Some of the best client-related content you can read often involves behavioral biases and how consumers react to events, headlines, and success or failure. I plan to sprinkle in a deeper dive into many of these over the next few Advisory Monthly articles. Representative bias is where we will start as I feel it is most relevant to clients who are overly focused on the news. With elections coming up next month, it makes sense to start here.
Representative bias can lead us to incorrect conclusions based on something representing a stereotype of what is expected instead of logical or statistical reasoning.
Base rate neglect is a concept related to representative bias. Ignoring base rates has a particularly insidious effect on how people perceive information. Bayes theorem,[i] albeit counterintuitive, is the mathematical formula we can use to help determine actual probabilities of events, given new information. Many of us already familiar with Bayes theorem got a real-life example in our faces during the pandemic when it came to test results for people who weren’t feeling any symptoms. Given an unrelated example of a test with a 98% true positive rate and a 0.5% prevalence in a general population, one might think that given a positive test, they are 98% likely to have the disease. However, using Bayes theorem, we can calculate that given a positive result on a random person in the population would mean less than a 20% (specifically, 19.76%[ii]) chance of having the disease which is wildly incongruent with the 98% accuracy expectation of the test if simply looking at that number.
In finance, this is more relevant to predictions. Whether they be recession indicators, a pundit, or a new tool to help drive alpha. For example, a screening tool with a 60% true positive rate of picking significantly outperforming stocks sounds great. But if you considered a base rate of 25% of stocks significantly outperforming peers, then the probability of the stock outperforming given a buy signal is only about one in three.
Clients also may hear about a major commentator predicting an imminent recession. Some of these may end up proving correct. However, wrong predictions don’t get many headlines after the fact. You are much more likely to hear the introduction, “Welcome to the show, Talkinghead Peters, who predicted the 2008 crash and is now saying a recession is on the way.” You rarely hear, “Welcome to the show, Talkinghead Peters, who has predicted 11 of the last two recessions, is once again saying that another crash is on the horizon.” It is important to note that recessions don’t happen every year, or even every other year, historically. A chart on this Investopedia article shows 12 recessions since 1948, although there should not be an expectation of recessions occurring at regular intervals.[iii] Base rates should always be a consideration when determining likely outcomes of uncertain events. Recessions can occur because of many causes. When a pundit says something is about to happen, it is important to judge the merit of the case and likelihood of the thesis they are putting out there rather than simply the appeal to authority argument that so in so had a correct prediction in the past.
Similarly, another concept related to representative bias is sample-size neglect. Sample-size neglect often occurs when new information comes into the fold, in a way that invites people to overemphasize it instead of looking at it in the context of a larger sample. For example, you may see headlines of several major companies announcing layoffs and conclude that we may start to see an uptick in the unemployment rate and increasing potential for recession. In reality, yes, layoff announcements aren’t ideal, but it doesn’t mean there is a broad decrease in jobs across the entire economy. It could indicate previous over-hiring or a slowdown in a sector, among other things.
We all experience representative bias, but when you or your clients feel a need to drastically alter long-term plans based on short-term concerns, it is important to address whether the change in strategy is sound, or if you may be succumbing to a form of representative bias.
[i] Wikimedia Foundation. (2024, September 5). Bayes’ theorem. Wikipedia. https://en.wikipedia.org/wiki/Bayes%27_theorem
[ii] Positive test probability from example:
P(E)=(0.98×0.005)+(0.02×0.995) P(E)=0.0049+0.0199=0.0248
Using Bayes’ theorem:
P(H∣E)= (P(E|H) * P(H)) / P(E)
Plugging in the values:
P(H∣E)= (0.98*0.005) / 0.0248 = 0.1976
[iii] Investopedia Team (2024, May 31). US recessions throughout history: Causes and effects. Investopedia. https://www.investopedia.com/articles/economics/08/past-recessions.asp
