A key outcome of the crisis has been to reduce externalities imposed by banks on taxpayers via moral hazard induced by the too-big-to-fail problem. The solution so far has been to mandate that G-SIBs internalize such costs by having higher capital requirements, and to shield taxpayers by mandating a clear structure of bank liabilities with sufficient loss absorbing capacity (TLAC). These measures are meant to reduce costs ex post but clearly they will change banks’ incentives and business models ex ante. The question is: how?
As an economist I tend to have a two-handed view of issues, and since I don’t have to engage in forecasting or storytelling for a living I will enjoy the luxury of presenting two sides of a few arguments and let the reader decide which one sounds more convincing. I will also not bore the reader with citations but she will surely recognize where most arguments come from.
I will deal with three interconnected issues: a) TLAC and bank assets: will the requirement of increased levels of loss absorbing capacity for G-SIBs and its structure impact on the composition and riskiness of bank assets? b) TLAC composition and bank assets: how do different instruments used to satisfy TLAC requirements affect banks’ asset allocation and risk taking? c) TLAC and banks’ ALM: will banks’ traditional ability to transform illiquid and risky assets into liquid and safe liabilities (such as demand deposits or short-term wholesale funding) be affected by the requirement of increased levels of loss absorbing capacity?
a. Will the requirement of increased levels of loss absorbing capacity for G-SIBs and its structure impact on the composition and riskiness of bank assets?
The first question one should ask is: to what extent can banks choose to be a G-SIB and therefore also subject to TLAC requirements, and would they rather want to be in or out?Because the Basel methodology is based on simple balance sheet indicators, to some extent banks can position themselves – however given that the score of each bank depends on the values of the indicators for other banks it is unlikely that banks’ balance sheets will be much affected by formal considerations derived from this methodology. Besides, the banks that are borderline are relatively few.As for whether to be in or not, there is a trade-off. On one hand being in means a higher loss absorbing capacity requirement, which is costly (in a non-MM framework which is what most practitioners assume, although they might be wrong in the broad sense that the weighted average cost of capital might not be very different if banks were to hold much more equity). On the other hand, being in can be seen as a marketing tool: this bank is a systemic player of the highest relevance, will not be let go bankrupt whatever happens (although ironically strictly speaking TLAC is actually about lining up creditors to bear losses), is in the A-league, etc. This could be beneficial in terms of attracting business, especially in periods of uncertainty – which is when business tends to flee banks. So trying to be in if their competitors are might make sense. In fact, supervisory judgement was used in two cases to classify as G-SIBs banks whose score put them relatively far from the lower threshold of the methodology. One might wonder whether supervisors were being extra prudent or had also some competitive issues in mind.So at least for banks close to the threshold the new regulations might affect M&As strategies, as getting closer might imply becoming a G-SIB, which in turn implies as stated above higher costs but also new opportunities. One might expect more M&As among G-SIBs, which therefore might become even larger and more systemic, and less among almost-G-SIBs. If this is the case, market structure will be polarized between ever larger, more complex and internationalised institutions on one side and medium-sized, at most D-SIBs on the other – although the resolution authority has to give a green light to such deals based on the resolvability of the new structure – and can even mandate divestitures – so this could actually limit polarization.Once it is determined that a bank is G-SIB, how will this affect its business model? Again, there are two possible views.On one hand, the combination of G-SIB buffer and TLAC should increase its resilience ex post, i.e. there is more loss absorbing capacity per unit of risk. This assumes that even if the bank increases its risk taking this is captured by capital requirements and requirements to issue TLAC-eligible liabilities so in fact the bank should be relatively indifferent to requirements that affect its situation in resolution as long as going concern rules are sufficiently binding.On the other hand, since these new requirements are costly banks might try to improve returns by taking on risks that are not adequately captured by the current regulatory framework, i.e. tail risk and other forms of systemic risk.
If the current environment of low interest rates persists, and if markets keep demanding high returns on bank equity, the second option might become more cogent.
The issue is then whether markets will react to all these changes in regulation by lowering their expectations about banks’ overall cost of funding, since they have become safer ex post for most creditors. However it is unclear that this would happen, since banks are safer mostly for creditors, existing shareholders are being diluted and returns to future shareholders depend also on how risk taking will change. The cost of equity will decrease only if investors perceive banks to have become safer, in the sense of more like a utility that provides services than an investment business – but this doesn’t seem to be the case (yet). So as long as the spread between expected return on equity and the risk-free rate is high, and banks are required to hold more capital, there is an incentive to increase risk-taking in forms which are not adequately captured by regulation.
b. How do different instruments used to satisfy TLAC requirements affect banks’ asset allocation and risk taking?
The issue of how the composition of TLAC affects risk taking is key right now as banks are gearing up to choose how to absolve this requirement but it is probably too specific for a meaningful answer at this stage. Markets expect banks to fulfil TLAC requirements overwhelmingly with new equity, new instruments such as CoCos and subordinated debt, rather than with senior unsecured debt – hence the effects described above. Senior unsecured debt could come back into play with an important role if, following the example of German law, a statutory subordination clause would make it automatically eligible for TLAC – this would need to be done via EU law and would dramatically reduce the existing shortfall at least of SSM G-SIBs, but it is not foreseen yet.
