Junk Banks?!

(Adapted from Are Systemically Important Banks Junk Credits? Industry Commentary of the Global Association of Risk Professionals)

A great challenge for bank credit analysts is the degree to which banks rely on “extraordinary support” (a euphemism for “bailout”) from their governments.  A bond investor may build a brilliant quantitative model to understand how a bank’s leverage, asset volatility, and other financial, operational, and economic characteristics impact the bank’s estimated default likelihood.  But there’s a final overlay that is vexing:  what is the probability that a government will step in to make creditors whole if the bank fails?

As with almost all such investment analysis questions, there is no definite, unambiguous answer.  But recent research of FitchRatings provides a fascinating observation.  (See the Fitch Special Report “The Evolving Dynamics of Support for Banks,” September 11, 2013.)  For the period 1990-2012, Fitch assessed both the default rate and the failure rate for “senior creditors of systemically important” global banks with Fitch ratings.  The five-year cumulative default rate in the period (1.15%) is six times lower than the failure rate (6.95%).  Fitch defines failure as “defaulted or would … have defaulted without extraordinary support.”

A quantitative model builder might be pleased with this historical data point.  She will use her financials-based risk model to estimate failure rate and then multiply by a new “no-bailout” parameter of one-sixth (the Fitch result) to get the model’s estimated default probability.  This approach is feasible, but we’re struck by a different observation.

Stepping outside the model-building exercise, the five-year failure rate of 6.95% (call it 7%) is striking.  Referencing Fitch’s Default Study of 2012, the Global Corporate Finance Average Cumulative Default Rate for double-B rated entities is 6.91% (call that 7% also).  (See the table on page 9 of the Fitch Special Report “Fitch Ratings Global Corporate Finance 2012 Transition and Default Study,” March 2013.)  Thus, in the absence of “extraordinary support,” the world’s systemically important banks behave like junk credits.  At least, this is the blended effect of the Fitch universe for the period 1990-2012.  Neither qualification, though, is disquieting.  The universe is large and the 23-year time period is long and indicative of recent history.

To our knowledge, the dominant rating agencies do not assign underlying bank ratings (such as “bank financial strength ratings”) that are non-investment grade.  For example, Fitch itself shows a list of 28 global systemically important banks in Appendix 4 of “The Evolving Dynamics of Support for Banks” and gives a junk "viability rating” to the Bank of China only.  Hence, the rating agencies likely disagree with the characterization that “systemically important global banks are junk” absent government support.  But we appreciate the Fitch study precisely because it shows observed data rather than just potentially optimistic models and judgment.  We’re not aware of any alternative historical studies of this sort.  (If such studies exist, please tell us!)

Of course, this question matters greatly to bank regulators, investors, and taxpayers as well as to rating agencies.  The world is now tightening prudential regulation by elevating capital requirements, pondering liquidity enhancements, considering a maximum leverage ratio (absent risk weighting), forcing central clearing of derivatives, et cetera.  Do regulators and the broader community agree that the “starting point” of systemically important global bank credit quality is junk?  Or, as I suspect, is this point contentious?  We need discussion and debate!  Our view is that the Fitch study is critically important.  It deserves much wider attention and acclaim.  Like a published scientific result, the world needs other rating agencies, bank regulators, and academics to perform similar studies with other data sources to validate or dispute the finding that “banks are junk” without government support.