1 Bachelor’s Thesis
Contingent Convertibles: how do
their risks affect banking firm’s
stock prices?
JEL Classification: G12, G13, G18, G21, G28, G32
Abstract
This research argues whether contingent convertibles capital instruments (CoCos) are suitable to increase the financial stability in the banking sector. The empirical results are consistent with existing literature, in which it is shown that badly designed CoCo bonds in
practice are related to stock prices and therefore increase the risk of a systemic crisis.
Jan Willem Stal
Student number: 5674468
BSc Economics and Business
Specialization: Economics & Finance
University of Amsterdam
Supervisor: I. Sakalauskaite MSc
Date: 29 June 2016
2 Statement of Originality
This document is written by Student Jan Willem Stal, who declares to take full responsibility for the contents of this document. I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it. The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
3 List of Contents
1. Introduction ... 3
2. Literature Review ... 6
2.1. CoCos as a capital instrument ... 6
2.2. Regulatory treatment of CoCos ... 8
2.3. Risks of CoCo bonds ... 9
2.4. CoCos and banks’ risk-taking behavior ... 10
2.5. Systemic risks related to triggering CoCos ... 11
2.6. Recent turbulence on financial markets regarding CoCos and hypothesis .... 11
3. Empirical research ... 13
3.1. Data set ... 13
3.2. Method ... 13
4. Results & Analysis ... 16
4.1. Descriptive statistics ... 16
4.2. Estimated betas ... 18
4.3. Difference in difference analysis ... 23
5. Conclusion ... 25
4 1. Introduction
The financials crisis of 2007 – 2009 revealed that equity-capital requirements for banks had not been sufficient. To be able to survive negative macro-economic shocks, banks are urged by regulators to strengthen their balance sheets. The Basel Committee on Banking
Supervision therefore defined additional measures in the Basel III regulatory framework, in order to increase the stability of the banking sector. Among the additional measures is the admission for banks to issue contingent convertible capital (CoCos), which under certain conditions qualify as Additional Tier 1 (AT1) capital. CoCos maybe included as Tier 2 (T2) capital when these conditions are not met.
CoCos are hybrid instruments that provide banks with higher capital cushions when private investors refuse to. Upon the occurrence of a so-called ‘trigger event’, CoCo bonds either turn into equity or are written off partially or completely. These hybrid instruments are designed this way, so that in situations of financial distress they provide an effective and quick way to relieve the burden of having to raise capital in situations (Flannery, 2002). In times of financial distress, private investors are reluctant to provide additional capital to banks. As we have seen during the financial crisis, this could lead to public capital injections or even a government resolution.
Since under the Basel III regulatory framework CoCos have become part of AT1 capital, the issuance of CoCo bonds has especially become popular among banks in Europe. According to Avdjiev (2015) the CoCo market has been rapidly growing in recent years, although it still is a relatively small market segment. The asset quality review (AQR), which in 2014 was undertaken by the European Central Bank (ECB) on the largest
European financial institutions, revealed that over one third of equity issues between July 2013 and August 2014 were through the issuance of CoCos.
Although CoCos are increasingly popular among banks as an instrument to meet higher capital requirements, there is still a lot of uncertainty on their regulatory treatment. In an opinion paper that was published in December 2015, the European Banking Authority (EBA) says that the supervisor should consider the prohibition on the distribution of profits in all circumstances where, in a given year, no profits are made and the capital requirements are not met. Investors had their doubts on the interpretation and the implications of the EBA paper, especially regarding the ability of European banks to fulfill future coupon payments on CoCo bonds. These concerns led to a sell-off in the first months of 2016 on the CoCo
market, which was accompanied by a drop of banks their stock prices.
Investing in CoCo bonds is especially attractive for investors in the current
low-interest environment. Questions are raised however, whether investors have underestimated the risks of CoCo bonds in their search for yield. The growing popularity of the AT1
5 research on the CoCo market.
A sudden depreciation of banks’ equity could possibly lead to a conversion of CoCo bonds, if these hybrid instruments are badly designed (Sundaresen & Wang, 2010). This research therefore aims to answer the question if the recent turbulence on financial markets can be considered as a revaluation of CoCo bonds and how this affects the stock prices of banks? To answer the research question, a data set is collected on CoCos and banking firm’s equity for the period from December 2014 to May 2016. The empirical part of this research aims to measure a possible change in the risks banking assets since the publication of the EBA opinion paper.
The remainder of this paper will start with a literature review on CoCo instruments, in which the institutional background, the regulatory treatment, practicalities and risks of CoCos will be discussed. A simple analytical framework will be described in section 3, by which this research tries to link the data on CoCo bonds to the existing literature. In section 4 the outcomes of the research will be presented, which leads to the conclusions in the final section.
6 2. Literature Review
In this section, characteristics of CoCo bonds will be discussed. First, the design features of CoCos will be explained. In the second subsection it will become clear why regulators allow banks to let these hybrid instruments be part of their Common Equity Tier 1 (CET1) capital. In the next subsection some characteristics of CoCo bonds will be covered, followed by a discussion on systemic risks that are related to triggering CoCos. The final subsection gives a brief overview of the recent turbulence on financial markets regarding CoCos and how this relates to the existing literature.
