Manage your bank, not your model

Double-check your discount rates in the rates-down stress-test

Great headline today from the Financial Times:

Fed study puts ideal US interest rate at -5% - Financial Times – April 27, 2009
The ideal interest rate for the US economy in current conditions would be minus 5 per cent, according to internal analysis prepared for the Federal Reserve’s last policy meeting.  The analysis was based on a so-called Taylor-rule approach that estimates an appropriate interest rate based on unemployment and inflation.  A central bank cannot cut interest rates below zero. However, the staff research suggests the Fed should maintain unconventional policies that provide stimulus roughly equivalent to an interest rate of minus 5 per cent.

It’s a silly attention grabbing headline, but it got me thinking about modeling rate floors.  Anybody running an asset/liability model has no doubt adjusted their model to take into account rate floors.  They’re most visible in the earnings simulation when modeling core deposit rates.  Banks can’t pay much less than 10 or 20 basis points.  Rate floors may also show-up in the bank’s loan portfolio - making their rates down earnings simulation behavior look more like a fixed-rate portfolio.

One place where I’ve seen rate floors noticeably absent is in the balance sheet (EVE) simulation.  You should double-check that the discount rates used for your present value simulations don’t fall below zero.  Take a look at the rate down shocks; increases in portfolio values should slow down noticeably as you get past the –100bp shock.  Of course it depends on the type and base-case duration of the portfolio - it should be much more noticeable on your shorter-term portfolios.

Quit complaining, it’s what you wanted…isn’t it?

I wish I could capture today’s results of my Google web search for “credit value adjustment” (Thursday, April 23, 2009.)  The first ten or eleven pages returned what I think are relevant hits.  The reason of course is because 1Q earnings are being reported by all the big companies.

I had to laugh-out-loud after reading some of the news reports.  I don’t really care who posted gains or losses, and I don’t really care why.  What interests me is the total hypocrisy of the “fair value devotees”.  On one hand they want companies to report the fair value of “toxic” assets (or whatever the latest euphemism is) because it gives the investor a true “transparent” picture of the bank’s financial health.  On the other hand they think it’s wrong for a bank to mark-to-market their debt because it communicates a confusing message. 

Here’s one article from the NYT April 17, 2009

After Year of Heavy Losses, Citigroup Finds a Profit

After more than a year of crippling losses and three bailouts from Washington, Citigroup, a troubled giant of American banking, said Friday that it had done something extraordinary: it made money.

But the headline number — a net profit of $1.6 billion for the first quarter — was not quite what it seemed. Behind that figure was some fuzzy math.

Like several other banks that reported surprisingly strong results this week, Citigroup used some creative accounting, all of it legal, to bolster its bottom line at a pivotal moment.  But the long-struggling company also employed several common accounting tactics — gimmicks, critics call them — to increase its reported earnings.

…

One of the maneuvers, widely used since the financial crisis erupted last spring, involves the way Citigroup accounted for a decline in the value of its own debt, a move known as a credit value adjustment. The strategy added $2.7 billion to the company’s bottom line during the quarter, a figure that dwarfed Citigroup’s reported net income. Here is how it worked:

Citigroup’s debt has lost value in the bond market because of concerns about the company’s financial health. But under accounting rules, Citigroup was allowed to book a one-time gain approximately equivalent to that decline because, in theory, it could buy back its debt cheaply in the open market. Citigroup did not actually do that, however.

“It’s junk income,” said Jack T. Ciesielski, the publisher of an accounting advisory service. “They are making more money from being a lousy credit than from extending loans to good credits.”

And another from a blog post on The Daily Markets, April 20, 2009

Big Bank Profits Are Bogus! It’s A Massive Public Deception!

Gimmick #3. The great debt sham. Consider this scenario: A financially distressed real estate developer owes the bank $4 million. His revenues have plunged. He’s lost a fortune in his properties. And he’s on the brink of bankruptcy.

Therefore, in the secondary market, traders recognize that loans like his are worth, say, only half their face value, or about $2 million. So far, a very common situation, right?

But now imagine this: He walks into the bank one morning and claims that he really owes only $2 million. Why? Because, in theory, he says, he could buy back his own loan for that price, thereby reducing his debt in half.

In practice, of course, that’s a pipedream. If he actually had the cash to buy back his own loans on the market, then he wouldn’t be financially distressed in the first place. And if he weren’t financially distressed, his loans wouldn’t be selling on the market for half price.

The reality is that he can’t buy back his own debt and never will. And even if he could someday, he will still be on the hook for the full $4 million unless and until he files for bankruptcy and the bankruptcy judge decides otherwise.

That’s why the government would never let real estate developers - or hardly anyone else, for that matter - mark down the debts on their books and still stay in business. But guess what? The government lets banks do precisely that!

It’s the ultimate double standard: The banks get away with inflating their toxic assets. But at the same time, they’re allowed to mark to market their own debts, which happen to be trading at huge discounts on the open market precisely because of their toxic assets.

Accountants call it a “credit value adjustment.” I call it cheating.

And here’s a bank that posted a loss because their debt had a higher market value, from Yahoo! Finance, Wednesday, April 22, 2009:

Morgan Stanley loses $578M in 1st quarter

The bank was also hit, counter-intuitively, by an improvement in the value of its own debt in the first quarter. This improvement essentially increased the amount of debt on Morgan Stanley's books…So if Morgan Stanley had to buy its debt back at the end of the first quarter, it would have had to pay more for it than it would have at the end of last year. And accounting rules require this change to be recorded as a loss.

