DISCUSSION

In our analysis in chapter 2 and 3 on the new regulation and the MREL amount, we saw that the T3 instrument comes with a high degree of complexity. This is not necessarily due to the characteristics of the instrument itself, but because of the regulatory environment of which it is a part and its position in the capital structure.

This makes it difficult to estimate how the spreads on T3 should be impacted by price influencing factors, and there is evidence that the market is currently pricing the T3 on the basis of T2.

As the results of our model show that the spreads on T3 and T2 are to a certain extent driven by some of the same factors, we are not overly critical about this approach, since the T2 price does after all reflect the bank’s credit risk. Neverthe-less, our model does also show that this method requires some caution, as one could easily overlook certain price mechanisms of the T3 tranche that impact the value of the T3 relative to the T2 instrument.

In particular, the model shows that certain factors have an impact on the absolute spread level, while not having much of an effect on the spread ratio between T3 and T2. These parameters are the volatility of assets and the CET1 ratio. On the other hand, some parameters do impact the relative spread ratio between the T2 and T3: the number of AT1, T2, and T3 debt tranches; the costs of the resolution process; the location of the PONV; and the MREL ratio. These results might help explain some of the variation we see across current T3 instruments.

As discussed in the analysis on the shortfall of MREL-eligible debt instruments, we do expect to see a great amount of T3 issued in the future, much of which will come from banks that have not yet introduced the T3 tranche into their balance sheet. Our model provides quantitative predictions of how these new T3 instru-ments should trade compared with current T3 instruinstru-ments, and provides an indi-cation of the relative spread ratio to T2 instruments.

Moreover, our model is able to provide quantification for some of the great uncer-tainties that surround the topic of bank capital regulation. Specifically, we address the fact that determining the PONV is not straightforward, and we are able to present concrete figures of the uncertainty regarding this decision. This is an im-portant contribution of our model.

We do not claim that we have developed a pricing model that a credit trader can

such a model. We do, however, believe that we have developed a model that can be used to evaluate the factors that drive credit spreads on bank bonds, to analyze how uncertainties regarding regulation affect prices, and to describe how these bonds should be valued relative to each other.

Model Assumptions and Limitations

Bank regulation is a complicated field, in particular due to its ever-changing nature and the myriad of different measures, definitions, and technical twists. At the same time, matters are further complicated by the fact that regulation can be difficult to generalize, because some requirements are set on a case-by-case basis, whereas oth-ers have not even been finalized yet. To finally arrive at our waterfall model, we have had to cut some corners along the way in this regulatory labyrinth. Therefore, we have made a number of simplifying assumptions. Furthermore, we have had to stop ourselves from following other interesting trails of research, and focus on the objective of this thesis: namely, to price T3 bonds. In the following, we will discuss the simplifying assumptions of our model and the limitations of our study

Consideration of the post-resolution bank in the case of bail-in

One of the main objectives of the BRRD and the MREL is that a bank should be able to function and continue as a new bank after a bail-in event and recapitaliza-tion. For simplicity, we have had to ignore the post-bank setup. When reaching the PONV, we have assumed that losses are allocated to creditors according to our established waterfall, but we have not considered what happens afterwards. Thus, if the PONV is reached at time 𝑡= 3, for instance, we stop the simulation and allocated losses, and investors receive their recovery value. Ideally, we should make assumptions regarding the recapitalization of the bank and continue the simulation until time T. However, this would demand a considerable amount of assumptions regarding the capital structure of the new bank, the conversion of bail-in liabilities to equity or other instruments, issuance of new liabilities, etc., and would also require us to price both equity and AT1 capital. We have therefore chosen not to make such assumptions.

Inability to Capture All Regulatory Complexities

In practice, bank regulation is a continuous and discretionary discipline that in-volves close dialogue with a bank’s management. This feature cannot be incorpo-rated into a model, where trigger points need to be “hard” and circumstances can-not be evaluated. Unfortunately, our model consequently appears slightly stylized.

One could argue that our model needs “situational judgement.” This is of course a general problem in academia but, as the new regulatory framework is to a great

extent based on discretion, soft triggers, and early intervention actions, it is a par-ticular challenge in our setting. It means that we have had to ignore several aspects of the new regulation, such as the P2G trigger.

Furthermore, this means that we have had to make simplifying assumptions re-garding fundamental regulatory aspects, such as the PONV. In this thesis, we have chosen the PONV to be at 6% CET1 ratio, which is in line with market research.

However, as we saw in the results section, the calculated spreads are sensitive to changing the PONV. While the 6% CET1 ratio might be our best estimate for the PONV, it is important to note that this point could be different and vary signifi-cantly during abnormal times. During the financial crisis, many of the troubled banks were “well capitalized” according to Basel standards right before they had to be bailed out (Flannery, 2015). Hence, it is important to keep in mind that our model does not capture the complexities of real life bank regulation. This leads to the important discussion below of leaving out other real life aspects of bank regu-lation that can affect the risk of a bank.

Failure to Capture Other Aspects of Risk and Regulation

The financial crisis emphasized the importance of liquidity risk for both financial institutions and regulators. However, in our model we only consider credit risk.