2.1. CoCos as a capital instrument
Contingent convertible capital instruments (CoCos) are hybrid capital securities that absorb losses in accordance with their contractual terms, when the capital of the issuing bank falls below a certain level. If this level is not breached however, CoCos tend to behave like ordinary bonds; they can be considered as fixed liabilities with predetermined coupon payments. The design of CoCos ensures a source of capital to the banks in distress when private investors are reluctant to supply in external capital (Avdjiev, 2015). According to Avdjiev (2013) there are two defining characteristics: the loss absorption mechanism and the trigger that activates this mechanism.
A CoCo bond has the ability to absorb losses of the issuing bank by either converting into common equity or by suffering a principal write-down. Triggers that activate the loss absorption mechanism can be based on a mechanical rule or on the supervisors’ discretion (Avdjiev, 2013). With a mechanical trigger, the mechanism is activated when the capital of the bank falls below a fraction of its risk-weighted assets (RWA). This fraction is
pre-specified and the capital measure can be based on book-values or market values. Typically, book-value triggers are set contractually in terms of the book-value of a bank’s CET1 capital as a ratio of RWA. Market triggers however are set at a minimum ratio of the CoCo issuing bank’s stock market capitalization to its assets.
7 Figure 1: Main design features of CoCos (Avdjiev, 2013)
The effectiveness of book-value triggers, which are also known as accounting-value triggers, depends on the frequency at which the CET1 as a ratio of RWA is calculated and publicly disclosed (Avdjiev, 2013). The consistency of the internal risk models can vary significant across banks and time, and therefore book-value triggers may not be activated in a timely fashion. The shortcoming of inconsistent accounting measures and valuations may be tackled by setting triggers in terms of the market value of a bank.
However, in practice very few of the contractual terms regarding CoCo issues rely purely on market valuations. In most cases, the loss absorption mechanism is not activated upon a specific and predetermined numerical value, but rather upon the supervisors’ judgment on bank solvency and hence whether the trigger has been breached, leading to discretionary “point of non-viability” (PONV) triggers (Avdjiev, 2015). The PONV is defined as the time when a bank is not able to support itself and a public capital injection or a government resolution is unavoidable (Pennacchi, 2014). That is, the bank is a “gone-concern”. The inclusion of PONV clauses in CoCos is primarily motivated by regulatory capital eligibility considerations.
It is clear that a thicker layer of CoCos and the further a trigger is set from PONV, the more likely CoCo issuance will ensure that the balance-sheet of a bank will be strengthened on a going-concern basis (Avdjiev, 2015). However, CoCos with triggers that are set
relatively low will offer more favorable terms to holders of CoCos than to equity holders. Especially in the current low interest environment, this might make CoCos interesting for investors in their search for yield. With triggers that are set relatively low it is less likely that these triggers are breached (Avdjiev, 2013). Therefore CoCo holders are less likely to absorb losses.
8 2.2. Regulatory treatment of CoCos
In the aftermath of the 2007 – 2009 financial crisis, the Basel III regulatory framework was formed. It had the objective to repair some significant deficiencies in the Basel I and Basel II Accords that were revealed by the crisis. According to the Basel III regulatory framework, banks have to keep their total capital a level of at least 8% of the RWA. The total capital of a bank consists of CET1, Additional Tier 1 (AT1) capital and Tier 2 (T2) capital (Basel
Committee, 2010).
Figure 2: CoCos’ position in the Basel III capital requirements (Avdjiev, 2013).
In the current Basel III framework, a CoCo bond can qualify as Additional Tier 1 (AT1) capital if its design contains two key elements: (i) a PONV trigger requirement and (ii) a “going-concern” contingent capital requirement with a trigger level of 5.125% in terms of
CET1/RWA (Boermans, 2014). In addition, all AT1 capital must be a perpetual. However, CoCos can count for no more than 1.5% of AT1 capital and they cannot fulfill the additional capital required for the largest, so-called “globally systemically important” banks (GSIB) (Pennacchi, 2014).
In November 2015 however, the Financial Stability Board has released its Total Loss Absorption Capacity (TLAC) Standard, worded such that CoCos could be used to fill in the additional capital requirements (Chan & Van Wijnbergen, 2016). As the TLAC Standard roughly requires banks to double the loss absorption levels of Basel III and allows them to do so entirely by issuing CoCos, one can expect that the market for these hybrid instruments
9 will continue to grow.
2.3. Risks of CoCo bonds
Boermans (2014) argues that the trigger level that is associated with the “going-concern” capital requirement of 5.125% in the Basel III framework, is relatively low. When capital ratios worsened to a level of under 6%, this indicates that a bank has suffered serious losses. Only CoCo bonds with a relatively high trigger level will therefore contribute to the continuity of a bank. In times of financial distress, these high trigger level CoCos will have a preventive function regarding the equity position when they are triggered in a going-concern situation. This means that CoCos should be triggered well before a possible bank resolution is unavoidable.
One might expect that regulators prefer market based trigger events over book-value triggers, since these are not considered as efficient. However, previous theoretical research shows that market based triggers might not be desirable, as they have characteristics to which they can be considered as risky.
With CoCos that are based on market value triggers, a natural hedge for investors might be to push down the stock price by short selling the equity of banks that issue CoCos. Investors then try to profit form the resulting dilution of the bank’s stock following the CoCo bond conversion, which is triggered by the drop of the stock prices (Goodhart, 2010). This self-generated decline in stock prices is referred to as the ‘death spiral’. A conversion of contingent capital with a market trigger that is punitive to equity holders, may introduce instability because it creates multiple equilibria (Sundaresen & Wang, 2010). The holders of CoCo bonds and the holders of equity have precisely the opposite motives in multiple equilibria. This may lead to manipulation of market prices when the stock price approaches the trigger level of a CoCo issuing bank.
