Please excuse me (and feel free to delete my previous comment EE Admins!)
—- Some small edits from what I posted above…
Thank you for your talk, Matthew
I read your working paper and I think your talk does a great job of presenting it. Some criticisms:
It seems you hold EE to a higher bar than EUT. For EE the growth rate must be certain (taken at the limit) and provide a guaranteed outcome in order to inform a decision. For EUT you seem to relax this requirement and allow for a spread of possible outcomes. Put another way:
– EE is useful only when it eliminates all risk
– EUT can be useful when evaluating a range of outcomes with a known risk distribution
The additive “risky” gamble you present is certainly useful in illustrating how a “real” growth rate in finite time can be radically different from the “certain” growth rate at the limit. However, for a multiplicative gamble like Peter’s coin toss, the typical growth rate converges in finite time reasonably quickly to approximate the “certain” growth rate at the limit. These two examples seem qualitatively very different.
Thank you for this interesting talk. In my opinion, clarifying the differences and similarities between EUT and EE is a really important task, and though I do not agree with your overall conclusion, I think your contribution is very valuable. Here are two comments:
First, I think there is a fundamental difference between the epistemological background of the two theories, which you do not seem to mention: EUT (and its subsequent variations such as prospect theory) attempts to assign mathematical formulations to human behavior according to observed preferences. On the contrary, EE’s emphasis on growth rate maximisation stems from the idea that economic and statistical intuition has emerged through the process of evolution, in which natural selection has favored specific preferences. In other words, EE is build on A Priori knowledge of the world, whereas utility functions can be arbitrary. In your introduction, you say that both EE and EUT suggests that ‘more is better’ but they do so for very different reasons.
Second, all models are wrong but some are useful. I think that the main problem with the current state of EE (with its focus on wealth processes) is that it is not useful for a wide range of standard economic problems – for example the consumption-savings decisions. Utility can be derived from all kind of things (consumption, leisure time, environmental goods, social cohesion ect.), not just wealth, and hence, utility theory can be informative on a much wider range of economic problems.
I am looking very much forward to the debate at the conference!
I’m sorry we weren’t able to get to this question in the live discussion!
EE and EUT are indeed very different theories on a fundamental level, so what you’re saying here is certainly correct. That said, if we observe behaviour that is evolutionarily-driven and use that to build a model then arguably we are doing something similar (but more robust) to using our a priori knowledge. So perhaps the differences aren’t as great as they appear.
I also think that *all* EE gets us is a justification for ‘more is better’ — as I discuss, for finite cases where uncertainty remains it becomes much less persuasive. (And even for some infinite cases…) I can see that it’s nice to have a theoretical justification for why more is better, but I don’t think this ranked as one of the great mysteries of the human condition!
Gah! I submitted that previous comment too soon!
Please excuse me (and feel free to delete my previous comment EE Admins!)
—- Some small edits from what I posted above…
Thank you for your talk, Matthew
I read your working paper and I think your talk does a great job of presenting it. Some criticisms:
It seems you hold EE to a higher bar than EUT. For EE the growth rate must be certain (taken at the limit) and provide a guaranteed outcome in order to inform a decision. For EUT you seem to relax this requirement and allow for a spread of possible outcomes. Put another way:
– EE is useful only when it eliminates all risk
– EUT can be useful when evaluating a range of outcomes with a known risk distribution
The additive “risky” gamble you present is certainly useful in illustrating how a “real” growth rate in finite time can be radically different from the “certain” growth rate at the limit. However, for a multiplicative gamble like Peter’s coin toss, the typical growth rate converges in finite time reasonably quickly to approximate the “certain” growth rate at the limit. These two examples seem qualitatively very different.
Thank you for this interesting talk. In my opinion, clarifying the differences and similarities between EUT and EE is a really important task, and though I do not agree with your overall conclusion, I think your contribution is very valuable. Here are two comments:
First, I think there is a fundamental difference between the epistemological background of the two theories, which you do not seem to mention: EUT (and its subsequent variations such as prospect theory) attempts to assign mathematical formulations to human behavior according to observed preferences. On the contrary, EE’s emphasis on growth rate maximisation stems from the idea that economic and statistical intuition has emerged through the process of evolution, in which natural selection has favored specific preferences. In other words, EE is build on A Priori knowledge of the world, whereas utility functions can be arbitrary. In your introduction, you say that both EE and EUT suggests that ‘more is better’ but they do so for very different reasons.
Second, all models are wrong but some are useful. I think that the main problem with the current state of EE (with its focus on wealth processes) is that it is not useful for a wide range of standard economic problems – for example the consumption-savings decisions. Utility can be derived from all kind of things (consumption, leisure time, environmental goods, social cohesion ect.), not just wealth, and hence, utility theory can be informative on a much wider range of economic problems.
I am looking very much forward to the debate at the conference!
I’m sorry we weren’t able to get to this question in the live discussion!
EE and EUT are indeed very different theories on a fundamental level, so what you’re saying here is certainly correct. That said, if we observe behaviour that is evolutionarily-driven and use that to build a model then arguably we are doing something similar (but more robust) to using our a priori knowledge. So perhaps the differences aren’t as great as they appear.
I also think that *all* EE gets us is a justification for ‘more is better’ — as I discuss, for finite cases where uncertainty remains it becomes much less persuasive. (And even for some infinite cases…) I can see that it’s nice to have a theoretical justification for why more is better, but I don’t think this ranked as one of the great mysteries of the human condition!
If anyone wants to learn more about this, the working paper is available here: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4140625
I’m @matthewcford1 on Twitter and available at matthewcford@icloud.com