers, the higher the adjustment in the capital requirements. More specifically, an increase of 1 pp in the capital requirements to Non-SII leads to an increase of 2.5 pp in the CET1 capital ratio. In fact, this increase is statistically significantly higher than 1, which indicates that the increase of the capital ratio is significantly higher than the one required by macroprudential policies. This result is consistent with Cohen and Scatigna (2016) who document for the period 2009-2012 that smaller (i.e., non-systemic) banks held higher capital ratios compared to the systemic banks. However, this behavior is not observed for the group of SII, suggesting that there are not differences in the way they adjust their capital requirements after the increase in the macroprudential buffers. To further confirm the robustness of our results, we conduct a matching analysis in the spirit of Abadie and Imbens (2006, 2011). Namely, we estimate the average treatment effect of facing a macroprudential buffer above the median buffer within a given group of banks (i.e., SII or Non-SII). The matching among banks within each group is based on the same set of control variables employed in equation (1). Column (1) of Panel B in Table 1 contains the results for the group of SII whereas column (2) corresponds to Non-SII. The results are fully consistent with those contained in column
Note that the dummy variables indicating the bank’s systemic importance can be considered as proxies for the size among other dimensions reflecting the complexity and the importance of the bank for the financial system. 3
4
(1) of Panel A and indicate that within the group of Non-SII, the higher the macroprudential buffers, the higher the adjustment in the capital ratio. Nevertheless, it is important to note that these results do not imply that the SII do not react to macroprudential buffers and keep their capital ratio unchanged. In fact, the increase of the average capital ratios of SII and Non-SII between 2013-Q4 and 2017-Q1 was equal to 1.56 pp and 2.38 pp, respectively. These figures correspond to the unconditional change in the capital position. To provide further evidence on this issue, we estimate a variation of equation (1) in which we exclude the fixed effects at group level (𝛼𝑏−𝑆𝐼𝐼 ) and split the interaction between MPBb and the group of banks dummy in two: the average buffers within each group interacted with the group dummy and the deviation from the buffers faced by bank b from the average interacted with the group dummy. The coefficients obtained for the first type of interaction terms denote the average reaction within each group of banks whereas the coefficient obtained for the second type of interaction terms is equivalent to the one contained in column (1) of Table 1. It enables us to confirm that the average capital ratios of SII increase due to macroprudential requirements. More specifically, we find that given an average macroprudential buffer of 1 pp, the average capital ratio of SII exhibits an increase of 0.7 pp, which is significantly higher than zero. A buffer of a similar magnitude leads to a larger average increase for the case of Non-SII (1.7 pp).4 As summarised by Brei and Gambacorta (2016) and explained by Myers (1984), Marcus (1984), Berger et al. (2008), and Jokipii and Milne (2008); there are three reasons to explain a capital ratio in excess of the regulatory threshold, which particularly apply to the Non-SII and explain the overreaction to macroprudential buffers. Non-SII are less diversified which makes them more exposed to specific shocks and as a consequence to unexpected costs. In addition, they could use capital ratios as a signal to the market that could help them to obtain funds quickly and at a low cost in case the bank faces unexpected profit opportunities. Finally, the bailouts of large banks could exacerbate their moral hazard but Non-SII, which are smaller, might discount a lower expected bailout probability conditional on default and so, prefer larger capital buffers. However, the banks facing higher macroprudential buffers within the group of SII, do not adapt their capital ratios accordingly. One potential explanation for this is that the capital ratios were above 4
The results obtained from this regression analysis are not tabulated in the interest of brevity.
5
the ones required and as consequence, there was no need for an immediate adjustment in a period when raising new equity for financial institutions was very costly. 5
3.2.
How do banks improve their capital ratios?
Banks can achieve this goal by adjusting capital ratios through recapitalizing (i.e., increasing core equity capital), de-risking (i.e., reducing risk-weighted assets), or deleveraging (i.e., decreasing total assets). For a better understanding of this issue, we split the CET1 ratio in its two components relative to total assets. Thus, in column (2) of Panel A, we estimate equation (1) using as the dependent variable the CET1 relative to total assets whereas in column (3) of Panel A the dependent variable is the ratio of RWA over total assets. The CET1 over total assets ratio has not suffered significant changes in response to regulatory requirements in any of the two types of banks. However, as shown in column (3), both SII and Non-SII attempt to optimize the ratio of RWA over total assets. To analyze if there is also an adjustment through total assets, we repeat the analysis of equation (1) using as the dependent variable the change in the logarithm of total assets between 2013-Q4 and 2017-Q1. Results are reported in column (4) of Panel A and show that there is not a significant variation in the bank total assets of Non-SII, confirming that the enhancement in capital ratios is through a de-risking strategy. Nevertheless, the decrease in the SII’s ratio of RWA over total assets is due to the increase of total assets and that is why we do not observe a significant effect of macroprudential buffers on their capital ratio.
