Correlation Is NOT Causation
Something seemingly innocuous happened last week—my wife stubbed her toe. After a day of swelling and discoloration, she went to the doctor. He examined the toe, and concluded that it wasn’t broken. His advice was to go home and put ice on it for a few days.
Later that week, my wife wakes up in the middle of the night with her toe throbbing. Getting out of bed to get an aspirin, she knocks a glass of water off the bedside table. It shatters on the floor and a piece of glass catches the top of her already-injured foot. The next morning, back to the doctor who said it was just a small cut.
Flash-forward a week her toe was visibly worse. Of course, we both drew the immediate conclusion that her toe was broken and that she had been misdiagnosed. So we went to a podiatrist for an MRI. No broken toe. She has a severed tendon and needs surgery.
So what’s the point? We convinced ourselves that the problem was a broken toe, caused by kicking the coffee table, which was cemented by all the correlating facts. We believed that all the evidence corroborated this view. When in reality, a piece of broken glass severed a tendon, which later tore after she walked on it.
What does this story teach us about investing? Our minds look for ways to associate events in order to establish causation. This correlation mind set is ingrained in us. Looking backward, correlation factors prove only one thing in my view and that is that there is a benefit in diversifying. What it does NOT prove is that any particular set of asset classes will exhibit the same correlation characteristics in the future. I see many investors, advisors and robo-advisors building models based on backward-looking assumptions that completely ignore the root “causation” that made those relationships achieve the desired result during that time period.
In my mind, the best advice is to recognize that causation is what matters. In the financial world, everyone has choices about where to under- or over-weight capital. This rotation between asset classes is not limited to just public equities and public debt, but to a whole host of investable options. This correlation phenomenon is the causation, in my view and it is the target we should be aiming for. No one can predict this correlation alignment between assets, but investing across all options (or as many as one can) helps us to get closer to capturing the driving causality behind disparate risk-and-return streams of asset classes over time.
So, a broken toe that wasn’t broken led to a glass of water falling that led to an insignificant cut that caused a torn tendon resulting in a crooked toe. While we believed that a coffee table caused the whole problem—and adapted our behavior in an effort to avoid that piece of furniture because we believed that it was the causality—it was only a minor contributor. Does this anecdote sound eerily similar to the way that many investors and even many financial pundits view the financial world?
Send a Comment
Your feedback is important to us. We encourage you to send us your comments. Although we cannot provide direct responses or post comments online, all feedback will be evaluated carefully. Your responses could influence the development of future Perspectives articles.