A fair amount of the risk management consulting work I do is on the “hard” side of risk management; building and reviewing models, establishing KRA’s, stress testing, regulatory requirements, etc… But in reality risk and risk management are soft subjects resulting in approximations, guesstimations, probabilities and other “uncertainties”. These uncertainties result in people having to make decisions about how to act.
How you define things leads to your results…whether they are right or not
Let me give you an example. We were recently working with a bank that wants to transition from a single factor credit rating system that is only applied to the commercial portfolio to a dual factor credit rating system that can be applied to all loans. One of the first topics that came up for discussion was, “What is a loan default?”. At first glance, this is a fairly easy thing to define. Until you look at all of the possibilities. A first possibility is passing a regulatory threshold of, say, 120 days past due (for closed end mortgage loans). At this point the loan is required to be charged off. Another possibility is that the bank has exhausted all avenues to get repaid on the credit and decides to charge off the credit. This could take place over many years or could occur quite quickly. The time element is very important.
But what about all of the additional work the bank has to put into a delinquent loan? At this point it is no longer very profitable to manage a delinquent credit. Even if the loan is foreclosed on, the bank can only be made whole on the credit (including costs). So maybe a default should be thought of as the point when the borrower is past due X number of days or is graded at a substandard level. That is when the work begins. I have often said that a loan should really be treated as a defaulted loan much earlier than the point of charge off. Bank failures happen because the bank is out of money or go below the defined level of capital, not when the credit is charged off and liquidated. But this is just me talking.
Let’s not forget about Loss Given Default (LGD) and the true loss associated with the loan. This involves the recovery aspects, which could take place many years into the future. One bank I work with saw there recoveries go up several years after the Great Recession because their customers really did want to pay them back. How do you define the net loss from default when you may get paid back some day, but you don’t know when that day will be?
The Illusion of Certainty
What I am getting at with this convoluted example is that ultimately risk is a fuzzy topic that plays on the illusion of certainty. The phrase, “the illusion of certainty”, I have borrowed from the book, Risk Savvy, How to Make Good Decisions (2015) by Gerd Gigerenzer. It is another book that explores the psychology of risk management. The principal lesson from this book is that we need to be taught how to interpret risk based statistics because they can be misleading, misinterpreted or just plain wrong and they are never precise. Take a single $100 loan with a 1% PD and a 50% LGD. To calculate the Expected Loss (EL) we basically take PD times LGD. So for the $100 loan we expect to lose 50 cents. Even though it will be a rare day when that is what the actual loss is.
What is the chance of Rain?
Another example, and one that may be less convoluted, comes from the weather forecasting arena. This example is cited in the book. What is the chance of rain? We all hear this discussed on TV. A 50% chance, 20%, 80%…what does it actually mean and how do we interpret it. When I hear there is an 80% chance of rain I intuitively bump it up to 100% and think there will be a nice little downpour. And when I hear a 20% chance, I bump it down to zero. Why do I do that? Because of the fuzzy nature of risk. When a forecaster says there is a 30% chance of rain, they are saying that considering similar days in the past, it has rained on 3 out of every 10 days. So on the 3 (past) days there was a 100% incident of rain and on 7 of those days there was a 0% incident of rain. Note that the time unit is days. If it first rains after midnight then it is on the next day. How much rain is needed? As little as 0.01 inches. Where is the forecaster predicting rain? They are not looking at the little black raincloud that sits above my head but rather the entire viewing area. As I said, it is all rather fuzzy.
This fuzziness leads us to personally assess risks in many different ways. And if our behavior is based on incorrect or misinterpreted information, then we are putting ourselves, or our organization, or others into a situation that is much more poorly understood than we think it is.
Obligation to understand risk statistics
I believe it is our obligation to understand how risk is measured, what it means and then educate others. Our brains need this kind of knowledge in order to reduce the fear we experience of the unknown, make certain we aren’t being lied to and enable us to better interpret what we are being told in this fast paced . In some speeches or lectures, I will cite the example that, “50% of the people make up ½ of the population”. I use it to demonstrate that something that is said with authority (I use my best baritone authoritative voice) doesn’t necessarily mean anything or add any value.
It is good to take a break from reading books with formulas and be able to delve into and share information on behavioral risk management. I hope you find some nugget in this blog that helps you stay safe and make good decisions.