Personal Finance

‘Finance theory needs a revolution’

Anand Kalyanaraman | Updated on January 21, 2019 Published on January 20, 2019

Gerd Gigerenzer, Director, Harding Center for Risk Literacy, Max Planck Institute for Human Development

In an uncertain world, we need precise models that you can simulate and test: Gerd Gigerenzer of Harding Center for Risk Literacy

The simple often leads to much better results than the complex, says Gerd Gigerenzer, Director, Harding Center for Risk Literacy, Max Planck Institute for Human Development. Speaking with The Hindu BusinessLine at CFA Society India ‘9th India Investment Conference’, Gigerenzer explains the importance of heuristics and why less is often more. Edited excerpts:

To begin with, can you explain the difference between risk and uncertainty?

Risk refers to situations where the future is like the past. That is, where you know all future events including their probabilities and consequences. An example would be a lottery or some kinds of games. Uncertainty is when this knowledge is not available; most situations that are interesting are of that kind, say investing or whom to marry or whom to trust. These are all situations of uncertainty. The key insight is that probability theory is a theory that is sufficient for risk but not for uncertainty. For uncertainty, you need other tools like heuristics, and here, less is more.

You advocate simplicity over complexity in investments. Can you elaborate?

A key example here would be Harry Markowitz who won a Nobel Prize for his mean-variance model to construct portfolios. But when it came to his own investments for his retirement, we might think he would have used his own Nobel Prize winning optimisation methods — no, he did not. He used a simple heuristic that is called 1/n. That is, invest equally over n assets. There are studies that show that you can do better, make more money by simplifying. So, 1/n doesn't need any data while Markowitz's optimisation method needs years of data to somehow estimate the weights of assets.

This is a case where no-data can make more money than big data. The interesting question is, can we identify the situations where simple heuristics work? Or where complex methods work?

So, when does simple work and when does complex work?

So, at a very general level, if the world is stable, then fine-tune and do optimisation kind of models. If it is unstable and there are surprises out there, lack of knowledge and more uncertainty, then simplify. More precisely, if the world is fairly unstable, like say the stock market or investments in general, if you have many options or you need to estimate many parameters and relatively small samples, this is a situation where you need to simplify. Otherwise, you get worse results.

Literature in finance is mostly about complex models which are not easy for an ordinary investor to understand. Is there a need for change?

Finance theory needs a revolution. And the revolution needs to start with the insight that we are dealing with an uncertain world, and get out from the illusion that we have to deal with a world that is foreseeable. So, you need different kinds of mathematical tools. One class of tools is heuristics. By heuristics, I mean precise models, not words like 'availability' or 'representativeness' where nobody knows exactly what it means.

We need precise models that you can simulate and test. 1/n is an example. Further, there are many other heuristics. For instance, one class of heuristics is for situations where there is one powerful cue that dominates everything else. In such a situation, you can forget about all big data, machine learning and complex models because the moment you have identified that one cue, you know that complex models cannot do better. They may do worse because they introduce more error from over-fitting.

Financial literacy is important. We know from studies that just teaching a concept doesn't help much. But teaching heuristics may help. So, for instance, put a third of the money in stocks, a third into bonds and a third into real estate. That's a simple heuristic. Second, the investment strategies of professionals should be studied and taken seriously.

There are people like Soros who basically laugh about finance theory as something useless. It would do well for finance theory if they looked closer how these people make money. Third, heuristics could play a useful role at the level of regulation. For instance, replace the complex rules of Basel 3 with transparent simple rules that actually create more safety.

It is important to study the ecological rationality of the risks to find out under what circumstances are they are likely to work. So it is important for any finance model to ask where any mean-variance model will work and where it will not. As opposed to this, most textbooks teach this today as a kind of a holy grail. That's just totally wrong. Also, finance needs to become more evidence based. Respect the evidence rather than running on the prejudice that complex is always better. It's not better.

There has been much discussion about emotional biases and how to overcome them. What are your views?

Many of the so-called biases that behavioural economics and behavioural finance teache us are measured against standards that come from the role of risk. So, in the real world of investment, these may not be biases at all. For instance, you could consider 1/n as a bias, and say that people use that because they are not very smart or their cognitive capacities are low. But this totally misses the point. The point is that in an uncertain world, you need to simplify. It's a smart move, not a stupid move.

What is your opinion about the works of (Nobel laureate) Daniel Kahneman?

Daniel Kahneman and I have had, for many many years, a lively controversy. Where we agree is that people rely on heuristics. Where we do not agree with is whether it's a good or bad thing. So, in Kahneman's thinking, logic and probability theories are always a correct solution. That is not my thinking. Under uncertainty, it is clear that it is incorrect. So, he thinks that heuristics are a second best. While I have shown that in many situations, heuristics are just the most reasonable thing you can do.

The real question is a different one. It's not whether optimisation or probability theories are the best ones, or heuristics is the best one. The question is to define the situations where one of the two work. This is why I talk of a toolbox. We need to have a toolbox where the tool can be adopted to the situation.

Is it the case that the finance and investment industry rely on complexity to maintain its importance and increase business?

Yes. The useless part of the industry would fall away (if things became simple). Look, if we make everything complex, we would decrease safety. For instance, traffic signs are made in a way that they are simple to understand. So, in many situations, we need to make things simple and transparent.

But there is a belief that big data will be always better, more complex will be always better. In finance regulation, we have a situation where the regulatory frameworks like Basel 3 are so complex that even my colleagues from the Bank of England don't understand the consequences. I am working with the Bank of England on alternatives that would be safer and simpler.

Published on January 20, 2019
This article is closed for comments.
Please Email the Editor