Investors increasingly embrace factor investing and the amount of capital allocated to these strategies has grown rapidly in recent years. Factor investing is not a new invention and the topic has been the base of a lot of research since it was documented decades ago. Even though there is a growing awareness among asset owners of the diversification potential of combining factors, the existing literature predominantly examine the factors individually and is often limited to the U.S market. The purpose of this thesis is to investigate how investors best can harvest the equity factor premiums, focusing on a multifactor strategy and tactical factor allocation.
The thesis examines the five common factors: Value, momentum, size, low risk and quality in global equities from 1993 through January 2016, and find in alignment with previous literature, attractive returns in all factors with exception of the size factor. The premiums cannot be explained with traditional risk-based measures, therefore the existence of the factors must rely on unknown risks or arguments based on behavioral finance.
The mean-variance framework is used to transform the factors into a multifactor strategy and finds that the optimal factor weights are heavily dependent on whether or not investors remaining portfolio is taken into account. With a classic portfolio consisting of 60% stocks and 40% bonds investors should allocate more to momentum and value, which is a result of strong negative correlation between the two factors.
Market timing within factor investing is seen as notoriously difficult, and even though it is limited with supportive literature, tactical implementation is widely evaluated among institutional investors. This thesis finds that market timing is, to some degree possible. The future returns of the individual factors are regressed against well-known market indicators and valuation ratios, but even though decent results are achieved, the transformation to a dynamic trading strategy is not straight forward. Lost diversification is not always overcome by timing benefits, but the regression result can be combined with the mean variance framework by estimating a set of dynamic expected returns. This method has, with modest tilts, improved returns historically with 94 basis points.
|Educations||MSc in Finance and Accounting, (Graduate Programme) Final Thesis|
|Number of pages||99|