A multifactor model of investment trust discounts.

Abstract

A closed-end fund, known as an investment trust in the UK and closed-end fund in the US, is a collective investment company that invests in shares of other companies. This study attempts to describe and explain the persistence of the excess discount return on UK investment trusts and US closed-end funds. The ability to identify which factors best capture return variation is central to applications of multifactor pricing models. So the main purpose of this thesis is the application of a multifactor risk model that will explain the-existence of the excess discount return. Hence, the title of the thesis: "A Multifactor Model of Investment Trust Discounts. A Comparative Study of UK Investment Trusts and US Closed-End Funds" First, the time-series properties of the closed-end funds' net asset values (NAVs) and discounts are investigated. In terms of normality, we find that the UK and US excess NAV returns and discounts are approximately normally distributed. In addition, through Augmented Dickey-Fuller tests, we find that the UK and US discounts are non-stationary, but the excess discount returns and the excess NAV returns are stationary. In terms of multicollinearity, we find that the independent variables included in our models are not closely correlated, so we do not have problems in using them in the regression models in Chapters 7 and 8. Finally, there are no significant differences in the discount during the month of January and other months. In Chapter 7, we study the importance of management performance in terms of excess NAV returns and discount persistence. We use three approaches: Fama and French's (1993) three-factor model, an extended Fama and French model which incorporates a market timing variable, and a performance persistence model used by Carhart (1997) and Dimson and Minio-Kozerski (2001). On average, the six-factor model developed in the thesis can explain 67% of the variation in the excess discount return in the UK market by taking into consideration the market effect, size, the book-to-market effect, momentum, sentiment and expenses. In contrast, Fama and French's (1993) three-factor and Carhart's (1997) four-factor models explain only 42% of the variation of the excess discount return. Similarly, the six-factor model can explain 66% of the variation in the excess discount return in the US market by taking into consideration the same six independent variables. In contrast, Fama and French's (1993) three-factor model explains 59% of the excess discount return variation and Carhart's (1997) four-factor model explains 65% of the variation.