Markov regime-switching models

: implications for dynamic and long-short strategies

Abstract

Conventional wisdom posits that investors should refrain from market timing. Challenging this, recent work using Markov regime-switching models argues that the market and a set of other risk premiums can be effectively timed and thus investors can avoid periods of high volatility diminishing losses, whilst still taking advantage of positive returns in a stable economy. Existing literature focuses mostly on in-sample predictability. This thesis tests those conclusions in an out-of-sample setting, contributing to existing literature by further investigating how to time the market’s conditions using a Markov regime-switching model. More specifically, I estimate regime presence based on different variables, such as market turbulence, inflation, and economic growth. Regimes relate to either a highly volatile state, defined by economic contraction, or a more stable state, defined by economic growth. Additionally, in this thesis, I analyze different assets and risk premiums’ performance out-of sample, by comparing both a regime-based dynamic and a long-short strategy against a static one. The regime-based strategies’ allocations are adapted to maximize returns whilst diminishing volatility given the presence of the regimes identified. I find that using a Markov regime-switching model to time the market and adjust portfolio allocations significantly decreases volatility, greatly improving risk-adjusted performance. However, regime-based strategies do not appear to yield higher returns than a simple static allocation. Truly, they struggle to outperform a static strategy and market benchmarks. Nonetheless, some strategies based on market turbulence produce interesting alphas, suggesting that more research on regime-based strategies is warranted.

Date of Award	25 Jan 2024
Original language	English
Awarding Institution	Universidade Católica Portuguesa
Supervisor	Pedro Barroso (Supervisor)