In addition, Koziol & Lawrenz (2012) demonstrate that equity holders have incentives to take excessive risks as CoCos distort risk-taking incentives. Although investors anticipate distorted incentives and demand a corresponding higher premium, investing in CoCos still can be value maximizing for equity holders and therefore form an optimal financing choice. According to Koziol & Lawrenz (2012) the earlier a bank is forced into financial
reorganization because of higher trigger levels in AT1 instruments, the more severe is the negative effect for the equity value. Stricter financial constraints result in incentives for an individual banking firm to avoid excessive risk taking. Risk-shifting incentives always increase relative to ordinary bonds as long as equity holders get to keep some shares after the conversion of AT1 instruments.
These results depend crucially on the assumption that the conversion of AT1 instruments coincides with the bankruptcy trigger. If the decrease in asset value is strong
10 enough to trigger bankruptcy of an individual banking firm, replacing some of the debt by CoCos will obviously leave shareholders better off. In a case with equal debt and an equal decline in asset value, they would have lost everything. In addition, Berg & Kaserer (2015) argue that risk shifting arises as wealth transfers from CoCo bond holders to equity holders increase. They show that the price at which the CoCo conversion takes place has a direct impact on the magnitude and sign of the wealth transfers.
When a CoCo converts, reducing its leverage will raise the residual value of the bank. This is the purpose for which regulators allowed AT1 instruments to be part of banks their CET1 capital. A higher residual equity value may however under many CoCo bond designs that are issued in practice accrue to the old shareholders. Thereby the CoCo
holders will be partially or completely wiped out before equity is reached. The risk decision of the shareholders allows these investors to make their desired event more likely, whether this is conversion or not. According to Chan & Van Wijnbergen (2016) higher levels of risk
reduce the wealth transfer conditional on conversion. Therefore, with the design of CoCos it is not enough to focus only on the probability of wealth transfers taking place when the risk level changes. The overall impact of CoCos on risk shifting incentives is the net effect of two opposing forces: the rise in conversion probability and the fall in the wealth transfer.
According to Chan & Van Wijnbergen (2016) the first effect dominates the second when initial risk levels or leverage ratio’s are sufficiently high. In other words, these are the circumstances that should alarm regulators.
2.4. CoCos and banks’ risk-taking behavior
There are a few existing papers that have examined how issuing CoCo bonds might affect banks’ risk-taking behavior. Martynova & Perotti (2015) argue for instance that CoCos that trigger when a bank is still solvent will reduce banks’ risk-shifting incentives.
Chan & Van Wijnbergen (2016) argue that whenever principal write-down CoCos are in the capital structure, the risk levels chosen by the banks are higher than what they would have chosen when the same amount of subordinated debt would have been. A higher write-down percentage leads to a higher wealth transfer to banks’ equity holders upon conversion, thereby giving them a larger interest in the actual occurrence of such an event. Principal write-down CoCo bonds therefore do not sufficiently increase the loss absorption capacity of a bank, compared to a case in which the same amount of subordinated debt would have been issued. Besides, the principle write-down CoCos increase risk taking incentives and so they certainly do not provide enough incentives for the banks to not let the conversion happen, they actually increase risk ex-ante for given levels of leverage (Chan & Van Wijnbergen, 2016). The authors even argue that principal write-down CoCos should not qualify as AT1 capital.
11 In the case of CoCo bonds that convert to equity however, the incentive turns
negative for a sufficiently dilutive CoCo because the wealth transfer itself becomes negative while shareholders in aggregate obtain a higher residual equity upon conversion (Chan & Van Wijnbergen, 2016). When a CoCo bond is sufficiently dilutive, the existing equity holders and the CoCo holders that become equity holders due to the conversion must share the total residual value. This leaves the existing shareholders worse off than before the conversion. Chan & Van Wijnbergen (2016) argue that the risk level that is chosen by a bank, will be lower when sufficiently dilutive CoCos are compared to the case where the same amount of subordinated debt would have been issued.
2.5. Systemic risks related to triggering CoCos
Since the short selling of the underlying stock is a natural hedge for CoCo bond holders, this could lead to a ‘death spiral’ and a systemic crisis (Maes & Schoutens, 2012). Triggering one bank’s CoCo bonds could lead to speculation about other banks assets. Due to high
correlation, this increases the probability of triggering their contingent capital during a systemic crisis. The main results of the research performed by Koziol & Lawrenz (2012) show that although CoCo bonds can be Pareto-optimal from the bank’s perspective, they have the potential to substantially increase a bank’s probability of financial distress as well as expected proportional distress costs.
In addition, Chan and Van Wijnbergen (2014) argue that when an unanticipated decline in asset returns lead to a CoCo conversion, this has the immediate effect of raising the probability of a bank run. The trigger-event sends a negative signal to depositors about the expected asset returns of a bank. The conversion of a CoCo bond is meant to absorb losses of an issuing bank and to strengthen the balance sheet in times of financial distress, but in practice a trigger event might lead to the opposite.
