5
$\begingroup$

While working on different research projects I got fascinated by RL which got applied to many fields that are focused on agent based modeling. Though it is a field in Machine Learning what interests me is the overlap between that apporach with problems in economics. The latest awesome paper I know is the "AI Economist" by Socher et al.. Do you know more fields/papers where RL is applied to economic decision making? And do you think that this approach can lead to a paradigm shift in economics ?

$\endgroup$
5
$\begingroup$

I would say applications of RL in economics are still in their infancy, though probably have applications in many state-based models, agent-based models, or fields with repeated interactions like game theory.

There has been some work done on what you might call the "microfoundations" part - which is studying "regret". Susan Athey, for example, does a lot of work in this space, and an example paper is "Policy Learning with Observational Data" found here: https://arxiv.org/abs/1702.02896

There are also some papers that shoes RL results converge to what one might obtain via rational expectations / equilibrium. For example, "Learning leads to bounded rationality and the evolution of cognitive bias in public goods games" by Leimar, O. & McNamara, J. M.

$\endgroup$
1
  • $\begingroup$ Thanks! Ill read that papers. To me it looks like RL is quite a low hanging fruit in economic research... $\endgroup$ – Nikolai Kl Mar 5 at 6:41
2
$\begingroup$

Given the set-up of the problem in reinforcement learning there is also a relatively close link with evolutionary stable strategies in game theory. Granted evolutionary game theory is not economics per se, but there are overlaps.

See for example this paper by Tuyls et al. on the relation between multi agent systems, reinforcement learning and evolutionary game theory

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.