The impact of new equity on banks’ asset allocation and risk taking is again unclear. More equity allows banks to take on more risk and creates incentives to do so; the meaningful question is whether this increase in risk is more than proportional to the increase in capital. Again, if regulatory requirements are correctly anchored to risk this shouldn’t be an issue, but if not banks might engage in covert risk taking. so we are back to hoping that the Basel framework is correctly specified in terms of mapping a complex, evolving multidimensional concept such as risk into a single variable such as capital, however layered.
How CoCos affect risk taking is also unclear. In theory they could lead to more risk-taking (and there is some academic literature that explains why and how), but there is no evidence so far. Depending on whether they are principal write-down or conversion to equity the balance of risk between senior unsecured debt holders and shareholders is very different (CoCos holders are assumed to at least break even since they are buying a new class of securities, the issue is whether there is risk shifting among existing stakeholders).In the first case they protect debt holders without diluting shareholders so the impact on risk-taking should be small, in the second case the perspective of dilution in case of negative events might lead bank managers to be more prudent ex ante. It should be noted however that even in the case of principal write-downs in the medium term the bank will probably need to recapitalize, so in the end the difference between the two sorts of instruments might be small and/or mostly in the short term. The issue is clearly an empirical one.
If banks will have to issue senior unsecured debt in significant amounts to comply with TLAC requirements, this debt will probably be re-priced to take into account its bail-in-ability. The key legal issue, which has financial consequences, is that such debt needs to be either statutorily, structurally or contractually subordinated. US banks can easily use structural subordination (debt issued by the holding company) but in Europe this is less easily done and could lead to higher costs.
Again, either banks take this as exogenous and increase risk to increase returns and restore margins, or they reduce risk in a sufficiently credible way that it passes through in their funding costs. The danger is that of hysteresis: if at the beginning markets are sceptical and price upwards senior unsecured securities, banks might be tempted to increase risk and enter therefore into a negative spiral. It would be important then to first give credible signals of risk reduction before issuing such instruments. The timeline of TLAC would allow for this but the urge to frontload to show strength vis-a’-vis competitors might work the other way round.
In the nineties many advocated market discipline as a way to keep banks in check; however the experience of the crisis has been that market discipline is most lax when it should be severe and most severe when the economy as a whole would need some level of forbearance.
TLAC mandates the issuance of the securities which are the most information-sensitive. A possible unintended consequence of such choice is an increase of the role of informational asymmetries and conflicts of interest – phenomena such as risk shifting across classes of liabilities and hidden risk taking might be on the rise, and as we have seen with the crisis when there is a negative shock they act as amplifiers as investors realize that they were fooled and sell off en masse: what started as an effort to reduce systemic risk might end up increasing it in some situations.
This might be second-order compared to the benefits of increased overall resilience but needs to be better understood and monitored. In particular it might entail demand for greater transparency (which would be good and also decrease the ability of banks to take on hidden risks) but also more inefficiencies in capital allocation if such asymmetries are perceived to be too great and lead to debt overhang issues at lower levels of debt than currently.
c. Will banks’ traditional ability to transform illiquid and risky assets into liquid and safe liabilities (such as demand deposits or short-term wholesale funding) be affected by the requirement of increased levels of loss absorbing capacity?
One issue which would need to be better understood is how TLAC will affect collateralized funding by banks. Mechanically TLAC increases the share of banks’ liabilities which cannot be collateralized (equity, CoCos, senior unsecured debt) as the purpose is precisely to increase “generic liabilities” to protect the rest of the balance sheet. This would lead to a decrease of collateralized funding such as repos – and therefore of the transformation of illiquid assets into liquid and safe liabilities. However if banks deem this form of funding particularly convenient they might keep it going and decrease other forms of funding, mainly deposits (which are also liquid and safe). So there might be a decrease in the liquidity transformation function of banks, and concurrently a change in the composition of liabilities depending on the relative merits of the various instruments. For this second effect conjunctural conditions are likely to be the key drivers.
On the other hand, TLAC should make the other liabilities of a bank safer, and to the extent that there is excess demand for safe assets, which for the time being is not being directed towards banks’ liabilities, then this should encourage banks to provide more of such securities.
Financial innovation might also play a role here (as for the issues discussed above). New contractual forms might be designed to make the best use of existing balance sheets once the TLAC part is taken out, and reduce whatever slack there is.
A related question would be: how does TLAC affect the liquidity transformation performed by the banking system at large, since it will affect only the largest banks? In fact these banks tend to be those whose assets are already more liquid. They have larger trading and securities portfolios, smaller loan portfolios and are more skilled in creating structured products. So if TLAC reduces their ability to provide such a service in the face of excess demand, they might transfer their skills to smaller banks, either by acquiring them (however see point above on M&As) or by selling advisory services.
So the bottom-line is: TLAC will change G-SIBs’ incentives and affect both sides of their balance sheet. How this will play out will depend among other factors on how well the rest of the regulatory framework holds up to increased incentives to risk taking, and on how markets perceive banks’ moves. The overall impact on the financial system is difficult to gauge, but TLAC should also change the relationship between G-SIBs and the rest of the banking system. Supervisors will need to dialogue closely with all players to understand changes in business models and not be caught off-guard by developments that are usually more about intangibles such as risk appetite than about quantifiable variables.