Stressing Bank Tests

Perhaps we have it easy in the life insurance industry.  Solvency is defined on two criteria: risk-based capital, and a variety of cash flow testing schemes, both contingent and noncontingent.  Truth is, it’s not that easy, but the life insurance industry has been more proactive on risk management than the banks.

I have not talked much, if at all about the “bank stress tests” for one major reason:  in life insurance, there are detailed rules for performing cash flow analyses.  With the bank stress tests, the adverse scenario posited higher unemployment, lower residential housing prices, and lower real GDP than “baseline” estimates.

Okay, that’s nice, but as is often said, the devil is in the details.  Did the banks get relief from the scenarios?  It seems so.  Why?  When I first saw the adverse scenario, I said to myself, “Not adverse enough.  Aside from that, how do you translate the adverseness into actual credit losses?”

The latter question is a critical assumption, particularly for complex financial institutions.  There is no immediate good answer, so how did the US government simplify matters?  We do not know, but we do know that the financial institutions pushed back.

What concerns me the most is that the stress scenarios did not explicitly consider weakness in commercial mortgage pricing.  This is a process that is in its early phases.  Much as REIT stock prices have fallen, and CMBS prices have fallen, the impact has yet to be realized on commercial whole loans on bank balance sheets.

It is very difficult to transform the macroeconomic assumptios of the stress test into usable credit loss data.  Reasons:

  • Differences in bank lending practices makes uniformity tough.
  • Attempts at getting accurate on a company-specific basis introduces the ability of the company to tilt the analysis their way.  Also, company specific loss estimates lack credibility.
  • Loss estimates on new lending classes also lack credibility.
  • Estimates of how sensitive loss estimates are to unemployment, GDP and residential housing prices lack credibility for most lending classes.  We don’t have enough data.

Now I have done stress tests at life insurance companies.  You estimate how much you can take in credit losses without having to dip into surplus assets over a 1, 3, 5, 10, etc-year periods.  You compare those statistics to worst few credit losses over 1, 3, 5, 10, etc-year horizons to get an idea of the likelihood of such large losses.  That has its troubles, but it is better than nothing.  The life insurance industry keeps pretty extensive statistics on its asset losses.

I didn’t get too encouraged by the results of the stress tests.  They were easy tests to pass for many because:

  • The stress scenario isn’t that severe.  I give it better than 50% odds of occurring.  A real stress test has perhaps a 5% chance of occurring.
  • The stress scenario isn’t very prolonged, like the Great Depression.
  • Creating the models that connect the economic assumptions to the loss costs is problematic.  Errors are unlikely to be on the conservative side — both the banks and the regulators are incented to be aggressive, because they don’t want to cause specific panic over their company, or general panic over the banks.  Remember, their is a large  number of people who think this panic is merely confidence/liquidity, and not solvency.   (Then why are we raising capital or selling assets?)
  • There are many new lending classes that have not gone through a full asset default cycle, so their default loss properties in an era of debt deflation won’t be calculable.  We don’t have the data.

When I look at the modest cost of $75 billion of capital to raise, I think of all the capital raised prior to this — and now a measly $75 billion will assure the future solvency of the system.  This is only an opinion, but I think that number is too low, particularly with the troubles in commercial real estate being so early in its cycle.  Remember 1989-92?  The degree of overbuilding now is greater than then.  The losses should at least be proportionate.

My simple bit of investment advice is to underweight the securities (bonds, preferred and common stocks), of the companies that failed an easy test.  That means underweighting:

  • Bank of America
  • Citi
  • Fifth Third
  • GMAC (debt, there is no public common)
  • Keycorp
  • Morgan Stanley
  • PNC
  • Regions Financial
  • SunTrust
  • Wells Fargo

At least, this will be worth watching as a basket from 5/8 on.  It may give us clues to the economy as a whole.  I expect that it will underperform, but I am more certain that it will covary very highly with the market as a whole.  Let’s see what happens.

Theme: Overlay by Kaira