New & Noteworthy

CAMELSolutions New Article of the Month - Monday, March 14, 2011

Read more...
 

 

CAMELSolutions New Article of the Month - Thursday, April 29, 2010

Read more...
 

CAMELSolutions New Article of the Month - Friday, March 5, 2010

Read more...
 
Client & Guest Log-In

2012 Bank Closure Forecasts:

Viewpoints from the CAMELSolutions Principals

 

From Joe Bartolotta:  54 Failures in 2012 

Basking in the glow of developing the most accurate forecast for closings in 2011 and reveling in the accolades from my CAMELSolutions colleagues (cue the sound of crickets chirping…) I’m confident in extending my winning streak to two with a forecast of 54 bank closings in 2012.

The model that I used in 2011 had the benefit of 2 years of weekly closing data  - and its accuracy improved significantly over earlier versions because of that solid history.  But there were “non-linear irregularities” [read:  events that didn’t follow the expected trend] which popped up from time to time during the year which required “heuristic adjustments” to the model’s weekly prediction [read:  good ol’ Kentucky windage].

The good news is that it became evident to me that these adjustments did indeed have a discernable pattern:  but it was not directly related to what would typically be thought of as a ‘driver’ variable like the number of institutions on the Troubled Bank list, Fed Beige Book data, or even the Dow Jones Industrial Average.   These deviations from the model’s predictions were instead tied to factors such as changes in the FDIC’s senior leadership; holiday / vacation periods; and perhaps the hardest of all to quantify, the Federal political landscape.

It’s this last factor – which seemed to have been gaining influence over the closure rate throughout the 4th quarter of 2011 - that is most strongly influencing my prediction for this year.  It’s not that the Feds will refrain from closing a bank that truly represents a threat to depositor safety.  But perhaps some will benefit from a period of ‘watchful waiting’.  That technique will help to prevent further erosion of the Insurance Fund balance, and more importantly, keep the number of failed banks low.  In the run-up to November’s elections, both of these effects will be seen by some voters as a sign that the economy is improving – an extremely powerful perception that’s difficult to replicate elsewhere.

How does this non-parametric, non-financial “factor” co-exist with a model which uses historic data to predict future bank closures each week?  A model is ultimately just a decision support tool:  it doesn’t create knowledge, judgment or experience any more than handing a man a scalpel makes him a surgeon.  But the combination of a sharp tool, a healthy dose of experience, and just the right amount of cynicism leads me to think that we’ll see just 54 closures this year.