An Empirical Investigation into Dynamics of Implicit Government
Guarantees granted to the Too-Big-To-Fail Banks: A Global Case
MSc Economics
University of Amsterdam
Author: Regis M. Rutagarama Student Number: 11376112
Thesis Supervisor: Dr. Ward E. Romp Co-reader: Dr. Christian Stoltenberg
Master’s Thesis December 2017
Statement of Originality
This document is written by Student Regis M. Rutagarama who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document are 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.
Table of Contents
1. Introduction 2. Literature Review 3. Methodology
3.1. Definitions
3.2. The Motive of Quantifying IGG from Senior and Subordinate CDS
3.3. Risk–Neutral Probability of Default 3.3.1. Assumptions of the Hull’s Formula
3.4. Measures of Implicit Government Guarantees (IGG) 4. Data Resources
5. Time-Fixed Effect Regression Analysis 6. Empirical Findings
6.1. The Dynamics of IGG from Descriptive Statistics 6.2. The Variability of IGG
6.3. Risk Impact on IGG
6.4. Too-Big-To-Fail (TBTF) Impact on IGG 6.5. Basel III Implementation Impact on IGG 6.6. Regional Impact on IGG
6.7. Bail-In Implementation Impact on IGG 6.8. Robustness Tests
6.9. Limitations of the study 7. Conclusions
8. References 9. Appendix
An Empirical Investigation into Dynamics of Implicit Government
Guarantees granted to Too-Big-To-Fail Banks: A Global Case.
Abstract
This thesis measures the implicit government guarantee of Too-Big-To-Fail banks using credit default swap (CDS) rates for the period 2009-2017. Empirical results confirm that these banks benefit from a higher implicit government guarantee than their counterparts. The results also reveal that these government guarantees have been decreasing at a slower pace than policy-makers expected.
Keywords: Too-Big-To-Fail banks (TBTF), Systemic risk, Implicit government guarantees, Credit default swaps, Probability of default.
1. Introduction
World’s biggest financial institutions have gained such a size, market importance, and
interconnectedness that if one of them were to fail, according to the Financial Stability Board and IMF's report of October 2009, it would cause a major financial distortion and real economic damage. Thus, those banks are viewed and labelled Too-Big-To-Fail (TBTFs), and leading the public authorities to bail out their creditors in case of failure.
These large implicit guarantees provided to private institutions creates a moral hazard and stimulates financial sector concentrations. In turn, this moral hazard expands to Too-Big-To-Fail problem, in that it becomes a menace to financial stability.
In reality, bailout actions are not only restricted to insured deposits by governments, as also noted by Demirgüç-Kunt et al. (2013). The main reason being that, even though bailouts are expensive, bailout policies are not only politically beneficial; such policies can prevent an ex-post problem, e.g., major market distortion, but they also create an ex-ante distortion by encouraging bad behavior. The outcome of the above paradox is that it becomes rather bizarre than the norm to end expensive government bailouts policies. It becomes unwelcome in public's eyes to end bailouts policies, and the governments become too lenient. Thus, becoming almost impossible to end entirely implicit government subsidies.
Most financial crisis and bank's failures are preceded by common indicators: financial
deregulations, high levels of leverage, and a bubble in the housing market as noted by Allen et al. (2009). Yet, it seems that lessons are not learned, and it seems like a Sisyphean task on its own. To put an end or at least limit those events, and reduce moral hazards in the financial sector, several guidelines have been suggested. Among them: Higher capital requirements which are included in Basel III requirements; Bail-in instruments which consist of writing down and converting equity and debt, placing the burden on the bank’s creditors and shareholders, instead on the public.
In this thesis, I will investigate whether the official sector's efforts to end the Too Big to Fail subsidies have borne fruit. I will try to answer the following specific question: Has the introduction of bail-in requirements, resolution plans and resolution instruments reduced the implicit government subsidies that large systemic banks benefit from?
The motives of this study are that: First, as far as I can tell, no one, barring Zhao (2014) study in a European case, has investigated different types of credit default swaps to extract embedded information and quantify Implicit government guarantees to financial institutions in a global case. I use Lei Zhao (2014) as my companion paper in this study. My contribution is to extend this cutting-edge methodology to global data; The merits of this method relative to earlier study goes to Lei Zhao.
Second, unlike other studies done before, which relied on equities and bonds market to measure the implicit government subsidies to systemically important financial institutions, this thesis uses CDS spreads to measure Implicit Government Guarantees (IGG). As Blanco et al. (2005) noted, credit default swaps are a more precise measure of credit risk as opposed to fixed-asset market, particularly in credit crunch periods as highlighted by Acharya et al. (2007). The reason given by Blanco et al. (2005), is that the credit risk price discovery is better observed in CDS market as opposed to the bond market.
Price discovery is defined as the process of finding the price of an asset in the market through the relationship between demand and supply of that asset. The explanation of Blanco et al. (2005) view is that: First, in the bond market, investors buy bonds and simply hold them to maturity. However, the liquidity in the secondary market is very weak, and hedging for credit risk of risky bonds is even harder and usually very short-term when found. On the other hand, credit
derivatives, more importantly, CDS, allow CDS investors to hedge against risk over a long-term period at a defined price. Thus, there appears to be a mismatch in the time horizon of the bonds asset and the period of credit risk protection in the fixed-asset market.
Second reason stated by Blanco et al. (2005) is that the credit derivatives market is still small, but very dynamic compare to bond market which is larger and better established.
According to the Bank for International Settlements, the outstanding CDS market size is
estimated at 11.8 trillion USD compare to 100 trillion USD in global bond market size as of the end of June 2016.
In addition, according to Blanco et al. (2005), CDS market has an advantage of being the easiest market to trade credit risk compared to bond market; In that, the CDS market does not have short-sale constraints. Unlikely in the bond market, investors can trade large quantities of credit risk in CDS market.
