Document Sample
One-Size-Fits-All Powered By Docstoc
					Bank Safety Soundness Advisor                                                          Monday, December 13, 2010

    Stress Testing Best Practice: For Good Results, Keep It Simple
    Published on: December 13, 2010

    If you want a complicated stress testing program, it isn’t hard to get one. In the last decade, any number
    of complicated new stress tests and variations of existing stress tests have emerged, promising a finer,
    more perfectly calibrated look into your portfolio. Resist the impulse to use them, experts say. The best
    testing program for most community and regional banks may be a small set of basic, well-chosen tests.

    “It’s tempting, when you have all these tools and all this technology, to look at your data and say,
    ‘Hey, look at all the scenarios we could run – We can find out a whole bunch of things,’” says Brad
    Olson, a stress test consultant and president of Olson Research Associates, Inc, in Columbia, Md. “But
    at the end of the day, if you’re running multiple scenarios, as many as 20, what you’re left with is a pile
    of data. That’s a lot of data you’ll have to sift through before making any decisions about your bank.”

    Using a simple set of stress test scenarios is a best practice, agrees Geoff Rubin, principal, Second
    Pillar Consulting in Glen Allen, Va. (Editor’s Note: Second Pillar and Bank Safety & Soundness
    Advisor’s parent company have an ongoing business arrangement.)

    “Don’t feel uncomfortable about adding a new test, but be comfortable that a small range of tests will
    provide a sufficient breadth of scenarios,” he adds.

    In regards to interest rate risk, the regulators and the industry have already whittled the universe of
    stress tests down to a small set of generally agreed-upon tests. These are the tests you’ll want to run,
    says Olson.

    “We had the same issues years ago when a joint policy on stress testing came out,” Olson says. “Stress
    testing is not new. Back then, the policy statement used stressed interest rate environments for tests.
    Since that time, everyone suggested new tests, [like] the ramped rate twist or the nonparallel shock.
    Not all of these suggestions are necessarily bad, but it’s better for most banks to look at the pretty good
    tests that regulators and the industry have already settled on.”

    Standardization hasn’t arrived for liquidity risk and credit risk stress tests, but that may not be too far
    off, he adds.

    When it comes to stress tests, is it really one-size-fits-all? “You can look at how the industry and the
    regulatory environment has solved this,” Olson says. “But for interest rate, liquidity and credit risk,
    then yes, I’m suggesting this.”

    Another good reason to keep stress testing simple: peer comparison. “If one bank is stressing this way
    and another is stressing that way, how do you compare the results,” Olson says. “Someone at the bank
    may say: ‘That’s not important for us, because we’re different. We’re unique.’ But comparing your
    data to peer data is what we, as an industry, do. That’s how we can analyze things.”

    What basic tests could or should you use? Here’s what Olson recommends:

    Interest Rate Risk Tests

    For interest rate risk, all you need is the parallel shock, Olson says.                                            Page 1 of 3
Bank Safety Soundness Advisor                                                          Monday, December 13, 2010

    “You can run multiple interest rate risk stress tests if you want,” he says. “You can run 20 or 30 of
    them, but we’ve settled on a single test, the instantaneous parallel shock.”

    Shock up and down at 100bp, 200bp, 300bp and 400bp and you’ll get the data you’ll need, he adds.
    “It’s a pretty good test. Did we test every possibility? No, but if you start testing too much you’ll begin
    to lose sight of the meaning of this.”

    Credit Risk Tests
    You may only need five tests to test your credit risk, Olson says.

    1. The Noncurrent Stress Test. “You’ve already reported a level of non-performing loans,” Olson says.
    “What if they all had to be charged off?” Here, you test for exactly that scenario. Make sure to include
    all of your bank’s non-performers, including those 90-day past due, less than 90-day past due and non
    accrual loans.

    2. The Peer / Market Stress Test. Test your bank against the average level of non-performing loans by
    sector in your peer group. If your bank’s charge-offs are higher than average in any given sector, sub in
    your bank’s numbers.

    3. The High Stress Test. This test assumes that charge-offs will go beyond the peer average by one
    standard deviation. “As a rule of thumb, this test covers nearly two-thirds of all noncurrent levels in the
    peer group,” Olson adds.

    4. The Extreme Stress Test. For this test, begin with peer average and multiply by three. “The numbers
    will start to look really big, but if you look at peer data, there are banks with numbers that high,” Olson

    5. The Sector Stress Test. Here is where you test your bank’s concentrations. Begin with peer / market
    levels and, where you’ve identified concentrations, sub in extreme test numbers for the sectors with
    those concentrations. If you have a high concentration of consumer lending, you might have an issue
    with credit scores. Test for that, Olson says. If you have a high concentration of commercial lending,
    you may want to model for cash flows going bad. Test the areas of the portfolio that need it most.

    Liquidity Risk Tests
    For liquidity risk, start with the same shocks you’ve used to test for interest rate risk, Olson says.
    Shock up and down at the same intervals. To this, Olson suggests adding a pair of scenarios:

    1. Deposit Run-Off. Basel suggests modeling for a 7% deposit run-off over a several month time-frame
    and we agree, Olson says.

    2. Capital Market Stress. This test assumes that portion of the banks bond portfolio becomes
    unmarketable due to the lack of market depth. “What if the market depth isn’t there for the bonds you
    hold?” Olson asks. “We experienced that in 2007 and 2008.”

    “For most community banks or credit unions, that’s a lot of data,” Olson adds. “If they do this, it will
    meet what I think regulators want.”                                           Page 2 of 3
Bank Safety Soundness Advisor                                                           Monday, December 13, 2010

    Using Stress Test Results
    Be careful how you interpret or react to your test results. The point of stress testing is to give context
    and occasion to critical thinking, not to identify a scary number and make radical management
    decisions to correct that number, says Second Pillar’s Rubin. You won’t know if any one scenario is
    any more likely than any other scenario and overreacting to one may leave you exposed to a range of
    alternative outcomes. The biggest peril in stress testing is in managing to the scenario, he adds.

    “If you run a test and find that, in that scenario, your capital position would be $10 million short of the
    regulatory minimum, you might be inclined to raise $10 million in capital. That isn’t the purpose of
    stress testing,” he says.

    Incidentally, the regulators didn’t follow this advice when they stress tested the country’s largest banks,
    Rubin adds. After they ran the Supervisory Capital Assessment Program (SCAP) tests they required
    that those banks raise capital. “There’s always a danger that your examiner will want you to manage to
    a specific scenario and to structure your portfolio as if that scenario will transpire. That’s a bad

    Stress testing should instead occasion discussion, agrees Olson. For example, a deposit runoff test will
    probably turn out some unpleasant numbers. What could or should management and the board do with
    this information? “You might want to talk about early withdrawal penalties for CDs,” Olson says.
    “Maybe it’s been sitting at 3 months of interest for years. Maybe the bank should switch to one year’s
    worth of interest. Absent this analysis, many banks wouldn’t have this discussion.”

    The best way to leverage stress testing is to make sure it’s integrated into the strategy and risk
    discussions your bank is already having, Olson adds. Banks already plan budgets and forecasts, he
    says. Stress testing can fit within this process.

    “If you use stress testing as a separate tool, it’s easy to toss the results into a drawer as soon as the
    regulator leaves,” he says. “If testing isn’t part of the management process, it won’t work.”                                             Page 3 of 3

Shared By: