Wednesday, January 15, 2020

The Gambler’s Fallacy and How Credit Decisions are Affected by Human Bias

Life is complicated, and we are busy.  People just don’t have the time or the energy to analyze everything.  So, we employ “heuristics” or “rules of thumb” to help us make judgments.  While these rules of thumb are often helpful, they also cause systematic biases which adversely affect our decisions.  This happens to all of us at home and at work.

And yes, if you work in any role related to credit underwriting, the chances are that these heuristics affect your decisions and the decisions of your coworkers.  Here are a few of the most common biases:

  • Availability: Our judgments are affected by the ease of remembering or retrieving events, and more vivid or more recent events tend to be more memorable.
  • Regression to the Mean: We assume that we can predict the future from the past, that this year’s cash flow will be similar to last year’s cash flow.  However, any extreme performance is likely to regress to the mean.
  • Confirmation Bias: We tend to selectively search for information which supports the conclusion which we desire to reach (or which is located wherever we have found the information previously).
  • Overconfidence: People tend to be overconfident in the accuracy of their predictions.
  • Hindsight Bias: We assess the quality of a decision based upon whether the outcome was good or bad, not by whether the original decision-making process was sound.

The Gambler’s Fallacy is another bias, and this one has been shown by academic research to impact credit decisions.

People underestimate the likelihood of sequential streaks occurring by chance.  We tend to believe that things will even out.  This is the Gambler’s Fallacy.

For example, in roulette, if red has come up four times in a row, then people are more likely to bet on black.

If a Major League umpire has just called a strike, then he is slightly less likely to call a strike on the next pitch (after controlling for the actual position of the ball).

And, a study of lenders in India found that when an officer approved a loan, he or she was 8% more likely to decline the next loan application.

How can you combat this bias?  First, learn about it.  For an important decision, develop an algorithm that helps you decide, and have another set of eyes examine the outcome.  The right types of incentives can help too.  In fact, the lending study found that adverse effects were reduced when stronger incentives for accuracy (versus the number of approvals) were implemented.

If you are interested in learning more, then please contact me here.  I have been researching behavioral science for a few years along with reviewing actual loan approval processes and decisions.  I have lots of interesting examples and solutions (with a focus on commercial loans).

I will begin formally presenting this information later in the year, but in the meantime, I’d like to perform a couple of free trial presentations.  If your institution is interested, then please contact me.


1 comment: