The Influence of Investor Sentiment on Mutual Fund Portfolio Composition

Abstract

This thesis addresses the relationship between investor sentiment and mutual fund portfolio composition and investigates whether prospect theory explains the behaviours of institutional investors like mutual funds. Most of the existing literature focuses on mutual funds’ performance related to investor sentiment or uses a mutual fund-based sentiment proxy. There is limited research on how investor sentiment impacts the composition of mutual fund portfolios. To fill this gap, this study creates a new composite investor sentiment index that includes survey-based, market-based, and internet-based sentiment indices and compares the effectiveness of the various sentiment measurement methods in reflecting mutual fund portfolio composition. This focus aligns with the broader objective of elucidating the behavioural underpinnings of investment decisions in the face of market uncertainties.

In the first empirical chapter, I examine the different effect of multiple investor sentiment proxies on mutual fund portfolios based on the panel ARDL method. The portfolio compositions are represented by two dependent variables: the percentage of capital invested in the common stock market and the percentage of cash holdings in mutual funds, respectively. The results indicate that all investor sentiment indicators will have an influence on mutual fund portfolio composition and their investment decisions; however, some of them have a more profound effect during crisis periods than stable periods. In addition, the results suggest that the composite sentiment index outperforms single indices when impacting mutual funds’ portfolios, and the survey-based index will significantly influence mutual funds’ investment decisions in both crisis and stable periods. Nevertheless, the internet-based and market-based indices have a partial effect when the market is in crisis or remains stable, but they still carry information about how they significantly influence mutual fund portfolio composition.
The second empirical chapter applies the nonlinear panel ARDL method to investigate asymmetric effects of investor sentiment on mutual fund portfolio composition. In addition, I investigate whether the prospect theory explains the behaviours of institutional investors. The empirical results imply that mutual funds may have different reactions to increases and decreases in investor sentiment. The observed reactions can be rationalized using the prospect theory, which proposes that people may have different subjective valuations towards gains and losses. This indicates that mutual fund investors might value losses more than gains and suffer from loss aversion bias.

The last empirical chapter employs the principal component analysis (PCA) method to build a composite investor sentiment index, consisting of various sentiment proxies such as surveys, the Internet, and market-based measures. The proposed index may carry more information than the well-known BW Index, constructed based on pure market-based sentiment indices. Furthermore, this chapter also applies panel ARDL and nonlinear panel ARDL methods to investigate the influence of the new PCA composite index on mutual fund portfolio composition. The results indicate that the new sentiment index has a significant correlation with both categories of mutual fund portfolio allocation and has a strong relationship than the BW Index. The results of the nonlinear analysis also indicate that mutual funds react to this index asymmetrically and consistently with the value function proposed by prospect theory.

The collective results suggest that all investor sentiment indicators can significantly impact mutual fund portfolio composition. Shifts in investor sentiment, whether positive or negative, have disparate effects on mutual funds’ allocations of assets between stocks and cash. These findings may have important implications for investing decisions and portfolio management strategies.

Date of Award	2025
Original language	English
Awarding Institution	University of Dundee
Supervisor	Martin Jones (Supervisor) & Andrzej Kwiatkowski (Supervisor)