3.3.
RWA optimization and asset reallocation
To study how lending and investment policies are affected by the Non-SII’s de-risking strategy, we propose a variation of equation (1) in which the dependent variable is the variation in the ratio of loans over total assets between 2013-Q4 and 2017-Q1. Results are reported in column (1) of Table 2. We observe that Non-SII, which diminish their RWA to comply with macroprudential requirements, reduce their lending to a higher extent. In a context of low, even negative, interest rates; the lower profitability obtained from their loan portfolio could affect ultimately the capacity of this group of banks to further build capital organically through the profits obtained from their lending. 5
We carry out an additional matching analysis in which we compare Non-SII and SII facing similar macroprudential buffers and with similar characteristics based on the control variables used in equation (1). Results (not tabulated) confirm that Non-SII tend to increase their capital ratio to a larger extent than comparable SII.
6
In this scenario, banks aiming to increase their capital ratios could opt by a change in the structure of their balance and cut lending to enhance the quality of their capital through the purchase of safe assets such as sovereign debt. [Insert Table 2 here] Therefore, for a better understanding of banks’ efforts to improve capital ratios by reducing lending, we estimate equation (1) using as the dependent variable the variation in the ratio of holdings of sovereign debt and loans granted to central governments in the euro area (both with zero risk weights) over total assets between 2013-Q4 and 2017-Q1. The results are contained in column (2) of Table 2 and show that Non-SII enhanced their capital ratios by optimizing their RWA using assets with zero risk weights, which could have crowded out credit. Several robustness tests, which are not tabulated, confirm that the enhancement of RWA is not due to a mechanic adjustment due to the migration from a Standardised Approach (SA) to an Internal Ratings-Based (IRB) approach. Moreover, when accounting for the requirements of additional buffers in the context of the Pillar 2, we confirm that the enhancement of the capital ratio by Non-SII is due to the macroprudential buffers and not to the Pillar 2 requirements.
4.
Conclusions
We document that although the average capital ratios of both SII and Non-SII increase due to macroprudential requirements, the magnitude of this average increase differs across them, being much higher for Non-SII. In addition, those banks in the latter group that face higher macroprudential buffers increase their capital ratios to a higher extent, well beyond the requirement of macroprudential policies. The RWA optimization conducted by Non-SII leads to a decrease in their credit supply and to an increase in the investment in safe assets subject to zero risk weights (i.e., sovereign bonds and loans). These results suggest a trade-off between the effectiveness of macroprudential requirements with regard to financial stability and the ultimate effects on the real economy. This trade-off is more evident in a context of excessive capital buffers built by banks that depend to a larger extent on capital for signaling purposes, for absorbing shocks, or for their proper functioning in periods of stress.
7
AuthorContibutionStatement Sergio Mayordomo: Conceptualization, Methodology, Software, Formal analysis, Writing- Original Draft, Writing Review & Editing; María Rodríguez-Moreno: Conceptualization, Methodology, Software, Formal analysis, Writing- Original Draft, Writing Review & Editing;
References Abadie, A., and G. Imbens (2006). Large Sample Properties of Matching Estimators for Average Treatment Effects. Econometrica, 74, 235−67 Abadie, A., and G. Imbens (2011). Bias-Corrected Matching Estimators for Average Treatment Effects. Journal of Business and Economic Statistics, 29, 1−11. Aiyar, S., C. W. Calomiris, and T. Wieladek (2014). Does macro-prudential regulation leak? Evidence from a UK policy experiment. Journal of Money, Credit and Banking, 46, 181−214. Aiyar, S., C. W. Calomiris, and T. Wieladek (2016). How does credit supply respond to monetary policy and bank minimum capital requirements? European Economic Review, 82, 142−165. Berger, A.N., R. DeYoung, M.J. Flannery, D. Lee and Ö. Öztekin (2008). How Do Large Banking Organizations Manage Their Capital Ratios? Journal of Financial Services Research, 34, 123–149. Brei, M., L. Gambacorta (2016). Are bank capital ratios pro-cyclical? New evidence and perspectives. Economic Policy, 86, 357–403.Bridges, J., D. Gregory, M. Nielsen, S. Pezzini, A. Radia, and M. Spaltro (2014). The Impact of Capital Requirements on Bank Lending. Bank of England Working Paper No. 486. Cohen, B. H., and M. Scatigna (2016). Banks and capital requirements: Channels of adjustment. Journal of Banking and Finance, 69, 56–69. ECB (2016). Macroprudential Bulletin, Issue 1. ESRB (2014a). The ESRB Handbook on operationalising macro-prudential policy in the banking sector. ESRB (2014b). Flagship Report on macro-prudential policy in the banking sector. Gropp, R., T. Mosk, S. Ongena, and C. Wix (2019). Banks Response to Higher Capital Requirements: Evidence from a Quasi-Natural Experiment. The Review of Financial Studies, 32, 266–299. Hyun, J. S., and B. K. Rhee (2011). Bank Capital Regulation and Credit Supply. Journal of Banking and Finance, 35, 323–330. 8
Jokipii, J. and A. Milne (2008). The cyclical behaviour of European bank capital buffers. Journal of Banking and Finance, 32, 1440–1451. Marcus, A. (1984). Deregulation and bank financial policy. Journal of Banking and Finance, 8, 557–565. Myers, S., (1984). The capital structure puzzle. Journal of Finance, 39, 575–592.
