In this series of post, we’ve considered three criticisms of behavioural economics, they are:
- Confusing ensemble probability with time series probability
- Assuming that rules-of-thumb (heuristics) are always, lazy, dumb and wrong
- Using a two-system model to create a false dichotomy between heuristics and other decision-making models
The key point common to all three criticisms is this: we can’t assume that people are behaving in a biased way or making wrong decisions just because they don’t follow the rules probability or use complex models.
There is considerable evidence that simple rules-of-thumb often result in faster and more accurate decisions. For example, we considered the 1/N diversification heuristic. Even Harry Markowitz, one of the fathers of modern portfolio theory uses this heuristic when constructing his own portfolio.
We shouldn’t just dismiss heuristics as biased, lazy and dumb “system one” thinking. Instead, we need a framework to help us to decide when and how to choose between using heuristics or more complicated methods.
Professor Gerd Gigerenzer, has featured throughout this series. In his book: ‘Risk Savvy – How to Make Good Decisions‘, Gigerenzer offers a simple three-variable framework:
How far we go in simplifying depends on three features. First, the more uncertainty, the more we should make it simple. The less uncertainty, the more complex it should be. The stock market is highly uncertain in the sense that it is extremely unpredictable… Second, the more alternatives, the more that we should simplify, the fewer, the more complex it can be. The reason is that complex methods are needed to estimate risk factors and more alternatives mean that more factors need to be estimated, which leads to more estimation errors being made… Finally, the more past data there are, the more beneficial for the complex methods… The various factors work together.
One of the reasons why I find Gigerenzer’s critique compelling is his use of a framework. His focus seems to be on helping decision-makers select the right tool for the job.
Gigerenzer’s framework considers three variables: uncertainty, the number of alternatives to choose from and the amount of data that’s available.
Uncertainty is not risk. Risk can be measured and to some extent modelled, when uncertainty can’t
Uncertainty is not risk. Risk can be measured and to some extent modelled, when uncertainty can’t. For example, a game of Roulette is high on risk but low on uncertainty. There’s a chance that you can lose money, but it’s possible to calculate that probability. Importantly, the probability remains the same each time that you play.
In contrast, uncertainty cannot be calculated. The distribution of possible outcomes is unknown and it changes over time. As Gigerenzer points out, investing is a highly uncertain activity.
Choice increases the number of estimates required. Remember these estimates are about the future and the future’s uncertain. More estimates equal more errors and bigger interaction effects between estimates. That’s why simpler models work better as the number of alternatives increases.
More data helps because it allows researchers to consider a wider range of circumstances. It also makes it possible to do more out-of-sample testing.
The last comment, “The various factors work together”, is important. For example, a smaller number of choices means that less data is required for a complex model to be appropriate. Or less uncertainty means that more alternatives can be considered without compromising the performance of a complex decision-making model.
So how does investing score using Gigerenzer’s framework?
- Uncertainty = high
- Alternatives = many
- Data = not enough given the uncertainty and number of choices
This argues in favour of using simple decision-making rules or heuristics over complicated models.
Principal and Agent Conflicts
Why don’t more investors use simple rules to make investment decisions? In Risk Savvy, Gigerenzer tells the story of a keynote presentation that he gave at the Morningstar Investment Conference. In his presentation, he explained the merits of using simple rules-of-thumb such as the 1/N diversification heuristic.
Afterwards, Gigerenzer was approached by a member of the audience. This person was the head of investment at a major global insurance company. He offered to test the 1/N heuristic against the results of the insurance company’s investment portfolio.
Three weeks later, the head of the investment division of the insurance company came to see Gigerenzer at his office. Here’s what he said:
I checked our investments beginning 1969. I compared 1/N to our actual investment strategies. We would have made more money if we had used this simple rule of thumb.
Then he added:
I have convinced myself that simple is better. But here’s my problem. How do I explain this to my customers? They might say, I can do that myself!
Anyone who’s worked at a large investment firm has seen defensive decision-making on a daily basis. Gigerenzer defines defensive decision-making this way:
A person or group ranks option A as the best for the situation, but chooses an inferior option B to protect itself in case something goes wrong.
Imagine that an institution is reviewing their multi-manager global equity portfolio. The portfolio consists of eight funds, all benchmarked to the MSCI All Country World Index. What would happen if the internal team simply recommended equally weighting (i.e. 12.5% in each) its allocation across the eight funds?
The trustees would ask the investment team why they have made this recommendation. The investment team would answer that high-uncertainty, a wide range of choices and lack of data meant that it was probably best to simply divide the allocation equally.
This is clearly an imaginary situation. Which investment team in their right mind would risk their jobs to do this?
Instead, they do a qualitative and quantitative review, engage an asset consultant to provide advice, use a qualitative risk model, etc.
Institutions that are intellectually honest and genuine about making better decisions should follow the example of the insurance company executive cited above. They should compare the performance of their portfolios against an equally-weighted portfolio. If they’re adding value, great. If not, simplify.
There are good reasons to conclude that investors aren’t necessarily as lazy, dumb and wrong as they’re sometimes portrayed by behavioural finance. Previous posts have examined three reasons why this is the case.
Investors need a framework to help them decide when to use a simple rule of thumb or a complex model. Professor Gerd Gigerenzer suggests that we should consider three factors when making this judgement: the level of uncertainty, the number of alternatives and the amount of data. As helpful as this framework is, it’s not enough. Especially in the case of institutional investors.
Institutions also need to manage the impact of conflicts of interest and the risk of defensive decision-making. Otherwise, they will be biased towards using complicated models that yield inferior results.
 The other is A.D. Roy who also published paper in 1952 entitled “Safety First and the Holding of Assets.” Econometrica, vol. 20, no. 3 (July):431–449.