Third, this thesis investigates whether the new regulations, such as Basel III and Bail-in instrument influenced the implicit government guarantees to TBTFs. Lastly, it is also worth noting that this thesis covers the aftermath of the 2007 subprime crisis and the recent European sovereign debt crisis which started from the end of 2010 to mid-2013, and I have extended the sample period to June 2017.
In the analysis section of this thesis, I use senior CDS spreads, and subordinate CDS spreads to measure and analyse the dynamics of IGG. My analysis shows that the IGG considerably increased after the subprime crisis in 2007. Average IGG in North America, Europe, Asia, and Australia increased roughly 400 basis points in the aftermath of the subprime crisis. Even though IGG differ across banks over time, it is important to note that, a few banks continually benefit from implicit government guarantees.
Additionally, the implicit government guarantees toward larger financial institutions are more sensitive to default risk. That is TBTFs benefit more from the implicit subsidies when the risk of default increases. Interestingly, Basel III implementation, which was regarded as the main tool to untangle the Gordian knot of financial instability, fails to deliver; at least for now. Until June 2017, Basel III implementation failed to reduce IGG.
The rest of my thesis is organized in following way: In section two, I review the literature and section three describe the methodology I use in this thesis.
Section four covers the data description. The empirical results are discussed in section five, and the thesis ends with the conclusion and appendix.
2. Literature Review
Three generations of empirical analysis of implicit government subsidies have appeared in last few decades. The first-generation focused at studying whether explicit or implicit government guarantees affected the risk sensitivity of subordinate debt spreads.
In a similar approach to first generation studies, an empirical analysis by Sironi et al. (2000), investigated whether bank's subordinate debt spreads, over the similar maturity risk-free bonds, reflect the buyers' awareness of the issuer's credit risk. They conduct a cross-sectional regression, with a regression function of six main variables: (1) Explicit or Implicit government guarantees, (2) the economic condition of the issuing bank, (3) the time to maturity of the issue, (4) the amount issued, (5) the time of the issue, and (6) the domination of currency.
This empirical study was done to European banks for the period of 1991 – 2000. Testing the impact of implicit government guarantees on subordinate debt spreads outside the U.S market is important not only as a robustness check of existing studies. On one hand, domestic bank regulations are incompatible with a more sophisticated and global presence of large banks. On the other hand, any attempt by policy makers in a single given country to regulate their banks would violate the principle of Basel Committee that promotes the Level-Playing-Field.
In this empirical study, in two sub-sample of the same size, they found that investors were not sensitive to bank specific risk, but afterward, the opposite occurred. Investors become risk-averse when faced different risk profile among European banks; Suggesting that Too-Big-To-Fail policies were present in the first part of the decade and become insignificant in the second part. That said, the results also reveal that European Public Banks enjoyed significant Implicit Government Guarantees from a lower cost of debt issuing, and the value of these guarantees have been increasing over time.
In a similar spirit to first generation studies, an empirical study, Calomiris et al. (1999), results show an adverse effect of current banks safety nets on the stability of financial system.
Their results show that that the government's lender of last resort policies gives banks incentive-to undertake excessive risk, particularly when the asset values are depressed. The study results also emphasize the importance of safety net policy and show how the safety net system must be well designed to maximize social welfare subject to both political and economic constraints.
Calomiris et al. (1999) safety net mechanism proposal, is that banks should maintain a minimum fraction of subordinated debt. The credit premium paid on this debt would bring much-needed market discipline in the financial system. In the sense that banks would be forced to finance themselves with the tiniest portion of the uninsured low-risk debt.
The other advantage of this safety net mechanism is that this policy instrument would protect banks from sudden runs by imposing only a gradual roll over of subordinated debt. A
combination of government deposit insurance and interbank lending would limit endogenous bank distress.
On the other hand, despite increasing measure to enforce market discipline, the expectation is that the biggest lender in the banking system are to be bailed out once they fail, therefore
enjoying implicit subsidies. Demirgüç-Kunt et al. (2013) empirical study also shows that there is evidence that the share prices of systemically important banks were heavily discounted in
countries with large fiscal deficits which highlight the problem and evidence that these large banks have become "too big to save."
Demirgüç-Kunt et al. (2013) findings reveal that systemically important banks have the incentive to reduce their size, particularly in fiscally distressed countries. This paper also shows that there is a private incentive trend to downsize in the wake of "too big to save" evidence, even in the absence of tax and financial regulation. Additional taxation and financial regulation as advocated by the April 2010 IMF report could robust this trend.
The second-generation studies aimed at providing estimates of the actual value of implicit
guarantees to financial industry during the financial crisis, the types of financial institutions who- benefited from these government guarantees, and the allocation between creditors and
shareholders.
In a study done by Tsesmelidakis et al. (2012), they investigate how the benefits received from reduced capital cost by bank's creditors and shareholders affect their behavior during the crisis. In this analysis, they use bonds spreads from primary market and secondary market, for U.S banks, in a sample of 74 banks, parent firms, and their subsidiaries. The value of implicit subsidies is estimated using two methodologies.
The first method assumes that if a bank pays a periodic interest rate on its debt, it is reasonable to assume that a bank hold equivalent saving in coupon payments for the entire debt levels, thus multiplying the outsourcing debt by the rate differential gives the value of the subsidies. The potential flaw in this assumption is that this method can lead to either over or under-estimation of the amount of the grants, in that the debt of a bank can only be observed on its balance sheet, which is a stock figure. Thus, this method cannot differentiate when the debt was given and at what Too-Big-To-Fail premium the debt was provided.
The second methodology, erase the above issues, by reevaluating the debt level in present values and holding everything else constant. The subsidy came from the difference between the market value and calculated the present value of debt spreads, using CDS data which includes one-year, five-year, and ten-year maturities.
The potential flaw in this methodology is that in 2012 the CDS data were scarce. The analysis, that was done in Tsesmelidakis et al. (2012), did not include a separate estimation of funding advantage from actual data of junior CDS spreads. Tsesmelidakis et al. (2012) study, only extended to subordinate CDS spreads, and this might underestimate the real implicit subsidies.