9
Figure 1: Changes in CET1 ratio and its components This figure illustrates the distribution of the changes in the CET1 ratio, CET1 to total assets and RWA to total assets between 2013-Q4 and 2017-Q1. The mean is denoted by a circle whereas the median, the 25th, and the 75th percentiles are denoted by the three corresponding horizontal lines in each box. The 10 th and 90th percentiles correspond to the lines in the two extremes.
SII
Non-SII
0 -15
-10
-5
0 -5 -10 -15
-15
-10
-5
0
5
10
RWA/TA
5
10
CET1/TA
5
10
Cet1Ratio
SII
Non-SII
10
SII
Non-SII
Table 1: Macro prudential buffers and capital ratio adjustments The results reported in this table aim to provide evidence on the way in which the banks improve their capital ratios. Column (1) of Panel A reports the results obtained from the estimation of equation (1). Columns (2) – (3) of Panel A reports the coefficients estimated from a variation of equation (1) in which the dependent variables are the CET1 and the RWA over total assets, respectively. In column (4) of Panel A, we report the results obtained when the dependent variable is the percentage change in total assets. Standard errors, in brackets, are robust to heteroskedasticity. Panel B contains the results obtained using the matching estimation technique developed in Abadie and Imbens (2006, 2011). It reports the average treatment effect of facing a macroprudential buffer above the median buffer within a given group of banks (i.e., SII in column (1) and Non-SII in column (2)) on the change in the CET1 ratio. The matching among banks within each group is done using the nearest neighbor matching of the same control variables employed in the regression analysis. Standard errors are reported in brackets. *, ** and *** indicate statistical significance at the 10%, 5%, and 1% levels, respectively. Panel A: Regression Analysis Dependent Variable SII x MPB Non-SII x MPB Bank Control Variables Country Control Variables Bank Type Dummy Variables Observations R-squared Panel B: Matching Analysis Sample of Banks Dependent Variable Banks with higher MPB Observations
(1) ΔCET1 Ratio 0.137 [0.301] 2.517** [1.007] Yes Yes
(2)
(3)
(4)
ΔCET1/TA
ΔRWA/TA
Δlog(TA)
-0.287 [0.194] -0.261 [0.829] Yes Yes
-1.500** [0.704] -3.670* [2.068] Yes Yes
0.057** [0.022] 0.022 [0.082] Yes Yes
Yes
Yes
Yes
Yes
144 0.31
144 0.16
144 0.097
144 0.091
(1) SII ΔCET1 Ratio 0.561 [1.361] 44
11
(2) Non-SII ΔCET1 Ratio 2.741* [1.476] 100
Table 2: RWA optimization and asset reallocation The results reported in this table aim to provide evidence on the impact of RWA optimization on the bank lending activity and the holding of sovereign bonds. Column (1) of this table reports the results obtained from the estimation a variation of equation (1) in which the dependent variable is the variation in the ratio of bank’s loans over total assets between 2013-Q4 and 2017-Q1. Column (2) reports the results obtained from the estimation a variation of equation (1) in which the dependent variable is the variation in the exposure to central governments in the euro area between 2013-Q4 and 2017-Q1 relative to total assets. Standard errors, in brackets, are robust to heteroskedasticity. *, ** and *** indicate statistical significance at the 10%, 5%, and 1% levels, respectively. Variables SII x MPB Non-SII x MPB Bank Control Variables Country Control Variables Bank Type Dummy Variables Observations R-squared
12
(1) ΔLoan/TA
(2) ΔGovExp
-0.969 [1.041] -4.768*** [1.605]
0.026 [1.083] 6.368*** [2.106]
Yes Yes Yes 144 0.212
Yes Yes Yes 144 0.108
How do European banks cope with macroprudential capital requirements
Sergio Mayordomo Conceptualization;Methodology;Software;Formal analysis; Writing- Original Draft;Writing Revie Mar´ıa Rodr´ıguez-Moreno Conceptualization;Methodology;Software;Formal analysis;Writing- Original Draft;Writing PII: DOI: Reference:
S1544-6123(19)31022-0 https://doi.org/10.1016/j.frl.2020.101459 FRL 101459
To appear in:
Finance Research Letters
Received date: Revised date: Accepted date:
19 September 2019 4 December 2019 11 February 2020
Please cite this article as: Sergio Mayordomo Conceptualization;Methodology;Software;Formal analysis; Writing- O Mar´ıa Rodr´ıguez-Moreno Conceptualization;Methodology;Software;Formal analysis;Writing- Original Draft;Writing How do European banks cope with macroprudential capital requirements, Finance Research Letters (2020), doi: https://doi.org/10.1016/j.frl.2020.101459
This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. © 2020 Published by Elsevier Inc.