Perform a web search for “Morgan Stanley credit value adjustment” - You’ll hardly see anyone complaining about it or calling this “voodoo” accounting.

Fair value just complicates things, it doesn't make things easier to understand
The stupid thing about this is that folks want some assets to be marked-to-market even if their isn’t a real or viable market out there (i.e. no trading values available).  So the reported “market” value of these assets is based on estimates and management judgment.  Yet the debt these banks are marking to market is actually being actively traded.  There are real buyers and sellers out there trading this debt.  If you’re interested in bank’s using marked-to-market accounting and if this debt is worth more (or less) shouldn’t that be reflected in the bank’s financial statements? (Just like you want it to be for assets).

You can’t have it both ways folks.  It’s called a balance sheet for a reason.  If you start applying different rules to different sides of the equation you are going to make things less “transparent” not more.

“fuzzy math”, “junk income”, “cheating”.  You’re kidding me right?  “Creative accounting” is the very definition of fair value accounting.  You either want banks to use it or you don’t, pick a side and stop complaining.

Just in case you’re wondering, I think we should ditch fair value accounting.  Management teams already have a pretty big bag full of accounting tricks to use.  We don’t need to double their supply and make it even harder for the rest of us to interpret their results.

What interest rate risk policy limits have other banks set?

Back in December I completed a series of posts about setting ALCO policy limits for interest rate risk.  One of the questions we often get asked is, what policy limits do you see at other banks?  We have some data that will help answer that question.  Each quarter for the past 3 years we’ve been asking our A/L BENCHMARKS clients, and canary report users to answer and update a survey of their IRR policy limits.

The first question we asked was: What is your bank’s Net Interest Earnings at risk policy limit given +/-200bp shock?  The largest response is in the “-5.1 to -10.0” category.  24% of those responding choose this range and based on my experience I’m guessing that most of them have a limit of –10.  I was a little surprised to see the size of responses to the “15.1 to 20.0” category.  Again I’m guessing that most of these responses mean –20.0 rather than something in between.  A -15.0 to –20.0 net interest earnings at risk is relatively high, but not extremely high.  And if a bank is growing, a –15.0 net interest earnings at risk limit or above is reasonable.

The second question we asked was: What is your bank’s EVE at risk policy limit given +/-200bp shock?  The largest response here was in the “Unknown or No Limit” category.  I could get really cynical here and say that I’m not surprised.  Most bankers really don’t pay that much attention to the EVE measurement anyway…but maybe I’ll save that for another post.

Beyond the “I don’t know” or “No limit” category, one of the largest responses is -15.1 to -20.0.  This seems quite normal because the rule-of-thumb is usually -20.0.  Again I was bit surprised by the volume of responses in the –5.1 to –10.0 category, especially for EVE at risk.  Equity, almost by definition, is a levered position.  Total equity to assets for community banks can be as low as 7%.  When you measure changes in total balance sheet value against the comparatively smaller amount of equity you tend to see much more volatility.  In general EVE at risk tends to be higher (i.e. more volatile)  than net interest earnings at risk.  I think an EVE at risk limit of –10.0 or less is pretty tight.  Bank’s with generally shorter-term assets can probably manage to a tight EVE at risk limit like that; but banks with longer-term assets would probably have a harder time.

The difference between interest rate risk measurement theory and practice

To keep her teaching certification current my wife attends various conferences and classes throughout the school year.  While attending each session she confesses that, while she enjoys hearing about new ideas and strategies, she’s secretly wondering when the last time this person was actually in a classroom.  Sounds good in theory, but have you actually tried this stuff?  Every one of us has probably echoed this sentiment at one time or another… “Sounds like a good idea, but…”.

Interest rate risk measurement for community banks is one of those areas where there is a HUGE gap between theory and practice.  Or at least, I think their should be.  For years the ALM modeling world has been awash in terms like “shocks”, “ramps”, “twists”, “duration”, “interest rate elasticity”, “stochastic-modeling”, “Monte Carlo simulation”, blah, blah, blah.  There are countless books and seminars offered that describe the “best” methods to use when measuring a bank’s interest rate risk. 

It’s been my experience that, for most community banks, any additional IRR analysis beyond the standard +/-100bp, +/200bp, and +/-300bp shock is a practical waste of time.  Now before the IRR measurement zealots out there start emailing me to tell me how shallow my thinking is, please note that I said “most” community banks, not all.  And I said “practical” waste of time, not total waste of time.

There is additional information to be gained, but in most cases it’s only academically interesting.  If my earnings at risk given a +200bp instantaneous shock is –8.5%, who cares if my earnings at risk given a rate ramp of +50bp per quarter is –7.1%?  What’s the difference, –8.5% or –7.1%?  If I’m a CFO or President of a community bank there isn’t any practical difference.  I know in general that I’m at risk to rising rates.  I know the severity of risk is in the neighborhood of 8%.  I’m not going to make any substantial balance sheet management decisions based on this difference.

There is a practical limit to the amount of time and energy community banks can spend on IRR measurement.  When the differences between using an instantaneous shock and some other approach become trivial, it’s time to stop the madness.