A series of pooled regressions are used in the above study, with a panel data, and they cluster residuals to avoid inflating T-statistics due to correlations between the residuals similar to what is done in Petersen et al. (2009).
The results suggest that debt holders' benefits from implicit government guarantees were larger, almost double, than those of shareholders during the financial crisis. The total of implicit government subsidies accrued to both shareholders and creditors was 365 U.S dollar billions, which is an astonishing number, but perhaps only part of the truth since these results rely on only 30% of the entire debt issued at that time in U.S.
Although early findings did not support this statement, Fraser and McCormack (1978) empirical analysis shows that the size of debt issued matters. Ceteris paribus, an increase in debt issuing or an increase in debt size reduce the required risk premium, which is consistent with the idea of too-big-to fail.
Avery et al. (1988) study also confirm Fraser and McCormack findings. In their study, one aspect of subordinate debts that make them an attractive instrument is their ability to impose a market discipline to banks due to that holder of subordinated securities are subject to large risk and losses in the event of bank failure.
The subordinated debt holders may impose market discipline on banks unlike uninsured
depositors since uninsured depositors can withdraw their funds once a bank's problem becomes apparent in the market. Thus, subordinated debt holders are subject to larger risk and losses which keep these investors on their toes by monitoring bank's activities very closely.
The other quite interesting finding in Calomiris et al. (1999) study is that during the 2007
financial crisis, financial institutions voluntarily reversed their funding strategy. The results show that banks used the upper hand of their implied Too-Big-To-Fail status by taking short-term debt at a fixed rate, as opposed to the usual strategy of uninsured companies to take a long-term position at a fixed rate during the uncertain period.
The third generation addressed information embedded in derivatives asset, such as credit default swaps, to estimate implicit government guarantees. The fact that credit default swaps are more standardized than bonds and explicitly reveal the credit risk associated with the underlying asset- makes them unique and gives a cutting-edge methodology that use CDS spreads to measure implicit guarantees.
In a study done by Lei Zhao in 2014, which is the companion paper of my study in this thesis, Zhao uses different segments of CDS to estimate implicit government subsidies in European financial institutions. Zhao also investigates whether the magnitude of such guarantees depends on the type of financial institution, precisely Zhao considers the difference between bank and insurance companies in Eurozone. Finally, Zhao uses Granger causality tests to study the possible cumulative effect of sovereign default risk and implicit subsidies. The methodology employed is explained in detail in my study; The results of Zhao's study reveals that the implicit government subsidies increased considerably in the aftermath of the recent financial crisis of 2008; the discount on credit default swaps was 89 basis points during the recent European debt crisis.
More interestingly, Zhao's study results show that even if the implicit government subsidies vary across firms, some European Financial Institutions benefit from these implicit guarantees
regularly; Suggesting that the effect of Too-Big-To-Fail status is present in European financial system.
Finally, Zhao's findings show that there is non-causality outcome between the implicit government guarantees and sovereign default risk in Eurozone for the period of
2005 – 2013. This is because: on the one hand, as the sovereign default risk increases, the implicit government guarantees decrease. On the other hand, as the sovereign default risk
increases, the default risk in financial sector increases, as large banks hold a high number of their government's bonds on their balance sheet, which leads to higher implicit government
guarantees. Suggesting that the two effects offset each other; Thus, non-causality outcome between the implicit government guarantees and sovereign default risk in Eurozone for that period.
The potential flaw of this study is the fact that CDS data are scarce, and mostly available only for financial institutions, at least for now; Attempting to measure the aggregate implicit guarantees-for both financial and insurance firms might result in a wrong and inaccurate estimation. In contrast, to avoid the above problems, I tried in my study to concentrate on the dynamics of implicit government subsidies and only for financial institutions.
3. Methodology
In this study, I use Zhao (2014) as my companion paper. The first part of this section, I demonstrate how the two types of CDS, senior CDS, and subordinate CDS, can be used to measure implicit government guarantees to financial institutions(IGG). In the second part, I illustrate the methods to measure the key element of IGG: the probability of default from which the IGG is formed.
Before proceeding, let us remind ourselves the meaning of technical terms that are mainly used in this section and how they work; Credit to Markit Group, the definitions of technical terms provided in following section are from Markit Group – CDS glossary.
3.1. Definitions
Swap – An agreement between two parties to exchange future cash flows or credit risk. (Markit Group CDS glossary)
Credit Derivative – A form of derivative transaction, for example, a Credit Default Swap, designed to transfer credit risk from one party to another. (Markit Group - CDS glossary)
Credit Default Swap (CDS) – A credit derivative transaction in which two parties enter into an agreement, whereby one party (the Protection Buyer) pays the other party (the Protection Seller) periodic payments for the specified life of the agreement. The Protection Seller makes no
If such an event occurs, it triggers the Protection Seller's settlement obligation, which can be either cash or physical. (Markit Group - CDS glossary)
Tier – Refers to one of four levels of debt in the capital structure of the reference entities. Each tier represents a different degree of seniority or preference in liquidation or bankruptcy. There will be varying levels of CDS protection for each of the tiers. (Markit Group - CDS glossary)
• Senior • Subordinated • Junior
• Preferred
CDS Spread – Also called a premium. The amount paid by the Protection Buyer to the Protection Seller, typically denominated in basis points and paid quarterly. For example, if the spread for The Widget Company is 200 basis points, the Protection Buyer will pay the Protection Seller 200 basis points multiplied by the notional of the trade annually (typically paid quarterly, on an actual number of days per period/360 basis). (Markit Group - CDS glossary)
3.2. The Motive of Quantifying IGG from Senior CDS and Subordinate CDS
Due to the policy response to the recent financial crisis, the public has become aware that banks debt has a special treatment, especially banks that are systemically important. This special-status of banks' debt prompt the policy maker to bail out systemically important financial institutions in case of failure. The reason being that, even if the government or policy makers deny that such guarantees exist ex-ante, thus "implicit guarantees" term, as previously seen, ex-post events might make it optimal to provide such assurances. In fact, the market believes so, when it comes to systemically important banks; This has also been confirmed by Santos et al. (2014).