Highlights Both Systemically Important Institutions (SII) and Non-SII have improved their capital ratios since the approval of the new macroprudential framework.
The sensitivity to macroprudential buffers varies between SII and Non-SII. Non-SII build up capital buffers to a higher extent than SII and tend to overreact to macroprudential requirements. Non-SII improve their capital position through a de-risking portfolio strategy, achieved by rebalancing portfolios towards safer assets.
How do European banks cope with macroprudential capital requirements?* Sergio Mayordomo Banco de España María Rodríguez-Moreno** Banco de España Draft: December 2019 * Mayordomo:
Banco de España, ADG Economics and Research, C/ Alcalá 48, 28014, Madrid, Spain. Email: [email protected]. Rodriguez-Moreno: Banco de España, Financial Stability and Macroprudential Policy Department, C/ Alcalá 48, 28014, Madrid, Spain. Email: [email protected]. We are very grateful for comments from Luis Gutiérrez de Rozas, Germán López-Espinosa, David Marqués-Ibáñez, Javier Mencía, Fernando Restoy, Rhiannon Sowerbutts and conference/seminar audiences at 2nd Annual Workshop of ESCB Research Cluster 3 (Frankfurt) and the Financial Stability Department (Banco de España). The views expressed in this paper are those of the authors and do not necessarily coincide with those of the Banco de España or the Eurosystem. Declarations of interest: none. ** Corresponding author.
Abstract This paper studies the effect of macroprudential requirements on capital ratios for a sample of euro area banks. We first document that banks’ capital ratios are typically above minimum regulatory levels. The banks in our sample differ in their degree of systemic importance and once we split the banks according to this criterion, we find that non-systemically important banks build up capital buffers to a higher extent than systemic banks and in excess of minimum requirements. The main channel through which these banks enhance their capital ratios is the optimization of risk-weighted assets, particularly by rebalancing portfolios towards safer assets.
JEL classification: G21, G28. Keywords: European banks; macroprudential requirements; capital ratios; risk weighted assets
1
1.
Introduction
The 2008 financial crisis brought about an unprecedented regulatory reform which has resulted in an increase in the quantity and quality of prudential requirements applicable to banks worldwide. In this new step-up, capital becomes a key cornerstone of the macroprudential regulatory framework whose primary objective is to enhance the resilience of the financial system and limit the build-up of vulnerabilities. At European level, the new capital requirements are defined by the Capital Requirements Regulation and Directive (CRR/CRD IV). These buffers include: i) countercyclical capital buffer (CCyB); ii) capital conservation buffer (CCoB); iii) systemic risk buffer (SyRB); and iv) capital buffers for global and other systemically important institutions (SIIB).1 Our paper contributes to the existing literature in three ways. We first study to what extent euro area (EA) banks boost their capital ratios when complying with these new macroprudential requirements. Although it has been widely documented that the consequence of stricter microprudential capital requirements is the adjustment of capital ratios (see Bridges et al., 2014; Gropp et al., 2019, among others), little is known about the effect of macroprudential measures on capital ratios, which contrary to microprudential requirements are publicly communicated and apply to all banks and not just to specific banks. Secondly, we analyze whether the enhancement of the capital ratio is achieved by increasing capital or by optimizing risk-weighted assets (RWA). Finally, we contribute to the literature that studies whether higher capital requirements hamper credit supply (i.e., Hyun and Rhee (2011), or Aiyar, Calomiris, and Wieladek (2014, 2016), among others) with the analysis of the effect of macroprudential requirements on bank’s credit supply and investments.