While it is common in the literature to focus on a single risk aspect in a model, it is important to note that our results do not capture factors that have a large impact on real life prices, such as liquidity (Hull, 2015)

This also means that we are ignoring a vast amount of bank regulation. An example of this is the leverage ratio proposed by the EBA as part of the determination of the MREL, which will serve as a backstop for total amount of leverage. Further-more, we are ignoring important measures of the Basel framework, such as the Net Stable Funding Ratio and the Liquidity Coverage Ratio. While these are important measures to prevent liquidity and maturity mismatches, we do not consider them for simplicity, and also because of our assumptions regarding the composition of the balance sheet. Moreover, including such measures into a model would require data that might be difficult to obtain. The impact of not including these measures is that we might overlook any liquidity or funding risks. In addition, by not intro-ducing any asymmetric relationship between the liquidity and funding availability in normal times versus crisis times, we might overestimate the banks’ ability to recover from a crisis.

Simplified Capital Structure

Most companies, including banks, actively manage their capital structure and their mix of financing to be set at certain targets. A shortcoming of our model is that across the five years of the bank simulation, we set a fixed debt and make the CET1 the fluctuating part of the balance sheet. Therefore, there is no upper bound on what the CET1 ratio becomes. This means that, in some situations, we observe high CET1 ratios at time T. Considering the cost of equity, it seems unlikely that a bank in practice would allow its CET1 ratio to increase far beyond the regulatory minimum requirement. A way to solve this issue could be to incorporate a mean reverting behavior into the level of equity, such that the bank in good times would decrease its level of equity towards its target, and in difficult times increase it correspondingly. In his model of contingent convertible capital, Pennacchi (2010) incorporates such a mean reverting feature of capital. Doing this would add a more realistic characteristic to our model, and help stabilize our capital levels.

Another interesting way to expand our model would be to change the assumptions regarding the balance sheet composition. For example, we do not consider the refi-nancing of liabilities, and make the simplifying assumption that all debt matures at time T and is issued at par value. In addition to refinancing debt, however, our model could be supplemented with stochastic interest rates, as Pennacchi (2011) does. However, while a stochastic interest rate could add interesting elements to the analysis and might be more appropriate if we had floating rate bonds, it would also lead to much greater complexities in the model (Kim, Ramaswamy and Sundaresan; 1993; Longstaff and Schwartz 1995). Moreover, these studies show that introducing a stochastic default-free interest process only has a relatively small effect on credit spreads.

Assumptions as a framework for understanding real world mechanisms

A theoretical pricing model can always be criticized on its assumptions, and apply-ing it in a practical settapply-ing can be difficult as it glosses over many of the real world dynamics. The user of the model must therefore assess whether or not the assump-tions allow for conclusions to be made. The waterfall model developed in this thesis is not exempt from such assessment, nor should it be. The assumptions we have made in developing our model make it a stylized example of a bank under fully implemented MREL. Nevertheless, we do believe that our model provides valuable insights regarding how investors and market players can think about the pricing of the T3 instrument.

Suggestions for Future Research

Based on the assumptions we deliberately chose, as discussed above, several inter-esting ideas emerge for future research and model expansion.

We believe that it could be interesting to model the post-resolution bank, and estimate how the performance of the T3 unconverted and converted debt can im-pact the spreads. This is because the post-resolution bank is an important part of the BRRD and could potentially impact the price of T3 instruments.

Furthermore, we highlighted our omission of other important bank regulation above. It could be interesting in the future to investigate the potential effects of measures such as the net stable funding ratio and the liquidity coverage ratio, and the interaction of these measures with the MREL.

Another idea could be to examine other factors that affect prices. Due to the large shortfall of MREL-eligible debt, it would be particularly interesting to investigate how the market would absorb this vast number of bonds. For example, Newman and Rierson (2004) studied the impact of new bond issues in the European telecom industry, and found that a large portion of bond issuances have sector-wise spread implications. Such a study could be interesting to duplicate for the EU bank bond market, considering the great amount of T3 that will be issued in the coming years.

Lastly, our goal of incorporating real life features into the bank bond-pricing model makes us interested in the effects of actively managing the capital structure. Our observation that certain asset paths result in extensively high CET1 ratios seems unrealistic, and we believe that additional insights into the risks that the T3 might face can be gained by analyzing the T3 spread impact of a target capital ratio and estimating the effects of refinancing risks for a bank on the T3 spreads.

Reflections on the BRRD and the MREL

The purpose of the BRRD was to increase the resilience of the financial system and reduce the likelihood of the widespread contagion experienced during the 2008-09 financial crisis, which led to massive bail-outs. With the MREL, banks should have necessary funds to support a resolution strategy to minimize its negative impacts.

Most of the T3 spreads are low and might attract certain investor groups, such as

the latter are indeed pension funds, for instance, the resolution would affect an important player in capital markets, which might then end up contradicting the purpose of the BRRD. This could lead to the tool of precautionary recapitalization being used more as the norm than as the exception.

All in all, given the additional complexities and the uncertainties surrounding pre-cautionary recapitalization, as well as the potential discretion in the rules across the EU, a natural question is whether or not it would have been more effective to increase the current capital requirements from the Basel framework instead of in-troducing a new set of complex regulations. Essentially, the MREL substitutes what could have been “genuine” going-concern T2 capital with a new type of gone-con-cern capital that can only be used in what seems to be a cumbersome, risky, and uncertain process of balance sheet restructuring.

To mitigate the clear challenges for the BRRD and the MREL, we believe that in the coming years, regulatory authorities will need to increase the reporting trans-parency and provide clear guidelines for investors to properly assess actions and consequences of breaching the different regulatory capital requirements. Therefore, even as it might be difficult and result in ramifications for EU member states, we believe that the European Commission and resolution authorities will have to pro-vide a strong signal to the market that resolution and the bail-in tool will indeed be used unless poor circumstances prevents it.