As any other guarantees, the implicit government guarantees also have economic value. The implicit subsidies benefited by senior creditors proceeds in lower senior CDS prices than it would have been without such subsidies.
Using a widely-used assumption in financial studies, i.e., Bernanke et al. (2013) approach, let us also assume that subordinate CDS spreads do not have such subsidies, the rebate can be extracted from information embedded in two types of CDS.
The thought process from the above approach is that, if only subordinated CDS gets wiped out in the case of a bank default, subordinated CDS prices take into consideration only the probability-of default. On the other hand, senior CDS prices take into account both the fact that the
government may repay senior CDS investors and the likelihood of default.
Having said that, if we assume that for a given bank, the loss in case of default always stays constant, Black et al. (2013), it is reasonable also to assume that the probability of default would remain constant. If, however, the probabilities of default proved to be different, within the same bank, it follows that the difference is the implicit government subsidies.
Through the information extracted from subordinate CDS and senior CDS spreads within the same bank, I calculate probabilities of default, which reveal the implicit guarantees enjoyed by senior creditors through their differences.
The CDS prices are mainly affected by two key elements: the risk-neutral probability of default and the loss-given-default. I assume that the loss given default stays constant for the same company. In the studies of financial institutions, it is common to assume a fixed loss given default (LGD), as defaults occur very rarely in the financial sector. This assumption was also confirmed by Weber et al. (2012); They state that it is almost impossible to calculate a stochastic loss given default from historical data.
As a result, a fixed loss given default coupled with the assumption that there are no subsidies for senior creditors, the risk-free probability of default embedded in senior CDS and subordinate CDS should match.
Likewise, Weber et al. (2012) state that even though subordinate CDS had on average lower prices on markets before the financial crisis, the difference becomes non-existent during –
the crisis. It is important to note here that this outcome was in a study done prior 2008 crisis. I have extended their studies until 2017. My results are consistent with their findings; My subordinate CDS spreads are as liquid as senior CDSs. In general, the liquidity between two spreads is almost the same; the reason being that the gap between senior and subordinate CDS is stable. In closing, in case there is any difference between the risk-neutral probability of defaults extracted from senior CDS spreads, and subordinate CDS spreads, should come from the IGG.
The potential flows are that as Longstaff et al. (1995) states, the loss given default, even if assumed to be constant, can also be stochastic. On the other hand, as Longstaff et al. (2005) found that, even if theoretically liquidity premium could be different, the nature of credit default spread contracts makes their prices less sensitive to liquidity premium. Both findings contradict each other.
3.3. Risk–Neutral Probability of Default
To calculate IGG for each bank-month, I followed the steps found in Zhao (2014). In this study, I use John Hull's formula, found in Hull et al. (2008) and Hull’s book, (Hull, 2008), to calculate the probability of default. This simplified version of the Zhao's formula which uses a more advanced approach that requires additional complex mathematical features and at this level of my study, it would shift the focus on the technical side instead of the main aspects of it.
The advantage of using Hull's probability of default formula instead Zhao's formula is that it provides insights more efficiently, at this level of study, than it would have been with Zhao's method; thus, providing a pedagogical utility.
The drawback, on the other hand, is that there might be a loss of accuracy in the process than it would have been with a more sophisticated and advanced mathematical formula. Nevertheless, my hope is that the simplified version of the formula still captures the main features of the probability of default estimation and a straightforward structure.
3.3.1. Assumptions of the Hull's Formula
John Hull assumes that the interest rate, the recovery rate, and default probabilities are
independent. It is important to note here that these assumptions might not always be accurate. For instance, if interest rate increases firms may experience financial distress, which in turn may- increase firms' probability of default. The positive relationship between probability of default and interest rate also may decrease CDS spread; As the higher the likelihood of default,
the higher the discount rate to the payoff of the underlying asset will be, which in turn reduce the CDS spread.
On the other hand, the higher probability of default tends to reduce the value of the underlying asset on the market which in turn increases its CDS Spread, as it becomes expensive to insure an asset with a high probability of default. Note that the above effects offset each other, and the negative impact of Hull's assumptions are minimal according to Moody's investor's service statistics referenced in this thesis.
First, the bank-month CDS spreads were converted into percentages and then passed through the implied probability of default formula found in Hull et al. 2008 and their book Options, Futures, and other Derivatives (Hull, 2008). They provide an equation for the risk-neutral probability of default implied in CDS.
There are a few different ways you can measure the probability of default; However, according to the seventh edition of John Hull’s book: Option Futures and Other Derivatives, a more exact calculation of the probability of default is to use the hazard rate:
"($) = '($) 1 − *
Where "($) is the average default intensity at time t or the hazard rate at time t, S is the spread of the underlying asset or debt spread. This is associated with the probability of default + from time $ to time $ + ℎ:
+($, $ + ℎ) = "($)ℎ + /(ℎ) Thus: + 0, 1 = 3 1 − + 0, 1 + $, $ + 2$ 4 = 3"($)(1 − + 0, 1 )2$ 4
Where the first part of the integral function has not occurred yet, and the second part of the integral function happens the following period. Which means that the probability of default + satisfy:
2+(0, $)
2$ = "($)(1 − + 0, $ )
As assumed in this thesis if the CDS is constant then the average default intensity at time t or the hazard rate, " is also constant and the CDS probability of default would be:
+ 0, $ = 1 − exp −'$ 1 − *
Thus:
PD
:,;<= 1 − exp
=>?,@A ∗CD=E
(1)
Where R is the recovery rate and
S
:,;< is the CDS spread for a bank i at time t, and it can either be subordinate or senior, j, CDS spread, assuming constant loss given default (LGD) within a same bank.PD:,;< is the probability of default for a bank i at time t extracted from a subordinate or senior CDS spread j; j can either be SEN for senior CDS spreads or SUB for subordinate CDS spread. Furthermore, I use LGDwhich stands for the loss given default, and constant.
