Factor based pension portfolio strategies for sustainable withdrawals

Abstract

This thesis studies the use of factor-based investment strategy to create pension fund portfolios for sustainable withdrawals. The first empirical chapter focuses on identifying factor-based anomalies in the UK stock market. This chapter identifies size, book to market ratio and profitability as factors that explain portfolios/stock excess returns other than the market factor, the sole predictor as stipulated by the capital asset pricing model (CAPM). In addition to these factors, this chapter explores the volatility factor which is currently gaining attention within the academic and professional sectors. Using the FTSE350 universe, the chapter then shows with the CAPM model the presence of size effect (small stocks outperforming large stocks in risk adjusted terms with smaller CAPM beta), profitability effect (high profitable stocks outperforming low profitable stocks in risk adjusted terms with smaller CAPM beta) and the volatility effect (low volatility stocks outperforming high volatility stocks in risk adjusted terms with smaller CAPM beta). The value effect (value stocks outperforming growth stocks in risk adjusted terms) was not identified. This chapter also finds evidence that the size, book to market and volatility factors could produce the observed excess return in an investment portfolio over the market portfolio in the UK stock market, but the profitability factor could not. However, in risk adjusted terms, only low volatility portfolios produced significant returns in excess of the market. Finally, this first empirical chapter documents that the returns of all portfolios constructed from the mentioned factors had a tendency of mean reversion. The chapter also examines the potential impact of transaction cost by using a methodology and statistic that give a reasonable indication of the scale of the likely costs of replicating an index which can then be used in observing if the potential cost of transaction will make the replication worthwhile; the chapter finds that transaction cost would not have changed the order of outcome observed. The second empirical chapter further investigates the effect of combining the factors that were studied in the previous chapter. I constructed 64 portfolios (and considered 53 of them) based on different combinations of these factors and observed that 23 of these factors-based portfolios outperformed the market portfolio. However, only 7 portfolios produced significant risk adjusted returns in excess of the market, 5 of which also produced observed excess return. Interestingly, all the 7 outperforming portfolios were constructed with the low volatility factor. This chapter also proposes a new ad hoc measure to capture the relative stability of an investment portfolio. The objective of this measure is to create a theoretical framework to assess the stability of a portfolio. This measure considers both the speed of mean reversion of the portfolio return and the ‘distance’ of deviation of the return from its mean. The chapter finds that 10 factor-constructed portfolios had better return stability than the market portfolio. Interestingly, all of these portfolios had a common characteristic—they were all constructed with the low volatility factor in the factor combinations. The third empirical chapter applies the results from the second study to pension fund investment. Using Monte-Carlo simulation, I explore how well the factor-based portfolios can provide sustainable pension-income withdrawals in addition to other performance indicators during the drawdown phase of an individual’s retirement. I identify four portfolios that could sustain a withdrawal rate of up to 10% with strong levels of success. Again, very interestingly, all of these portfolios were constructed with the low volatility factor; this seems to suggest that the low volatility factor is indeed a driver of return sustainability. In addition, I revisit the proposed relative stability measure introduced earlier and find that it has varying consistencies with the performance indicators of investment portfolios during the withdrawal phase in a pension fund. The chapter also finds that all else being equal, when the returns of a portfolio are highly persistent (slower reversion speed), a better success rate will likely be achieved; this is also the case when the deviation of shock is smaller, all else being equal. Overall, the findings of this thesis contribute to the literature in identifying factors that are particularly relevant to the withdrawal sustainability in pension fund investment. It provides evidence that constructing pension fund portfolios using low volatility and value factors tilts can generally provide a more stable and secure withdrawal experience to pension fund clients.