G-SIB Failures: Either Way, You Pay

Credit: Thierry Cheverney

I have been reading the public responses to the November 2014 Financial Stability Board (FSB) consultative document on "Adequacy of Loss-Absorbing Capacity of Global Systemically Important Banks in Resolution"   This consultative document frames the broad outlines of what will eventually constitute the international framework for the bank "bail-in" process.  A process meant to (hopefully) minimize the impact of a bank failure on financial stability, (hopefully) eliminate future bank bailouts by governments, and (hopefully) do away with the "too big to fail" concept.

In its most general sense, the idea of a "bail-in" is to have Global Systemically Important Banks (G-SIBs) maintain a layer of  Total Loss-Absorbing Capacity (TLAC)-eligible liabilities in an amount sufficient to convert (some or all) of the bank's obligations to unsecured and uninsured creditors into a confidence-inducing chunk of bank equity capital.  The debt-to-equity conversion would be triggered should the bank's existing shareholder's equity cushion be depleted by losses.  

Think of the G-SIBs as the supertankers of global banking and their skippers, our Captains of Finance.  Many years ago, changes were made to the structural integrity of our global fleet of petroleum supertankers.  Most of the fleet transitioned from single-hulled vessels to double-hulled vessels in the interest of greater public safety.

TLAC works in much the same way - the first hull (and line of defense) is shareholder capital, the second hull (and line of defense) are the monies owed to unsecured and uninsured creditors that can be converted to equity capital.

It's the intent of the FSB to identify a TLAC of sufficient thickness to help avoid a financial Exxon Valdez disaster - a supertanker that lacked a double-hull at the time of its grounding in Alaska in 1989.

The consultative document proposes a minimum TLAC requirement of approximately 16-20% of risk weighted assets and 6-8% of total assets or at least twice the Basel III leverage capital requirement, excluding any Tier 1 common equity used to meet any applicable capital buffer requirements.

The public responses, which come from banks, trade associations, academia, rating agencies, and others, are supportive of the "bail-in" idea in general, but differ on interpretations of the proper size of the TLAC cushion, eligible TLAC debt instruments, and the issue of pre-positioning TLAC  - both from a home country/host country perspective and from the perspectives of single point of entry (SPE) resolution and multiple point of entry (MPE) resolution.

There are two things that make me uneasy about the proposed bail-in regime.  One is the compartmentalized approach regarding bank insolvency and the other is this annoying hallelujah chorus that is proclaiming the end of costs to taxpayers by eliminating government bank bailouts of G-SIBs.

Bail-ins for G-SIB leviathans will likely shift the costs to taxpayers from one government expense bucket (targeted bank bailouts) to a different government expense bucket (economic stimulus programs).  Because of the immense size of the G-SIBs relative to any measures of GDP (country, regional, global), the losses sustained by a G-SIB failure will likely be destabilizing to real economies. When it comes to a G-SIB resolution, government pays, either way.


A compartmentalized approach

It is important to remember that banks can fail in two ways.  First, there is the classic equity insolvency.  Losses erode the shareholder capital of the bank until the point where government authorities trigger their on-the-shelf resolution plans and wind down the affairs of the bank.

The other bank failure vector is a liquidity insolvency.  A bank is unable to meet funding demands and the central bank (as lender-of-last resort) cannot continue to support the bank's liquidity due to lack of eligible collateral or it unilaterally decides to terminate lifeline funding to the bank.

TLAC deals with the classic equity insolvency.  It doesn't appear to me to deal with the potential of liquidity insolvency.  And most important, it doesn't discuss the calculus of the critical dynamic inter-relationships between capital and liquidity in times of financial stress.  Mandatory liquidity coverage ratios will buy time, but are not themselves a solution.

A key question: Will market participants lend funds (without asset pledges) to a "bailed-in" bank absent government guarantees, when there might be marketplace uncertainty about how deep the bank's capital hole really is?

In the fog of a financial crisis, transparency and truth are always the first casualties.  Information is incomplete, rumors abound, and it always takes some time for the second shoe to drop.  And like all rational players (who tend to prefer pleasure to pain), institutional funds providers, honoring their own fiduciary responsibilities, might just stand on the sidelines.

The paper obliquely hints at these difficulties, when the authors say:

"Resolution is not resurrection.  But nor is it insolvency: the institution or successor institution (e.g., bridge institution) has to meet at least the minimum conditions for authorization in order that supervisors may allow it to continue performing authorized activities, in particular critical functions.  Moreover, the reorganization or solvent wind-down that will be necessary following resolution may require a level of capitalization above that required by supervisors so that counterparties continue to trade with the resolved firm and provide funding to it."

Size Matters

Second, the hallelujah chorus ignores the implications of the total cost of a G-SIB failure.  It's not that living wills, the bankruptcy process, and orderly liquidation authority cannot wind-down one or more of any failing G-SIBs by bailing-in shareholders, unsecured creditors, and uninsured depositors.

But at the bail-in point, how large are the cumulative financial losses sustained by legacy shareholders and those now bailed-in?  And second, how will those cumulative financial losses be subsequently transmitted through the national and global real economies?

Remember the sub-prime Collateralized Debt Obligations (CDOs)?  Risk managers were patting each other on the back because they had sliced and diced these financial instruments and thought they had diversified away the risk by distributing the risk to thousands of investors all over the world.  So instead of a few big losers, the smaller distributed losses dragged down economies all over the globe. Sub-prime lending triggered, not only a recession, but this era's mother of all recessions - the Great Recession.

As a point of reference, a May 2012 working paper produced by the research department of the Federal Reserve Bank of Philadelphia pegs the damage (total write-downs) from sub-prime CDOs at $420 billion.

At year-end 2014, for example, the two most complex G-SIBs, JPMorgan Chase and HSBC Holdings, had total Tier 1 capital of $187 billion and $153 billion respectively.  Just pause and reflect for a moment on the size of these numbers.  These sizable Tier 1 capital numbers, when translated into potential bail-in-able dead-weight losses (not even considering the contagion/propagation effects of investor debt leverage) could, individually or collectively, induce recessions in the real economy.

To paraphrase Ross Perot, what you would likely hear "is the giant sucking sound" of aggregate economic demand dropping due to understandably defensive financial behaviors on the part of the injured legacy shareholders, the new bailed-in shareholders, and some portion of the now understandably frightened creditors in the G-SIB.

So either way, government pays for G-SIB failures.  In a bail-in, TLAC regime, government expenditures will likely be in the form of economic stimulus or rehabilitation programs instead of targeted bank bailouts.  Remember, the world has experimented with the bail-in concept during the banking crisis in Cyprus in 2013.  The deep scars from that country's economic upheaval have yet to heal.

It is the size of the cumulative loss to the bailed-in shareholders, creditors, and uninsured depositors that needs to be reduced in order to limit the impacts to the real economy of G-SIB failure.  That argues for less market share concentration in the market for banking services globally.  If we want to contain the economic damage from the failure of a G-SIB, and limit the costs to government, the smaller the all-in costs of G-SIB failure, the better.  Until G-SIBs get smaller, either way, government will pay.