In my calculation, I set the loss given default, LGD, at 0.6 following Black et al. 2013 ex-ante measures, which varies from 0.57 to 0.64 during seven years prior to 2013. This is also consistent with Weber et al. 2012’s loss given default estimation, which ranges from 0.57 to 0.61. This estimation is based on Moody’s Investor services of 2012 which sate the recovery rate at 0.4 for the period of 1982 to 2012.
3.4. Measures of Implicit Government Guarantees (IGG)
In measuring the implicit guarantees (IGG), I assume the absence of implicit subsidies rebate in the risk-neutral probability of default extracted from subordinate CDS spread while the rebate exists in the probability of default extracted from senior CDS. It is a fundamental assumption in the sense that, only senior CDS prices take into account the fact that the government support senior investors in case of bank failure.
I use here probability of default (PD) from senior CDS, as I assume that the percentage loss is subsidized only for senior debt holders. I also use here probability of default (PD) from subordinate CDS, as I assume that the percentage loss is not subsidized for subordinate debt holders.
If the loss given default is same or constant for a given bank, it is rational to assume the same probability of default across both types of debts. If these probabilities differ, the difference is the implicit government subsidies.
IJJ
K,L= M
LGD+N
K,LODPQR− +N
K,LOD>ST(2)
In my calculation, as I mentioned previously I set the loss given default, LGD, at 0.6 following Black et al. 2013 ex-ante measures, which varies from 0.57 to 0.64 during seven years prior to 2013. This is also consistent with Weber et al. 2012’s loss given default estimation, which ranges from 0.57 to 0.61.
This estimation is based on Moody’s Investor services of 2012 which sate the recovery rate at 0.4 for the period of 1982 to 2012. Since then this has been referenced in most risk-related academic literature such as the seventh edition of John Hull’s book, Option Futures and Other
Derivatives, (Hull, 2008, sect. 22.4).
4. Data Resources
My data consists of two types of credit default swaps (CDS) spreads, namely senior and
subordinate CDS. I started with the 100 largest financial institutions by assets in the world using Thomson Reuters Eikon. I then employed Black et al. (2013) framework which consists of eliminating banks without 24 valid consecutive monthly credit default swaps spreads.
At the end of that procedure, I had 43 banks left as my sample, from which 20 of them are systemically important financial institutions, which are viewed and labelled Too-Big-To-Fail (TBTFs) according to the recent list published every year in November by Basel Committee on Banking Supervision (BCBS) in consultation with the Financial Stability Board (FSB). I gathered monthly CDS spreads from Thomson Reuters Eikon and DataStream for the period of January 1st, 2009 to June 1st, 2017.
5. Time-Fixed Effect Regression Analysis
In this study, I use the following time-series regression to estimate whether the Basel III
implementation, introduction of the Bail-in instrument, and being a TBTF bank had an impact on IGG.
IGG
:,;= β
DRisk
:,;=D+ β
[TBTFdummy
K+ β
cBaselIIdummy
K,L+ β
fBaildummy
K,L+ β
gEuropedummy
Kk
+ β
lAsiadummy
K+β
nAustraliadummy
K+ β
pRisk
:,;=D∗ TBTFdummy
K+ β
qBaselIIdummy
K,L∗ TBTFdummy
KIn equation (3), t subscript denotes time, which is monthly in my study, and i subscript indicates the bank. I also use probabilities of default extracted from subordinate CDS spreads as risk variable (SubPD).
TBTF dummy variables are depicted from a list of Systemically Important Financial Institutions published every year in November by the Basel Committee on Banking Supervision (BCBS) in consultation with the Financial Stability Board (FSB).
Basel III dummy variable is also used, and it is equal to 1 if the bank is located in a country which has implemented Basel III since its announcement in November 2010, and 0 otherwise. Bail-in instrument dummy variable is also used to study whether the introduction of the Bail-in instrument has had an impact on IGG. The summary statistics of the regression of equation (3) are reported in the table (1).
6. Empirical Findings 6.1. The Dynamics of IGG from Descriptive Statistics
The implicit government subsidies extracted from senior credit default and subordinate credit default swaps can be interpreted as an advantage enjoyed by large banks when raising funding. Figure 1: Scatter Plot Figure 5: Mean of IGG Determinants
0 .2 .4 .6 IGG
Jan. 2009 June 2010 Nov. 2011 April 2013 Sept. 2014 June 2017
Date
As seen in Figure 1 and Figure 5, the IGG Scatter Plots and the probability of defaults
respectively, show that IGG, the difference between the two lines in figure 5, has not changed in general terms. Looking at those two graphs, you can conclude that since January 2009 until June 2017, IGG are the same.The other element is that from Figure 5, the probabilities of default with and without government support are correlated, the difference between the two being the IGG.
Figure 4: IGG by Region
As seen in Figure 4, when I plot the average IGG by region namely: North America, Europe, Asia, and Australia regions, it shows that IGG varies across regions. However, the figure shows that large Asian banks and European banks currently enjoy more funding cost advantage
provided by their governments.
Perhaps this may be explained by the European debt crisis, which is the only crisis that occurred during my sample period. On the other hand, this may be explained by how in general European countries are more supportive than the other regions when it comes to social or economic policies. .05 .1 .15 .2 .25 IGG
Jan. 2009 June 2010 Nov. 2011 April 2013 Sept. 2014 June 2017
Date2
IGG_US IGG_EU
IGG_ASIA IGG_AU
Australia currently has lower implicit subsidies toward large banks compare to its peers; This can be explained by the fact that Australia introduced tougher bank regulation before the others toward the end of 1999.
