Abstract
The finance landscape is evolving, with recent years witnessing a significant shift towards integrating advanced technologies and sustainable considerations into financial practices. These trends are driven by technological advancements and a heightened awareness of environmental, social, and governance (ESG) issues. This thesis is situated at this nexus, primarily focusing on three key investing issues: the application of machine learning in investment decision-making, especially portfolio optimization, the exploration of sustainable investing within mutual funds, and the impact of uncertainty on capital flows channeled through mutual funds.First, I focus on optimal asset allocation, the central concern of institutional investors. In this context, traditional portfolio optimization models (i.e., the seminal Markovitz mean-variance framework) lack attraction in practice due to parameter uncertainty. Therefore, I am committed to improving them with advanced machine learning methods and generating portfolios with superior out-of-sample performance. Specifically, I refine and revitalize the classical Bayes-Stein portfolio optimization model by leveraging well-tailored and explainable machine learning techniques. This work sheds light on how machine learning can overcome limitations and unlock value in conventional finance models.
Second, I delve into two research questions related to mutual funds. On the one hand, I explore actively managed US equity funds and investigate investors’ reactions towards two signals of fund sustainability, namely the ESG label and the Morningstar sustainability rating. I find that "silent" funds (top sustainability-rated ones without ESG labels) achieve comparable returns and better mitigate risks than ESG-branded counterparts. This suggests an overreaction by investors toward ESG labels, overshadowing the intrinsic value of sustainability within "silent" funds. Therefore, this study offers implications for industry and policymakers regarding marketing and defining ESG investing. On the other hand, I comprehensively explore the impact of different risk and uncertainty perspectives, including the geopolitical risk (GPR) index, economic policy uncertainty (EPU) index, and world uncertainty index (WUI), on capital flows measured by mutual fund flows. This study offers novel insights into how different dimensions of risk and uncertainty can shape capital flows differently and whether the compounding effect of risk and uncertainty differs from that of a single risk or uncertainty.
In a nutshell, this thesis contributes to portfolio optimization by incorporating innovative machine learning techniques from a theoretical standpoint and offers fresh insights about mutual funds regarding sustainable investing and capital flows from an empirical perspective, making it with not only financial value but also social impact.
Date of Award | 22 May 2024 |
---|---|
Original language | English |
Awarding Institution |
|
Supervisor | Dimitrios Gounopoulos (Supervisor) & David Newton (Supervisor) |