2.6. Recent turbulence on financial markets regarding CoCos and hypothesis Since the introduction of renewed regulations in the aftermath of the financial crisis, the practical application of the renewed capital requirements has not always been clear. For example discretionary or PONV triggers, are activated based on supervisors’ judgment on the issuing bank’s solvency prospects. The loss absorption mechanism can be activated if the supervisor believes that such action is necessary to prevent a bank from becoming insolvent (Avdjiev, 2013). PONV triggers allow regulators to repair any unreliability or timeliness of book-value triggers. However, such power creates uncertainty on the timing of the activation of the loss absorption mechanism by supervisors, unless the conditions under which regulators will exercise their power to activate the mechanism are made clear.
12 banks should ensure that they meet all of their capital requirements when calculating the maximum distributive amounts for dividend and coupon payments on Tier 1 capital instruments. This includes bank-specific requirements imposed by regulators (European Banking Authority, 2015). The EBA recommended that regulators should require disclosing all capital requirements, while the European Central Bank (ECB) wants banks to keep the bank-specific Supervisory Review and Evaluation Process (SREP) requirements
confidentially (European Central Bank, 2016).
After the publication of the EBA opinion, concerns have been raised about banks’ ability to make discretionary coupon payments on AT1 CoCos. To date, no CoCo bond has missed a coupon payment or has experienced a trigger event. However, the uncertainty on the regulatory treatment of CoCos has made investors worrying about the ability of
European banks to fulfill future coupon payments on CoCo bonds. This has led to a sell-off early 2016 of these hybrid instruments and other bank related assets. Due to the decline in the value of CoCo bonds and stock prices in the first quarter of this year, Deutsche Bank felt obliged to release a press statement on 8 February 2016 in which the bank declared that the payment capacity was amply sufficient to meet future coupon payments on AT1 instruments. Speculation regarding CoCo bonds which is followed by a sell-off could potentially have harmful effects for banks. According to Boermans (2014) a strongly negative signal on CoCo bonds combined with the lack of knowledge of investors about these hybrid
instruments and their regulatory treatment, could lead a sell-off of a banks’ senior debt and equity. As discussed earlier, short-selling the underlying equity can be a natural hedge for investors of CoCo bonds. Risk-shifting arises as wealth transfers from CoCo bond holders to equity holders take place. The equity sell-off could lead to a ‘death spiral’ according to Maes & Schoutens (2012), which eventually could lead to a trigger event which activates the loss absorption mechanism of a CoCo and this might have a bank run as a result (Chan and Van Wijnbergen (2014).
Early 2016 concerns have been raised regarding whether investors have correctly assessed the risks which are associated with CoCos. It seems to be possible that a missed coupon payment or the occurrence of a trigger event could affect valuations across the asset class. Therefore, this paper will try to examine whether the recent turbulence on financial markets has affected the valuation of CoCo bonds and stock prices of banks. Following the theory of the existing theoretical literature, this is hypothesized: when the prices of CoCo bonds decrease because of a sell-off, the associated equity will show a decrease in prices as well. This is because equity and CoCo bonds are highly related, according to the existing literature.
13 3. Empirical research
This section will start with an explanation on the data set that will be used in the empirical part of this research. In the second subsection, the research method will be described and lastly the hypothesis will be formulated.
3.1. Data set
Since there is a lack of papers that provide empirically based analysis on the CoCo market; this empirical study aims to fill the gap between the theoretical papers and the ongoing CoCo market development. This research is based on a unique and comprehensive set of hand collected data. The data on CoCos is obtained by using information that is in the press releases from banks and analyst reports from bank research desks. The data set consist of 72 CoCos which were issued between 2009 and June 2015 by 27 different European banks. For this research, the daily prices of these CoCos are obtained from Datastream for the period between December 2014 and May 2016. Because only 21 out of 27 banks are listed on the stock markets, the CoCo bonds data is merged with the daily data on stock prices of these listed European banks. Six of the CoCo issuing banks are either nationalized or are considered as a private or cooperative institution, therefore there is no data on equity available for these banking firms. The information on equity is extracted from Datastream as well.
In addition, data on the S&P 500, Eurostoxx Banks and the Bank of America Merrill Lynch Contingent Capital Index (BofAML CoCo Index) are obtained from Datastream as well. As a measure of the risk-free rate, the Euro area 30 year yield for maturity data of the European Central Bank (ECB) Statistical Data Warehouse is used. Since under Basel III all AT1 capital instruments must be perpetuals, this 30 year interest rate for AAA-rated
European government bonds seems to be the appropriate measure for the risk free rate.
3.2. Method
This research is based on the Capital Asset Pricing Model (CAPM), where beta will be estimated as a measure of risk. According to Ranaldo & Eckmann (2004) the CAPM is a good measure the risk for convertible capital, although it is not necessarily the best model to use.
Based on the average daily returns of the hand collected CoCo data set, a beta for the CoCos is obtained by using the formula (Berk & Demarzo, 2014):
𝑅𝑅𝑅𝑅 − 𝑅𝑅𝑅𝑅 = 𝛼𝛼𝑅𝑅 + 𝛽𝛽𝑅𝑅 ∗ [(𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅) − 𝑅𝑅𝑅𝑅] + 𝜀𝜀𝑅𝑅 (𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝑅𝑅𝑅𝑅𝐸𝐸𝐸𝐸 1)
14 equal to the market risk premium, which consists of a beta (βi) and the market return minus
the risk free rate (Rmkt – Rf). As a measure of the market returns and the risk free rates, the
data on the S&P 500 and the average of 30 year Eurozone bonds with an AAA rating will be used. The εi illustrates the error term and αi should not differ from 0 significant. Beta is the
expected percentage change of the CoCo return for a given change of the market return by one percent.