2.
Data
Bank’s balance-sheet information is at consolidated level and comes from SNL Financial. Our sample consists of 144 banks domiciled in 17 euro area countries with available information at 2013-Q4 and 2017-Q1. Germany (27), France (19), Italy (18), and Spain (16) are the countries with the highest number of banks in the sample. The sample consists not only of parent institutions but also includes national and foreign subsidiaries and
1
See the ECB (2016), ESRB (2014a, 2014b) for a description of the macroprudential policy framework.
2
foreign branches operating in a given country, providing additional variability to the macroprudential capital requirements at country level. We classify banks as Systemically Important Institution (SII) whenever the competent authority designated the institution as either global or domestic SII at 2017Q1 while the remaining institutions are classified as Non-Systemically Important Institutions (Non-SII).2 According to this classification, the sample consists of 44 SII whereas the remaining 100 banks are classified in the category Non-SII. Based on the differences in terms of capital requirements faced by the different groups of institutions, which have been manually collected from the European Systemic Risk Board (ESRB), and the dispersion across countries; we depict in Figure 1 the changes in capital ratios and the changes in the numerator and denominator of these ratios for each group. The average increase in the CET1 capital ratio of SII (1.56 percentage points, pp) is significantly lower than that obtained for Non-SII (2.38 pp). [Insert Figure 1 here] Importantly, the average CET1 capital ratios in 2013-Q4 for SII and Non-SII are 14.4% and 13.2%, respectively; which are well above the requirements faced by the two groups of banks at the end of our sample period: 6.8% and 5.8%, respectively. It suggests that there was no urgency on the part of the banks in our sample to increase their CET1 capital ratios to comply with the macroprudential buffers.
3.
Results
3.1.
Macroprudential buffers and the CET1 capital ratio
To study the effect of macroprudential requirements on the banks capital ratios, we run a regression analysis in which the dependent variable (∆𝐶𝐸𝑇1𝑏𝑐 ) is the variation in each bank b CET1 ratio from 2013-Q4 to 2017-Q1 and it is regressed on the capital buffers applied to each specific bank during the same time period (𝑀𝑃𝐵𝑏 ). Given that the banks in our sample are exposed to different macroprudential buffers and have different
2
SII can be split into global and domestic (also so-called other-SII). However, from the capital requirements perspective, both types of SII have to hold the highest of the following buffers: SyRB and SIIB. Thus, the fundamental distinction arises between SII and Non-SII institutions and consequently we treat global and domestic SII as a single group.
3
business models; we interact 𝑀𝑃𝐵𝑏 with the dummy that indicate the bank systemic treatment: SII or Non-SII. We run the following regression: ∆𝐶𝐸𝑇1𝑏𝑐 = 𝛼𝑏−𝑆𝐼𝐼 + 𝛽1 𝑀𝑃𝐵𝑏 𝑥𝑆𝐼𝐼𝑏 + 𝛽2 𝑀𝑃𝐵𝑏 𝑥𝑁𝑜𝑛 − 𝑆𝐼𝐼𝑏 + 𝛾𝐵𝐶𝑜𝑛𝑡𝑟𝑜𝑙𝑉𝑎𝑟𝑏
(1)
+ 𝛿𝐶𝐶𝑜𝑛𝑡𝑟𝑜𝑙𝑉𝑎𝑟𝑐 +𝜀𝑏𝑐
We use a set of control variables at bank level (𝐶𝑜𝑛𝑡𝑟𝑜𝑙𝑉𝑎𝑟𝑏 ) that include: the leverage measured as Tier 1 capital over total assets, the profitability (ROA), and the ratio of non-performing loans over total loans and use two dummy variables indicating whether the bank is a SII or a Non-SII (𝛼𝑏−𝑆𝐼𝐼 ).3 In addition, we include two control variables at country level (𝐶𝐶𝑜𝑛𝑡𝑟𝑜𝑙𝑉𝑎𝑟𝑐 ): GDP growth and 5-year sovereign CDS spread. All the control variables are defined according to their values as of 2013-Q4. The standard errors are robust to heteroskedasticity. Results are reported in column (1) of Panel A in Table 1. We observe that within the group of Non-SII, the higher the macroprudentual buffers, the higher the adjustment in the capital requirements. More specifically, an increase of 1 pp in the capital requirements to Non-SII leads to an increase of 2.5 pp in the CET1 capital ratio. In fact, this increase is statistically significantly higher than 1, which indicates that the increase of the capital ratio is significantly higher than the one required by macroprudential policies. This result is consistent with Cohen and Scatigna (2016) who document for the period 2009-2012 that smaller (i.e., non-systemic) banks held higher capital ratios compared to the systemic banks. However, this behavior is not observed for the group of SII, suggesting that there are not differences in the way they adjust their capital requirements after the increase in the macroprudential buffers. To further confirm the robustness of our results, we conduct a matching analysis in the spirit of Abadie and Imbens (2006, 2011). Namely, we estimate the average treatment effect of facing a macroprudential buffer above the median buffer within a given group of banks (i.e., SII or Non-SII). The matching among banks within each group is based on the same set of control variables employed in equation (1). Column (1) of Panel B in Table 1 contains the results for the group of SII whereas column (2) corresponds to Non-SII. The results are fully consistent with those contained in column