Figure 2: Average IGG
The other insight from graphs, it is the plot of the average IGG in Figure 2. It can be seen from the illustration how the implicit subsidies are positively related to the financial crisis;
IGG increases during financial crisis such the start of 2009, subprime crisis, and decreases in normal times when the financial sector is not distressed.
Figure 2 also shows that the average IGG at the start of my sample period in January 2009 has been decreased when compared to the average IGG at the end of my sample period, June 2017, roughly 800 basis points.
.1 .12 .14 .16 .18 me an _I G G
Jan. 2009 June 2010 Nov. 2011 April 2013 Sept. 2014 June 2017
Date
6.2. The Variability of IGG Figure 6: Basel III Box-Plot
Introducing Basel III reduced the variability of IGG in that the spread before Basel III
implementation was between 0 and 0.3 but that spread shrank to the spread between 0.04 and 0.18 after Basel III implementation without the outliers. Secondly by looking at Figure 6, one can see that IGG stays symmetric either before or after Basel III implementation.
Additionally, from Figure 6, the median IGG before Basel III implementation was about 0.12 and that decreased to 0.1 after the implementation of Basel III.
0 .2 .4 .6 0 1 IGG Graphs by Basel
Figure 7: Bail-In Instrument Box Plot
With the introduction of the Bail-in instrument, the IGG variability seems to express same behaviors as when Basel III was implemented. The spreads were around 0 and 0.3 before the introduction of the bail-in instrument and shrank to the spread ranging from 0.04 and 0.18.
The median also expresses same characteristics as of Basel III implementation effects; Before the Bail-in instrument IGG median was around 0.12 and declined to 0.1 after the implementation of the bail-in instrument.
0 .2 .4 .6 0 1 IGG Graphs by Bail_in
Bail-in Instrument
Table 1: Regression Results for the Model (with Robust Standard Errors).
(1) (2) (3) (4)
DV: IGG DV: IGG
VARIABLES DV: IGG robust SE DV: IGG robust SE
Lag Risk 0.254*** 0.254*** 0.266*** 0.266*** (0.00527) (0.00731) (0.00584) (0.00948) TBTF -0.0129*** -0.0129*** 0.0101** 0.0101** (0.00146) (0.00123) (0.00484) (0.00503) Basel III 0.00131 0.00131 -0.00425 -0.00425 (0.00772) (0.0104) (0.00769) (0.00971) Bail_in 0.0119*** 0.0119*** 0.0336*** 0.0336*** (0.00265) (0.00231) (0.00346) (0.00308) Europe 0.0213*** 0.0213*** 0.00925*** 0.00925*** (0.00262) (0.00191) (0.00284) (0.00185) Asia 0.0419*** 0.0419*** 0.0323*** 0.0323*** (0.00311) (0.00356) (0.00321) (0.00328) Australia 0.00421 0.00421 -0.0237*** -0.0237*** (0.00377) (0.00346) (0.00475) (0.00403) Inter. Basel_TBTF 0.0329*** 0.0329*** (0.00368) (0.00267) Inter. Bail_TBTF -0.0422*** -0.0422*** (0.00422) (0.00338) Inter. Risk_TBTF -0.0472*** -0.0472*** (0.00858) (0.0104) Observations 4,343 4,343 4,343 4,343 R-squared 0.915 0.915 0.918 0.918
Time FE YES YES YES YES
Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1
Table 2: Marginal Effects on TBTF
Bank Type Basel III Margin
Small Banks Before 0.1404851***
Small Banks After 0.1362345***
TBTF Banks Before 0.1133063***
TBTF Banks After 0.1419716***
The marginal effects on TBTF banks, are seen in third and fourth rows of Table 2. The third row shows the marginal effects on banks that are TBTF before the implementation of Basel III; the fourth row shows the marginal effects on TBTF banks after the implementation of Basel III. The conclusion here is ambiguous, perhaps it is still too early to draw any conclusions on the effectiveness of Basel III implementation on IGG. The third and fourth rows show that it is only statistically significant increase of IGG for TBTF banks after the Basel III implementation.
Table 3: Marginal Effects on TBTF
Bank Type Bail-In Margin
Small Banks Before 0.1280852***
Small Banks After 0.1617348***
TBTF Banks Before 0.1244295***
TBTF Banks After 0.1158569***
The marginal effects on TBTF banks are again seen in third and fourth rows of Table 3. The third row shows the marginal effects on banks that are TBTF before the implementation – of bail-in instruments; the fourth row shows the marginal effects on TBTF banks after the implementation of in instruments. The conclusion here is that after the introduction of bail-in bail-instruments, it appears that small banks are relatively better off compared to TBTF banks; As the margin’s column, in Table 3, shows that IGG increased for small banks after the
6.3. Risk Impact on IGG
Table 1 summarizes the results of the regression analysis with main effects only as well as including interactions to investigate whether the effects of risk, Basel III, and Bail-in requirements differ between banks that are too-big-to-fail and other (smaller)banks.
Looking at Table 1, the regression in first and third column shows that risk has a positive effect on dependent variable, IGG, for both banks that are TBTF and the banks that are not TBTF. The higher the risk in previous time period, the higher implicit government subsidies for both types of banks. However, an even more compelling case, the risk increased IGG more for banks that are not TBTF compare to banks that are TBTF. The results on this variable are significant at one percent level (p<0.01)
6.4. Too-Big-To-Fail (TBTF) Impact on IGG
The time fixed effects regression with the main effects shows that the banks that are not too-big-to-fail also had lower implicit government guarantees; However, when I include interactions the results change and the higher IGG levels observed for small banks are mainly driven by Bail-in requirements introduction which affected TBTF banks adversely compared to small banks.
The other element worth mentioning is that in the Appendix A, Table 4, which ranks IGG by bank, interestingly only small banks are constantly in the ranking throughout the whole sample period, as opposed to TBTF banks which might be in the ranking for a given year and not in the ranking for the following year. Precisely those banks are: Resona Holdings, Bayerische
Landesbank, Commerzbank, and KBC Group NV.