Beta is will be estimated for the hand collected CoCo data set, the associated equity of CoCo issuing banks, the BofAML CoCo Index and the Eurostoxx Banks index. These betas are estimated based on a single regression on the daily returns for each type, relative to the S&P 500 market index. For the data set on CoCo bonds and the associated stock prices, the average of the daily returns is regressed on the daily returns of the S&P 500 to obtain beta.
In this analysis, beta will be measured in different time periods. In December 2015 the EBA published their opinion on the regulatory treatment of AT1 capital. Therefore the first research period will be between June 2015 and December 2015, the second period is between December 2015 and May 2016 and the third research period will cover both of these separate periods. As a robustness check, the betas in three similar periods between December 2014 until November 2015 will be estimated.
To check whether the betas of CoCos and equity have changed after the publication of the EBA opinion in December 2015, the results will be shown in a difference in difference matrix:
Table 1 Difference in difference matrix a, b
Pre period
Post period
Difference
Research group
β0 + β1
β0 + β1 + β2 + β3 β2 + β3
Control group
β0
β0 + β2
β2
Difference
β1
β1 + β3
β3
aβ1 is the average of the research group in the period before a certain event occurred,
relative to the average of the control group in the period before a certain event occurred
bβ2 is the average of the control group in the period after a certain event occurred, relative
to the average of the control group in the period after a certain event occurred
By calculating the differences in the estimated beta for equity and CoCo bonds before and after the publication of the EBA opinion, it will become clear whether there is a change in the
15 trend of these estimations of risk. With this method, it can be checked whether there is a difference in growth rates between CoCo bonds and the associated equity.
A significant increase of beta for CoCo bonds would indicate that investors have underestimated the risks of CoCos. The same applies for the banking firm’s equity. Therefore the hypothesis for the difference in difference check will be:
H0: β2 = β3 = 0 H1: β2 ≠ β1 ≠ 0
The same research method will be applied on the BofAML CoCo Index and the Eurostoxx Banks Index, relative to the S&P 500 Index as a measure for the market return. This can be seen as a first robustness check. Another robustness check will be performed by estimating the betas of CoCos and stock prices of European banks in another time period.
16 4. Results & Analysis
In this section the results of the empirical research will be discussed. The first subsection will show some descriptive statistics of the data set that is being used. An overview of the
estimated betas will be given in the second subsection, which is followed by the results for the T-test. In the results will be interpreted with respect to the main research objective of this paper.
4.1. Descriptive statistics
An overview of the average daily returns of the hand collected CoCo data set in relation to the average daily returns of the S&P 500 market index is given in the scatter plot of figure 3. The first period is from June 2015 until November 2015 and the second period is between December 2015 and May 2016.
Figure 3 Scatter plot of the average daily returns in of the hand collected CoCo data set, compared to the average returns of the S&P 500a
a The First period is from June until November 2015, the second period is between
December 2015 and May 2016.
-0.025 -0.020 -0.015 -0.010 -0.005 0.000 0.005 0.010 0.015 0.020 0.025 -0.05 -0.04 -0.03 -0.02 -0.01 0.01 0.02 0.03 0.04 0.05 Co Co s d ai ly re tu rn s
S&P 500 daily returns
CoCo Dataset daily returns compared to the
S&P 500
Daily Returns CoCo Data Set Period 1 Daily Returns CoCo Data Set Period 2 Linear (Daily Returns CoCo Data Set Period 1) Linear (Daily Returns CoCo Data Set Period 2)
17 The figure shows two best-fit lines for the data on the first period (dotted line) and the
second period (striped line). The best-fitted line for the second period is steeper and lies above the best-fitted line for the first period. A steeper line indicates a higher beta and a more risky investment in the second period when compared to the S&P 500 market return. The closer the slope of the best-fitted line is to 1, the more related the returns of the CoCo bonds are related to the stock index returns.
Since the values of the best-fitted line for the second period are all above the values for the best-fitted line of the first period, one might conclude that the overall beta of CoCos was higher in the second period compared to the first period. This might indicate a
revaluation of the risk of CoCo bonds in the second period, after the EBA opinion was published.
Figure 4 Scatter plots of the daily returns in of equity regarding banks that are in the hand collected CoCo data set, compared to the average returns of the S&P 500a
a The First period is from June until November 2015, the second period is between
December 2015 and May 2016.
-0.065 -0.055 -0.045 -0.035 -0.025 -0.015 -0.005 0.005 0.015 0.025 0.035 0.045 0.055 0.065 -0.05 Eq ui ty d ai ly re tu rn s
S&P 500 daily returns
Equity daily returns compared to the S&P 500
Daily Returns Equity Banks In CoCo Data Set Period 1 Daily Returns Equity Banks In CoCo Data Set Period 2 Linear (Daily Returns Equity Banks In CoCo Data Set Period 1) Linear (Daily Returns Equity Banks In CoCo Data Set Period 2)
18 In figure 4 a scatter plot is shown of the average daily returns of equity regarding the banks that are included in the hand collected CoCo data set, compared to the average daily returns of the S&P 500 market index. The periods are the same as in graph 1, the dotted line is the best-fitted line of period 1 which is from June until November 2015. For the second period the striped line is the best-fitted line. The second period refers to the data on December 2015 to May 2016.
