Berkshire Hathaway had their annual shareholders’ meeting last Saturday (May 2), and Warren Buffett and Charlie Munger totally hated on “higher-order” mathematics used in finance. Come on guys, what did little ol’ math do to you? Math and modern portfolio theory were picked on by these investment gurus more than Arnold was picked on by the Gooch! Don’t worry math, I got your back.
The truth of the matter is while Mr. Buffet and Mr. Munger are right about Wall Street’s reliance on complex math, the real blame should be focused on the consultants and investment managers who hawked these models as the end-all, be-all, best thing since sliced bread. This is one case where it is totally fine to shoot the messenger in the face, however, we shouldn’t abandon using math to help us make better decisions. We just need to find a better translator, because the message has some very valuable insights.
The reason why we build financial models, or really any models, is to keep track of numerous and complex relationships, something that is very difficult to do in our heads. The world does not move in simple, predictable ways and the real value in modeling frameworks is to find the best representation of how the world actually behaves. Sometimes a simple relationship just doesn’t make sense; Mr. Buffet would surely agree that modeling investment growth as a simple linear change is not nearly a good as modeling it as an exponential change (there are a number of high school curriculum that consider this “higher math”).
The key is to fully understand and make transparent that as we increase complexities in models, we increase the number of things that can go wrong, and therefore decrease our certainty of performance. Think back to our first calculator, which for a lot of us often doubled as our first watch (wicked). Simple, easy, and reliable. Now add in a 2.66Ghz Intel Processor, 8GB RAM, 320GB of Storage, and a super-fly, aluminum cased, glow in the dark keyboard. We have a kick-ass laptop that let’s us do all sorts of things a whole lot better, but it’s not surprising that its average lifespan is somewhere around 2 – 4 years. And when it goes, we lose everything (yes, even that awesome illegally downloaded music collection that was the envy of our less tech savvy and risk adverse friends). The funny thing is that Casio can still multiply two five-digit numbers, even after 20+ years! But that doesn’t make it better.
Unfortunately, the certainty of performance only really bothers us in the worst of times, like when our computers crash and the stock market collapses. Now, just like backing up our hard-drives, there are ways that we can create more security around financial modeling. A few things that come to mind are good stress testing frameworks (if your models can’t do this easily for you, then be very cautious with its results), putting good translators (i.e., people who get how the model works AND understand its limitations) in front of decision makers early and often, and moving to a risk-based incentive compensation model (a discussion for another time).
Modeling frameworks are very useful, but they shouldn’t be used as a reason to stop thinking about what we are doing. The human element in analyzing data can never be replaced by a pure modeling framework. We shouldn’t site blantent disregard of rational thought by high-paid consultants and star investment analysts as failures in mathematical modeling. Because remember, when you point your finger at your model, there are three fingers pointing back at you … wait for it …. wait for it … okay, you got it, cool.