While the results obtained in the analysis are highly significant they should be treated with some caution as bank specific fixed effects, which were not included in the models, are likely to be an important predictor of IGG. Probably due to their accounting for bank interconnectedness, a factor shown by previous studies to have a substantial effect on IGG and which, due to data unavailability, was not included in the model.
It is also worthwhile mentioning, however, that even without their inclusion the model fits the data extremely well as the R2 of the model without and with interactions amount to 0.915 and
0.918 respectively. This implies that bank fixed effect could account for 8% of the differences in IGG levels at most.
6.5. Basel III Implementation Impact on IGG
Basel III dummy variable, shows that interestingly Basel III not only failed to reduce the implicit government subsidies for banks that are TBTF but seems to have increased it, at least for now. As for banks that are not TBTF Basel III introduction had no effect on IGG. It shows that the -market still believes that senior creditors are still supported by governments. It is a puzzling finding, and the results may change shortly, as it might be too early for effect to truly kick in for now.
6.6. Regional Impact on IGG
The regional dummies tell us that there are higher implicit government guarantees everywhere than the North America; To put it differently, Europe, Asia, and Australia are characterized by higher implicit government guarantees than North America.
Again, this shows that the U.S. banks' regulators are tougher than other regions. In U.S. bank's regulators are prepared to let a bank go under when it fails, e.g., Lehman Brothers Holdings Inc., which is not the case in Europe for example. My results again are significant at one percent level when one look at column one and three of Table 1.
6.7. Bail-In Implementation Impact on IGG
Bail-in dummy variable shows that interestingly Bail-in failed to reduce the implicit government subsidies for banks that are not TBTF; However, Bail-in requirements implementation succeeded in reducing IGG for banks that are TBTF. This might signal that the market is starting to believe-
that the hard stance adopted by regulation authorities on systemically important lenders is indeed real.
6.8. Robustness Tests
The regression is also estimated with robust standard errors, in case some of Ordinary Least Squares assumptions might have been violated. Columns two and four of Table 1 report the same results as columns one and three respectively but with robust, rather than simple, standard errors. The results obtained are the same, which confirms the robustness of the results reported above; that is none of the significance levels change, even if the, usually unrealistic, homoscedasticity assumption is violated.
6.9. Limitations of the study
Although in this thesis, I carefully conducted my analysis, I am still aware of its shortcomings and limitations. The potential flaws of my study are mainly due to lack of data, the sample size, and possibly methodological limitations.
First, I use CDS spreads data, which I could not find for some of banks I would have liked to include in my sample. In my study, I use only two CDS tranches, senior and subordinate, perhaps a model that can accommodate all tranches or even few more might give additional insights. That said, at worst I have underestimated IGG due to a smaller sample size.
Second, in this study, I started with 100 largest banks in the world by assets value, and using Black et al. (2013) framework, which consist of eliminating banks without 24 valid consecutive monthly credit default swaps spreads, I had 43 banks left as my sample. Even though this might be a good start, future studies need to be conducted to see whether an analysis with larger sample might give similar results.
Finally, my study might have omitted crucial explanatory variables, such as interconnectedness of banks variable, and Market risk aversion variable; Thus, the results of my study might be carrying omitted variable bias.
7. Conclusions
Events in the global financial sector that preceded the recent financial crisis support the theory of a paradox of instability: The world financial sector looked strongest when in fact it was most vulnerable. The 2008 subprime crisis was a wake-up call for the world financial system. Since then several measures have been adopted. The grand total of these banking regulations is quite remarkable but probably only part of the truth. Nonetheless, the eagerness to discuss the discomfort caused by recent crisis is a sign of healthy institutions.
It is sometimes not prudent and certainly not easy to end government support to the financial system, especially when the financial sector is in distress. On the flip side, it is important to figure out to what extent the government should intervene. This thesis provides a measure of implicit government guarantees using information embedded in CDS. This thesis also analyses the dynamics of IGG in different regions from January 2009 to June 2017.
My findings show that since 2009 until now IGG has not changed; IGG is positively correlated with economic malaises. It also shows that currently large Asian banks are more subsidized compared to their peers in other regions.
The risk is positively related to IGG and banks that are Too-Big-To-Fail also have higher implicit government guarantees. This study also shows that Basel III failed to reduce implicit government subsidies to large banks, at least for now. The results also tell us that there is higher implicit government guarantees everywhere than North America.
In the effort to balance risks and returns, and limit risk-taking appetite in the global financial sector, there is a need for banks regulations to internalize implicit government subsidies to large-
The stepping stone would be a mechanism to quantify Implicit government subsidies, such one used in this thesis, and use analytical tools to enhance system-wide financial stability.
Just like Caesar’s wife, banks’ accounts must not be under any suspicion in public eye. A mechanism that quantifies IGG in the financial sector on a regular basis would provide much-needed transparency in this industry.
Reference
Asli Demirgüç-Kunt, and Harry Huizinga, 2013, Are banks too big to fail or too big to save? International evidence from equity prices and CDS spreads, Journal of Banking & Finance, Volume 37, Issue 3, March 2013, Pages 875-894.
Roberto Blanco, Simon Brennan and Ian W. Marsh, 2005, An Empirical Analysis of the
Dynamic Relation between Investment-Grade Bonds and Credit Default Swaps, The Journal of Finance, Vol. 60, No. 5 October., 2005, Pages 2255-2281.
Viral V. Acharya, and Timothy C. Johnson, 2007, Insider trading in credit derivatives, Journal of Financial Economics, Volume 84, Issue 1, April 2007, Pages 110-141.
Charles W. Calomiris, 1999, Building an incentive-compatible safety net, Journal of Banking & Finance, Volume 23, Issue 10, October 1999, Pages 1499-1519.
Mehdi Beyhaghi, Chris D'Souza, and Gordon S. Roberts, 2013, Funding Advantage and Market Discipline in the Canadian Banking Sector, Journal of Banking and Finance, Vol. 48, September 2013, 47 Pages.
