BID® Daily Newsletter
Apr 16, 2010

BID® Daily Newsletter

Apr 16, 2010

MAGIC ANSWERS FOR MODELING TROUBLED TIMES


The Magic 8-Ball was invented in 1946 and is sold by Mattel to the delight of kids and adults worldwide. This advice-giving toy provides entertainment for hours and can often be found lurking around offices as well as in homes. Given all the turmoil in the industry as of late, bankers could certainly use a tool that accurately predicts the future. Unfortunately for us all, we only have to look at the past 2Ys to know that forecasting the future is impossible. After all, no one predicted the carnage of the credit crisis, the number of companies that would fail or the ultimate impact on the country and indeed the world.
That doesn't mean modeling isn't endemic in the banking industry. After all, bankers use models all day long in their business. Whether the models handle budgeting, project cashflows, analyze risk or simply calculate the interest due on an outstanding loan, models are prolific in the industry.
To use models better, we have to understand that a primary problem is that most of them depend on historical data to project the future. The problem with using historical data when the economy isn't very stable is that many older relationships simply don't hold up very well so the entire process can break down. Sprinkle in additional complexity along with that basic underlying problem and results can even be more skewed to actual experience as time goes by.
In short, it is very hard to predict the future when so many things are moving around. Model designers are the first ones to tell you that most models are designed for "normal" periods of time where things are pretty stable, but are not designed to handle significant shifts. Time is also the bane of modeling accuracy, since the longer the prediction, the worse it usually gets. Bankers that use models to forecast performance over periods beyond 3Ys know the actual result won't be anything like the modeled prediction. In abnormal times, even short-term forecasting can be very off the mark because many assumptions that feed models are based on the expectation that things will operate within reasonable bands.
To truly get a decent process underway, one has to recognize that all models are inherently flawed and while the future can be like the past, it is never exactly the same. It is also important to recognize that humans operating the models are also inherently flawed. We like to make sense of things to feel more in control, so we tend to extrapolate patterns and relationships from the data that may not exist.
We run lots of models for lots of bankers and experience leaves us with some tips to pass along. First, there are many sophisticated models that can't predict their way out of a paper bag. Just because a model is complex doesn't mean it is inherently better than one that is simple. Next, managers should report and understand the imprecision embedded in any forecast. Finally, humans run models and humans are all horrible at making predictions. Studies find so-called experts will not predict events any more accurately than moderately well informed regular people. We want a single answer because it is easier to communicate, present and discuss, but the only real answer is that the Magic 8-Ball may be the best model we already own.
Modeling is important to banks, so understanding some weaknesses of the process is crucial. Bankers must deal with uncertainty on a daily basis, so developing strategies and plans around an evolving business model is the best way we have found to date to deal with things. As we shake the Magic 8-Ball one more time, we ask it whether we should stop believing our predictions about the future and shift to doing business in a way that can respond to surprises. The answer is a resounding "Yes."
Subscribe to the BID Daily Newsletter to have it delivered by email daily.