Note that the dummy variables indicating the bank’s systemic importance can be considered as proxies for the size among other dimensions reflecting the complexity and the importance of the bank for the financial system. 3
4
(1) of Panel A and indicate that within the group of Non-SII, the higher the macroprudential buffers, the higher the adjustment in the capital ratio. Nevertheless, it is important to note that these results do not imply that the SII do not react to macroprudential buffers and keep their capital ratio unchanged. In fact, the increase of the average capital ratios of SII and Non-SII between 2013-Q4 and 2017-Q1 was equal to 1.56 pp and 2.38 pp, respectively. These figures correspond to the unconditional change in the capital position. To provide further evidence on this issue, we estimate a variation of equation (1) in which we exclude the fixed effects at group level (𝛼𝑏−𝑆𝐼𝐼 ) and split the interaction between MPBb and the group of banks dummy in two: the average buffers within each group interacted with the group dummy and the deviation from the buffers faced by bank b from the average interacted with the group dummy. The coefficients obtained for the first type of interaction terms denote the average reaction within each group of banks whereas the coefficient obtained for the second type of interaction terms is equivalent to the one contained in column (1) of Table 1. It enables us to confirm that the average capital ratios of SII increase due to macroprudential requirements. More specifically, we find that given an average macroprudential buffer of 1 pp, the average capital ratio of SII exhibits an increase of 0.7 pp, which is significantly higher than zero. A buffer of a similar magnitude leads to a larger average increase for the case of Non-SII (1.7 pp).4 As summarised by Brei and Gambacorta (2016) and explained by Myers (1984), Marcus (1984), Berger et al. (2008), and Jokipii and Milne (2008); there are three reasons to explain a capital ratio in excess of the regulatory threshold, which particularly apply to the Non-SII and explain the overreaction to macroprudential buffers. Non-SII are less diversified which makes them more exposed to specific shocks and as a consequence to unexpected costs. In addition, they could use capital ratios as a signal to the market that could help them to obtain funds quickly and at a low cost in case the bank faces unexpected profit opportunities. Finally, the bailouts of large banks could exacerbate their moral hazard but Non-SII, which are smaller, might discount a lower expected bailout probability conditional on default and so, prefer larger capital buffers. However, the banks facing higher macroprudential buffers within the group of SII, do not adapt their capital ratios accordingly. One potential explanation for this is that the capital ratios were above 4
The results obtained from this regression analysis are not tabulated in the interest of brevity.
5
the ones required and as consequence, there was no need for an immediate adjustment in a period when raising new equity for financial institutions was very costly. 5
3.2.
How do banks improve their capital ratios?
Banks can achieve this goal by adjusting capital ratios through recapitalizing (i.e., increasing core equity capital), de-risking (i.e., reducing risk-weighted assets), or deleveraging (i.e., decreasing total assets). For a better understanding of this issue, we split the CET1 ratio in its two components relative to total assets. Thus, in column (2) of Panel A, we estimate equation (1) using as the dependent variable the CET1 relative to total assets whereas in column (3) of Panel A the dependent variable is the ratio of RWA over total assets. The CET1 over total assets ratio has not suffered significant changes in response to regulatory requirements in any of the two types of banks. However, as shown in column (3), both SII and Non-SII attempt to optimize the ratio of RWA over total assets. To analyze if there is also an adjustment through total assets, we repeat the analysis of equation (1) using as the dependent variable the change in the logarithm of total assets between 2013-Q4 and 2017-Q1. Results are reported in column (4) of Panel A and show that there is not a significant variation in the bank total assets of Non-SII, confirming that the enhancement in capital ratios is through a de-risking strategy. Nevertheless, the decrease in the SII’s ratio of RWA over total assets is due to the increase of total assets and that is why we do not observe a significant effect of macroprudential buffers on their capital ratio.