In comparison to figure 3, it is clear that again the best-fitted line for the second period is steeper than the best-fitted line of period 1. This indicates that an investment in banking firm’s equity was more risky in the second period, after the publication of the EBA opinion paper in December 2015. In figure 4 there is no difference in the average location of the best fitted lines. Therefore, there is no indication that the overall beta of banking firms stock has increased.
4.2. Estimated betas
In table 2 the beta estimates of the CoCo data set are shown. The values of beta (between 0.112 and 0.180) implicate that CoCo bonds have a relatively low risk, compared to the S&P 500 stock market. Notice that beta has increased from 0.112 to 0.18 since in December 2015 the EBA published their opinion on the regulatory treatment of AT1 capital.
Table 2 Results of the regression analysis for CoCo bonds in three research periodsa,b
Jun 2015 –
Nov 2015
Dec 2015 –
May 2016
Jun 2015 –
May 2016
β
0.112
(0.021)
0.180
(0.052)
0.148
(0.028)
α
-0.002
(0.000)
0.000
(0.000)
-0.002
(0.003)
a Based on the hand collected data set on CoCo bonds for which the average daily returns
are calculated for the period between June 2015 and May 2016
b Standard errors are between brackets
The results that are shown in table 3 give the estimated betas for the Bank of America Merrill Lynch Contingent Capital Index (BofAML CoCo Index). The beta for the BofAML CoCo Index has increased from 0.116 to 0.193 since the publication of the EBA opinion. Like the results on the CoCo bonds data set in table 1, the beta of the BofAML CoCo Index is relatively low.
19 Again, this means that investing in CoCos seems to have a low risk relative to the S&P 500 stock market.
Table 3 Results of the regression analysis for the BofAML CoCo Index in three research periodsa,b
Jun 2015 –
Nov 2015
Dec 2015 –
May 2016
Jun 2015 –
May 2016
β
0.116
(0.020)
0.193
(0.048)
0.152
(0.025)
α
-0.002
(0.000)
-0.002
(0.000)
-0.002
(0.000)
a Based on the data set on the BofAML CoCo Index for which the average daily returns are
calculated for the period between June 2015 and May 2016
b Standard errors are between brackets
Table 4 provides an overview of the beta that is estimated for the stocks of the 21 European banks that are listed on the stock markets. These banks have issued CoCo bonds and are included in the hand collected data set on CoCo bonds. Just like the estimates of beta for the hand collected CoCo bonds data set, the stock prices of CoCo issuing bonds show an increase in beta from 0.775 to 1.15 since the publication of the EBA opinion in December 2015. The results show that bank equity has a higher beta than CoCo bonds, which implies that an investment in equity is more risky.
Table 4 Results of the beta estimates for stock prices of European banks that are listed on the stock markets and which are included in the hand collected data set on CoCo bondsa,b
Jun 2015 –
Nov 2015
Dec 2015 –
May 2016
Jun 2015 –
May 2016
β
0.775
(0.101)
1.152
(0.156)
0.949
(0.092)
α
-0.001
(0.001)
-0.001
(0.002)
-0.001
(0.000)
aBased on the data of stock prices for European banks that are included in the hand collected data set on CoCo bonds. For each stock the average daily returns are calculated for the period between June 2015 and May 2016
20 In table 5 the estimates of beta are shown for the Eurostoxx Banks Index, relative to the S&P 500 market index. The Eurstoxx banks index consists of European banks that have issued CoCo bonds and banks that did not. Since the publication of the EBA opinion in December 2015, beta has increased from 0.804 to 0.943 in the second research period. Again, the estimates of beta show that an investment in bank equity is relatively risky compared to the betas regarding an investment in CoCo bonds.
Table 5 Results of the regression analysis for the Eurostoxx Banks Index in three research periodsa,b
Jun 2015 –
Nov 2015
Dec 2015 –
May 2016
Jun 2015 –
May 2016
β
0.804
(0.355)
0.943
(0.303)
0.864
(0.023)
α
0.000
(0.004)
-0.003
(0.003)
0.000
(0.003)
a Based on the data set on the Eurostoxx Banks Index for which the average daily returns
are calculated for the period between June 2015 and May 2016
b Standard errors are between brackets
As a robustness check, the same data analysis is performed in a different time period on the CoCo data set, the stock prices of banks that are in the hand collected CoCo data set, the BofAML CoCo Index and the Eurostoxx Banks Index. The results of the regression analysis of the CoCo data set are shown in table 6. For the control period of December 2014 until November 2015, there is no change in the beta of CoCo bonds that is worth mentioning. Again, the low values of beta indicate that CoCo bonds are relatively low-risk investment, compared to the stock market.
21 Table 6 Results of the regression analysis for CoCo bonds in three research periods as a robustness check a,b
Dec 2014 –
May 2015
Jun 2015 –
Nov 2015
Dec 2014 –
Nov 2015
β
0.103
(0.024)
0.104
(0.020)
0.105
(0.015)
α
-0.002
(0.000)
-0.002
(0.000)
-0.002
(0.000)
a Based on the hand collected data set on CoCo bonds for which the average daily returns
are calculated for the period between December 2014 and November 2015
b Standard errors are between brackets
In table 7 the regression results for the BofAML CoCo Index are shown. The regression results show a decrease in the estimated beta from 0.122 to 0.114 over the control period. As the previous results have shown, an investment in CoCo bonds seems to have a relatively low risk compared to the S&P 500 index.
