Print This Post

Stress Testing Effeciency

Madison has been quite successful setting up custom stress test reports and analysis for clients. It seems that every organization looks at stress testing a little differently and therefore needs a solution that can be tailored to their specific requirements. Here is an example of a stress test report that we set up for a client. It took between 4 and 8 hours to prepare the stress test from their systems but only seconds once the Stress Test was set up in Madison.

Here is what they wanted to do.

  1. Look at the portfolio of commercial real estate loans based on four scenarios: the current portfolio with no stress, a 5% stress, a 10 % stress and a 15% stress.
  2. Create one set of stresses based on LTV and another set based on DSCR.
  3. For each stress scenario and for both LTV and DSCR separately, show the results broken out by three stressed ranges from least affected to most adversely affected.
  4. Group the data by the region in which the collateral was located.
  5. If a multi property loan had collateral in more than one region, allocate the loan balance to each region based on the most recent collateral value.

We set this up in a single report that presented all the information requested with the simple click of the mouse. It took seconds to produce rather than hours.

In addition, as additional value-added to the report:

  1. Results could be grouped by many other factors in addition to region – collateral type, loan purpose, loan type, loan officer, loan office, etc.
  2. By clicking to display audit details, all the loan-by-loan details were displayed so that the user could verify accuracy and understand how individual loans were treated. And most importantly, these results let the user see the loans that were most adversely affected by the stresses.
  3. An option also allowed the user to change the scenarios from 5% increments to any other increment they desired.

Stress testing is important, but bankers should be spending their time evaluating stress test results, not manipulating raw data to create the analysis. This custom report allowed the client to do just that.