3.3.
RWA optimization and asset reallocation
To study how lending and investment policies are affected by the Non-SII’s de-risking strategy, we propose a variation of equation (1) in which the dependent variable is the variation in the ratio of loans over total assets between 2013-Q4 and 2017-Q1. Results are reported in column (1) of Table 2. We observe that Non-SII, which diminish their RWA to comply with macroprudential requirements, reduce their lending to a higher extent. In a context of low, even negative, interest rates; the lower profitability obtained from their loan portfolio could affect ultimately the capacity of this group of banks to further build capital organically through the profits obtained from their lending. 5
We carry out an additional matching analysis in which we compare Non-SII and SII facing similar macroprudential buffers and with similar characteristics based on the control variables used in equation (1). Results (not tabulated) confirm that Non-SII tend to increase their capital ratio to a larger extent than comparable SII.
6
In this scenario, banks aiming to increase their capital ratios could opt by a change in the structure of their balance and cut lending to enhance the quality of their capital through the purchase of safe assets such as sovereign debt. [Insert Table 2 here] Therefore, for a better understanding of banks’ efforts to improve capital ratios by reducing lending, we estimate equation (1) using as the dependent variable the variation in the ratio of holdings of sovereign debt and loans granted to central governments in the euro area (both with zero risk weights) over total assets between 2013-Q4 and 2017-Q1. The results are contained in column (2) of Table 2 and show that Non-SII enhanced their capital ratios by optimizing their RWA using assets with zero risk weights, which could have crowded out credit. Several robustness tests, which are not tabulated, confirm that the enhancement of RWA is not due to a mechanic adjustment due to the migration from a Standardised Approach (SA) to an Internal Ratings-Based (IRB) approach. Moreover, when accounting for the requirements of additional buffers in the context of the Pillar 2, we confirm that the enhancement of the capital ratio by Non-SII is due to the macroprudential buffers and not to the Pillar 2 requirements.
4.
Conclusions
We document that although the average capital ratios of both SII and Non-SII increase due to macroprudential requirements, the magnitude of this average increase differs across them, being much higher for Non-SII. In addition, those banks in the latter group that face higher macroprudential buffers increase their capital ratios to a higher extent, well beyond the requirement of macroprudential policies. The RWA optimization conducted by Non-SII leads to a decrease in their credit supply and to an increase in the investment in safe assets subject to zero risk weights (i.e., sovereign bonds and loans). These results suggest a trade-off between the effectiveness of macroprudential requirements with regard to financial stability and the ultimate effects on the real economy. This trade-off is more evident in a context of excessive capital buffers built by banks that depend to a larger extent on capital for signaling purposes, for absorbing shocks, or for their proper functioning in periods of stress.
7
AuthorContibutionStatement Sergio Mayordomo: Conceptualization, Methodology, Software, Formal analysis, Writing- Original Draft, Writing Review & Editing; María Rodríguez-Moreno: Conceptualization, Methodology, Software, Formal analysis, Writing- Original Draft, Writing Review & Editing;
References Abadie, A., and G. Imbens (2006). Large Sample Properties of Matching Estimators for Average Treatment Effects. Econometrica, 74, 235−67 Abadie, A., and G. Imbens (2011). Bias-Corrected Matching Estimators for Average Treatment Effects. Journal of Business and Economic Statistics, 29, 1−11. Aiyar, S., C. W. Calomiris, and T. Wieladek (2014). Does macro-prudential regulation leak? Evidence from a UK policy experiment. Journal of Money, Credit and Banking, 46, 181−214. Aiyar, S., C. W. Calomiris, and T. Wieladek (2016). How does credit supply respond to monetary policy and bank minimum capital requirements? European Economic Review, 82, 142−165. Berger, A.N., R. DeYoung, M.J. Flannery, D. Lee and Ö. Öztekin (2008). How Do Large Banking Organizations Manage Their Capital Ratios? Journal of Financial Services Research, 34, 123–149. Brei, M., L. Gambacorta (2016). Are bank capital ratios pro-cyclical? New evidence and perspectives. Economic Policy, 86, 357–403.Bridges, J., D. Gregory, M. Nielsen, S. Pezzini, A. Radia, and M. Spaltro (2014). The Impact of Capital Requirements on Bank Lending. Bank of England Working Paper No. 486. Cohen, B. H., and M. Scatigna (2016). Banks and capital requirements: Channels of adjustment. Journal of Banking and Finance, 69, 56–69. ECB (2016). Macroprudential Bulletin, Issue 1. ESRB (2014a). The ESRB Handbook on operationalising macro-prudential policy in the banking sector. ESRB (2014b). Flagship Report on macro-prudential policy in the banking sector. Gropp, R., T. Mosk, S. Ongena, and C. Wix (2019). Banks Response to Higher Capital Requirements: Evidence from a Quasi-Natural Experiment. The Review of Financial Studies, 32, 266–299. Hyun, J. S., and B. K. Rhee (2011). Bank Capital Regulation and Credit Supply. Journal of Banking and Finance, 35, 323–330. 8
Jokipii, J. and A. Milne (2008). The cyclical behaviour of European bank capital buffers. Journal of Banking and Finance, 32, 1440–1451. Marcus, A. (1984). Deregulation and bank financial policy. Journal of Banking and Finance, 8, 557–565. Myers, S., (1984). The capital structure puzzle. Journal of Finance, 39, 575–592.