The results of table 6 and table 7 show that the beta of CoCo bonds has not changed as much in the control period as it did in the research period, for which the results can be found in table 2 and 3. This might indicate that the beta of CoCo bonds has increased since in December 2015 the EBA published its opinion on the regulatory treatment of coupon payments on AT1 capital
Table 7 Results of the regression analysis for the BofAML CoCo Index in three research periods as a robustness check a,b
Dec 2014 –
May 2015
Jun 2015 –
Nov 2015
Dec 2014 –
Nov 2015
β
0.122
(0.026)
0.114
(0.020)
0.119
(0.016)
α
-0.001
(0.000)
-0.002
(0.000)
-0.002
(0.000)
a Based on the data set on the BofAML CoCo Index for which the average daily returns are
calculated for the period between June 2015 and May 2016
b Standard errors are between brackets
22 banks that are listed on the stock markets, for the control period. The results show that bank equity has a higher beta than CoCo bonds in the control period, which again implies that an investment in equity is more risky. In the control period, estimated beta has decreased from 0.811 to 0.769. Again, an investment in banking firm’s equity seems to be more risky than an investment in CoCo bonds.
Table 8 Results of the beta estimates for stock prices of European banks which are included in the hand collected data set on CoCo bonds as a robustness checka,b
Dec 2014 –
May 2015
Jun 2015 –
Nov 2015
Dec 2014 –
Nov 2015
β
0.811
(0.106)
0.769
(0.102)
0.788
(0.073)
α
0.000
(0.000)
0.001
(0.001)
0.000
(0.000)
a Based on the data of stock prices for European banks that are included in the hand
collected data set on CoCo bonds. For each stock the average daily returns are calculated for the period between December 2014 and November 2015
b Standard errors are between brackets
The estimates of beta in the control period for the Eurostoxx Banks Index are shown in table 9. There is an increase in beta from 0.780 to 0.850 relative to the S&P 500 market index. As previous results have shown, an investment in equity is more risky compared to an
investment in CoCos.
Table 9 Results of the regression analysis for the Eurostoxx Banks Index in three research periods as a robustness checka,b
Dec 2014 –
May 2015
Jun 2015 –
Nov 2015
Dec 2014 –
Nov 2015
β
0.780
(0.118)
0.850
(0.109)
0.829
(0.080)
α
0.000
(0.000)
-0.001
(0.001)
0.000
(0.000)
a Based on the data set on the Eurostoxx Banks Index for which the average daily returns
are calculated for the period between December 2014 and November 2015
23 As with CoCo bonds, again there is a smaller difference between the betas of equity in the control period (table 8) than in the research period (table 9). This might indicate that since the publication of the EBA opinion, investing in banking firms’ equity became more risky than investing in CoCo bonds.
The data on the estimated betas will be used to create difference in difference matrices in the next subsection.
4.3. Difference in difference analysis
To check whether the betas of CoCos and equity have changed after the publication of the EBA opinion, table 10 gives an overview of the estimated betas for the CoCo data set and related equity. In the difference in difference matrix it is clear that a large difference of 0.377 is measured in the beta of the stock prices of the banking firms that are included in the CoCo data set. Since this increase is much larger than the estimated increase of 0.068 for the beta of the CoCo data set, it can be concluded that investing in equity of banking firms that have issued CoCo bonds became riskier since the publication of the EBA opinion.
Table 10 Difference in difference matrix for the CoCo data set and equity that is associated with the CoCo data set a
June - Dec 2015
Dec 2015 – May 2016
Difference
β Equity CoCo
data set
0.775
1.152
0.377
β CoCo data set 0.112
0.180
0.068
Difference
0.663
0.972
0.309
a Based on the estimated betas of paragraph 4.2.
In table 11 the difference in difference matrix is shown for the estimated betas of the BofAML CoCo Index and the Eurostoxx Banks Index. It is clear that the increase of 0.139 in the estimated beta for the Eurostoxx Banks Index is higher than the estimated increase of 0.077 for the BofAML CoCo Index. Again, this indicates that investing in banking firm’s equity became more risky after the publication of the EBA opinion on the regulatory treatment of AT1 capital.
24 Table 11 Difference in difference matrix for the BofAML CoCo Index and the Eurostoxx Banks Index a
June - Dec 2015
Dec 2015 – May 2016
Difference
β Eurostoxx
Banks Index
0.804
0.943
0.139
β BofAML CoCo
Index
0.116
0.193
0.077
Difference
0.688
0.750
0.062
a Based on the estimated betas of paragraph 4.2.
By comparing the results of table 10 and 11, it must be noticed that the increase in beta for stock prices of banks that are in the hand collected data set (0.377), is much higher than the increase in beta (0.139) of the Eurostoxx Banks Index. This difference might be explained by the difference in composition of the hand collected data set and the Eurostoxx Banks Index. The hand collected data set contains the daily returns on CoCo bonds and stock prices of banks that did issue CoCo bonds. The Eurostoxx Banks Index consists of the largest European banks that did not necessarily issue CoCos. Since we have seen that the beta of CoCo bonds increased after the publication of the EBA opinion, we might conclude that banking equity that is associated with firms that did issue CoCo bonds was clearly more affected by the revaluation on financial markets in the first quarter of 2016. This is consistent with the existing literature that CoCo bonds and banking firms’ equity are highly related.