Kwast et al 1999, Using Subordinated Debt as an Instrument of Market Discipline, Study Group on Subordinated Notes and Debentures Federal Reserve System, December 1999, Staff Study 172.
The Bank for International Settlements, Statistical release - OTC derivatives statistics at end-June 2016, BIS - Monetary and Economic Department, November 2016.
Donald R. Fraser and J. Patrick McCormack, 1978, Large Bank Failures and Investor Risk Perceptions: Evidence from the Debt Market, The Journal of Financial and Quantitative Analysis, Vol. 13, No. 3, September., 1978, Pages 527-532.
Robert B. Avery, Terrence M. Belton, and Michael A. Goldberg, 1988,
Market Discipline in Regulating Bank Risk: New Evidence from the Capital Markets, Journal of Money, Credit and Banking, Vol. 20, No. 4, November 1988, Pages 597-610.
Gary Gorton, Anthony M. Santomero, 1990, Market Discipline and Bank Subordinated Debt: Note. Journal of Money, Credit and Banking, Vol. 22, No. 1 February 1990, Pages 119-128
Mark J. Flannery, Sorin M. Sorescu, 1996, Evidence of Bank Market Discipline in Subordinated Debenture Yields: 1983- 1991. The Journal of Finance, Vol. 51, No. 4, September 1996, Pages 1347-1377.
Andrea Sironi, 2003, Testing for Market Discipline in the European Banking Industry: Evidence from Subordinated Debt Issues, Journal of Money, Credit and Banking, Vol. 35, No. 3, June 2003, Pages 443-472.
Sebastian Schich, and Kim Byoung-Hwan, 2012, Developments in the Value of Implicit Guarantees for Bank Debt: The Role of Resolution Regimes and Practices, OECD Journal: Financial Market Trends, Vol. 2012/1, November 2012, Pages 45-63.
Moody's Global Credit Research, Moody's reviews European banks' subordinated, junior and Tier 3 debt for downgrade, November 29, 2011.
Tu Nguyen, 2013, The disciplinary effect of subordinated debt on bank risk taking, Journal of Empirical Finance, Volume 23, September 2013, Pages 117-141.
Bhanu Balasubramnian, and Ken B. Cyree, 2013, Market discipline of banks: Why are yield spreads on bank-issued subordinated notes and debentures not sensitive to bank risks? Journal of Banking & Finance, Volume 35, Issue 1, January 2011, Pages 21-35.
Lei Zhao, 2014, Market-based Estimates of Implicit Government Guarantees in European Financial Institutions, European Financial Data Institute, August 2014, 48 Pages.
Kenichi Ueda, B Weder Di Mauro, 2013, Quantifying structural subsidy values for systemically important financial institutions, Journal of Banking & Finance, October 2013, Pages 3830-3842. Bryan Kelly, Hanno Lustig, and Stijn Van Nieuwerburgh, 2011, Too-Systemic-to-Fail: What Option Markets Imply about Sector-Wide Government Guarantees, American Economic Review, Vol. 106, No. 6, June 2016, Pages 1278-1319.
Andreas Jobst, and Dale F. Gray, 2013, Systemic Contingent Claims Analysis; Estimating Market-Implied Systemic Risk, International Monetary Fund Working Paper, No 13/54, February 2013, 93 Pages.
Zoe Tsesmelidakis, and Robert C. Merton, 2012, The Value of Implicit Guarantees, March 2013, Minneapolis Federal Reserve Bank Working Paper, 46 Pages.
Frederic Schweikhard and Zoe Tsesmelidakis, 2012, The Impact of Government Interventions on CDS and Equity Markets, European Financial Data Institute, June 2012.
Viral Acharya, Deniz Anginer, and Joe Warburton, 2013, The End of Market Discipline?
Investor Expectations of Implicit Government Guarantees, Working Paper, May 2013, 58 Pages. Moody's Investors Service of 1982-2012.
Martin Weber, and Lars Norden, 2012, When Senior meets Junior: Information in Credit Default Swap Spreads of Large Banks, Working Paper.
Francis Longstaff, and Eduardo S Schwartz, 1995, A Simple Approach to Valuing Risky Fixed and Floating Rate Debt, Journal of Finance, July 1995, Vol. 50, Issue 3, Pages 789-819.
Francis A. Longstaff, Sanjay Mithal, and Eric Neis, 2005, Corporate Yield Spreads: Default Risk or Liquidity? New Evidence from the Credit Default Swap Market, Journal of Finance, October 2005, Volume 60, Issue 5, Pages 2213–2253.
John C. Hull, 2008, Options, Futures, and Other Derivatives - 7th Edition.
João A. C. Santos, 2014, Evidence from the Bond Market on Banks’ “Too-Big-to-Fail” Subsidy, the Federal Reserve Bank of New York, December 2014, Volume 20, Number 2
Lamont K. Black, Ricardo Correa, Xin Huang and Hao Zhou, 2013, The systemic risk of European banks during the financial and sovereign debt crises, the Federal Reserve System, International Finance Discussion Papers, No 1083.
Markit Group, 2009, Credit Derivatives Glossary, March 2009
Ana Babus, Elena Carletti, Franklin Allen, 2009, Financial Crises: Theory and Evidence, Annual Review of Financial Economics, June 2009, Vol. 1, pp. 97-116.
Ben S. Bernanke, 2013, The too-big-to-fail problem, Federal Reserve Board, March 20, 2013 The Financial Stability Board (FSB), and Basel Committee on Banking Supervision (BCBS), 2016, List of global systemically important banks (G-SIBs), November 2016.
Isabelle Distinguin, Market Discipline and Banking Supervision: The Role of Subordinated Debt, University of Limoges, February 2008, HAL Id: hal-00916729.
IMF Global Financial Stability Report, 2009, International Monetary Fund - World economic and financial surveys, October 2009, 0258-7440.
36