9
Figure 1: Changes in CET1 ratio and its components This figure illustrates the distribution of the changes in the CET1 ratio, CET1 to total assets and RWA to total assets between 2013-Q4 and 2017-Q1. The mean is denoted by a circle whereas the median, the 25th, and the 75th percentiles are denoted by the three corresponding horizontal lines in each box. The 10 th and 90th percentiles correspond to the lines in the two extremes.
SII
Non-SII
0 -15
-10
-5
0 -5 -10 -15
-15
-10
-5
0
5
10
RWA/TA
5
10
CET1/TA
5
10
Cet1Ratio
SII
Non-SII
10
SII
Non-SII
Table 1: Macro prudential buffers and capital ratio adjustments The results reported in this table aim to provide evidence on the way in which the banks improve their capital ratios. Column (1) of Panel A reports the results obtained from the estimation of equation (1). Columns (2) – (3) of Panel A reports the coefficients estimated from a variation of equation (1) in which the dependent variables are the CET1 and the RWA over total assets, respectively. In column (4) of Panel A, we report the results obtained when the dependent variable is the percentage change in total assets. Standard errors, in brackets, are robust to heteroskedasticity. Panel B contains the results obtained using the matching estimation technique developed in Abadie and Imbens (2006, 2011). It reports the average treatment effect of facing a macroprudential buffer above the median buffer within a given group of banks (i.e., SII in column (1) and Non-SII in column (2)) on the change in the CET1 ratio. The matching among banks within each group is done using the nearest neighbor matching of the same control variables employed in the regression analysis. Standard errors are reported in brackets. *, ** and *** indicate statistical significance at the 10%, 5%, and 1% levels, respectively. Panel A: Regression Analysis Dependent Variable SII x MPB Non-SII x MPB Bank Control Variables Country Control Variables Bank Type Dummy Variables Observations R-squared Panel B: Matching Analysis Sample of Banks Dependent Variable Banks with higher MPB Observations
(1) ΔCET1 Ratio 0.137 [0.301] 2.517** [1.007] Yes Yes
(2)
(3)
(4)
ΔCET1/TA
ΔRWA/TA
Δlog(TA)
-0.287 [0.194] -0.261 [0.829] Yes Yes
-1.500** [0.704] -3.670* [2.068] Yes Yes
0.057** [0.022] 0.022 [0.082] Yes Yes
Yes
Yes
Yes
Yes
144 0.31
144 0.16
144 0.097
144 0.091
(1) SII ΔCET1 Ratio 0.561 [1.361] 44
11
(2) Non-SII ΔCET1 Ratio 2.741* [1.476] 100
Table 2: RWA optimization and asset reallocation The results reported in this table aim to provide evidence on the impact of RWA optimization on the bank lending activity and the holding of sovereign bonds. Column (1) of this table reports the results obtained from the estimation a variation of equation (1) in which the dependent variable is the variation in the ratio of bank’s loans over total assets between 2013-Q4 and 2017-Q1. Column (2) reports the results obtained from the estimation a variation of equation (1) in which the dependent variable is the variation in the exposure to central governments in the euro area between 2013-Q4 and 2017-Q1 relative to total assets. Standard errors, in brackets, are robust to heteroskedasticity. *, ** and *** indicate statistical significance at the 10%, 5%, and 1% levels, respectively. Variables SII x MPB Non-SII x MPB Bank Control Variables Country Control Variables Bank Type Dummy Variables Observations R-squared
12
(1) ΔLoan/TA
(2) ΔGovExp
-0.969 [1.041] -4.768*** [1.605]
0.026 [1.083] 6.368*** [2.106]
Yes Yes Yes 144 0.212
Yes Yes Yes 144 0.108