25 5. Conclusion
This research tries to fill in the gap between the existing theoretical research on CoCo bonds and recent market developments. The publication of the EBA opinion paper in December 2015 on the regulatory treatment of AT1 capital has led to turbulence on financial markets. Investors were worried on the ability of European banks to fulfil future coupon payments on AT1 instruments. This research shows that such a relative small event with respect to the uncertainty about the regulatory treatment of AT1 capital, could lead to a strong reaction on the financial markets.
However, CoCos are designed to increase loss absorption capacity of a bank. They are allowed to be a part of the CET1 capital under the Basel III regulatory framework in order to increase the stability of the banking sector, especially in times of financial distress.
Existing literature argues whether CoCos are a suitable instrument to prevent a systemic crisis. This heavily depends on the design of the hybrid instrument. When badly designed CoCos are part of a banks’ capital structure, this might not increase the loss absorption capacity of a bank sufficiently. There is evidence that when CoCos are part of a banking firms’ capital structure, this might be an incentive for a bank to increase their risk-taking behavior.
In addition, risk shifting incentives among investors might lead to a situation in which they have opposite motives. Since short-selling the underlying equity might be desirable for investors as a natural hedge when CoCo bond holders suffer losses on their investment, a trigger event might be breached. This is also referred to as a ‘death spiral’ that is considered as an undesirable feature of CoCos. The possibility of wealth transfers from CoCo bond holders to equity holders must be considered with the design of CoCos.
Possibly even more important, triggering a CoCo bond leads to speculation on the value of assets of other banks since these are highly correlated. This leads to the increase of systemic risks. These findings are supported by the existing literature in which it is shown that when contingent capital is triggered, the probability of a bank run increases.
The results of the empirical part of this research are consistent with these theoretical propositions. It is shown that the beta of CoCo bonds and banking firms equity have
increased since the EBA has published an opinion on the regulatory treatment of coupon payments on AT1 capital instruments. According to the empirical results in this paper, a relatively small event can trigger a reaction on financial markets that can be explained as a revaluation of banking equity and CoCo bonds. This indicates that CoCo bonds in fact are related to banking firm’s stock prices on the financial markets, which is not desirable from a regulatory perspective.
However, this research is based on a data set that is composed over a relatively small period of time. This is due to a lack of data on CoCos for the longer term, since these
26 hybrid instruments are relatively new. It is therefore recommended to improve and enlarge this research in the future.
Based on the results of this research, it can be argued whether CoCo bonds are a suitable instrument to increase the stability of the banking sector. Since under the TLAC Standards banks are allowed to fulfill additional capital requirements completely by issuing CoCos, one can expect that the issuance of these hybrid instruments will continue to grow in the nearby future. Therefore, further research on CoCo bonds is necessary.
27 References
Avdjiev, S., Bolton, P., Jiang, W., Kartasheva, A., & Bogdanova, B. (2015). Coco bond
issuance and bank funding costs. BIS and Columbia University working paper.
Avdjiev, S., Kartasheva, A., & Bogdanova, B. (2013). CoCos: a primer. BIS Quarterly
Review.
Basel Committee. (2010). Basel III: A global regulatory framework for more resilient banks
and banking systems. Basel Committee on Banking Supervision, Basel.
Berk, J. B., & DeMarzo, P. M. (2014). Corporate finance. Pearson Education.
Boermans, M., Petrescu, S., Vlahu, R. (2014). The future of CoCo bonds. Economisch Statistische Berichten, 4695, 616 – 619.
Chan, S., & Van Wijnbergen, S. (2016). Coco Design, Risk Shifting and Financial Fragility. Chan, S., & Van Wijnbergen, S. (2014). Cocos, contagion and systemic risk.
European Banking Authority, "Opinion of the European Banking Authority on the interaction of Pillar 1, Pillar 2 and combined buffer requirements and restrictions on
distributions", 16 December 2015.
European Central Bank, “SSM SREP Methodology Booklet”, 19 February 2016.
Flannery, M. J. (2002). No pain, no gain? Effecting market discipline via 'reverse convertible
debentures'. Effecting Market Discipline Via 'Reverse Convertible
Debentures' (November 2002).
Goodhart, C. (2010). Are CoCos from cloud cuckoo-land?. Central Banking, 21(1), 29-33. Koziol, C., & Lawrenz, J. (2012). Contingent convertibles. Solving or seeding the next
banking crisis?. Journal of Banking & Finance, 36(1), 90-104.
Maes, S., & Schoutens, W. (2012). Contingent Capital: An In‐Depth Discussion*. Economic
Notes, 41(1‐2), 59-79.
Martynova, N., & Perotti, E. C. (2015). Convertible bonds and bank risk-taking. Pennacchi, G., Vermaelen, T., & Wolff, C. C. (2014). Contingent capital: The case of
COERCs. Journal of Financial and Quantitative Analysis, 49(03), 541-574.
Ranaldo, A., & Eckmann, A. (2004). Convertible Bonds: Characteristics of an Asset Class. UBS Research Paper.
Sundaresan, S., & Wang, Z. (2015). On the design of contingent capital with a market
