Machine-learning a family of solutions to an optimal pension investment problem

TL;DR

This paper introduces a neural network-based approach to solve a family of optimal pension investment problems, allowing for flexible preference modeling and validation against classical methods within a Black-Scholes framework.

Contribution

It presents a novel neural network method for solving parameterized pension investment problems, validated against classical numerical techniques.

Findings

Neural network accurately approximates optimal investment solutions.

Method enables exploration of pension outcomes across different preferences.

Validated within a Black-Scholes model with proven convergence.

Abstract

We use a neural network to identify the optimal solution to a family of optimal investment problems, where the parameters determining an investor's risk and consumption preferences are given as inputs to the neural network in addition to economic variables. This is used to develop a practical tool that can be used to explore how pension outcomes vary with preference parameters. We use a Black-Scholes economic model so that we may validate the accuracy of network using a classical and provably convergent numerical method developed using the duality approach.