«Equity Style Investing RONG, WU How to cite: Equity Style Investing, Durham theses, Durham University. RONG, WU (2013) Available at Durham E-Theses ...»
A business cycle model usually uses economic variables to determine an economic state, such variables are latent in essence and hence the forecasted outcome as which state would prevail at each point of time can only be drawn on an adaptive manner. Quantitative-based adaptive trading techniques have already raised many interests from academics in the literature (c.f. Rabatin (1997), Hung et al. (2003), Chiu and Xu (2004)). In the context of the equity style investing, the adaptive style rotation model forecasts the equity style performance dynamics and identify the leading style trends, rather than on the individual stocks, in the current market state and opportunistically shifting to the most productive style. By striving to invest a number of top-performing stock groups in leading market segments in a specific period of time, the objective of adaptive style allocation is to achieve enhanced style investing returns via a more rewarding form of diversification.
There is a growing literature exploring the dynamic trading strategies based on the equity style cycle and the corresponding style switching in a given point of time. Birch (1995) shows plan sponsors can use style cycle information to manage equity style exposures. Reinganum (1999) demonstrates the massive economic benefits of controlling the variability of size premiums to improve returns as compared to the by-and-hold and rebalanced fixed-weighted investing strategies. Kao and Shumaker (1999) simulate the performance of three timing strategies based on asset classification (e.g. stocks versus cash, size and value-growth stocks) in the U.S. markets. They use a set of macroeconomic variables like yield curve, real bond yield, corporate credit spread, high yield spread, estimated GDP growth rate and earning yield gap for their business forecasting model to forecast subsequent year’s value-growth performance. Kao and Shumaker (1999) find that the rotation strategies based on stocks versus cash, small-cap stocks and large stocks could historically provide more opportunities to outperform the timing strategies based on value and growth stocks. Kao and Shumaker (1999) demonstrate that, based on the monthly rebalancing, a perfect selection by market values could add 20%-27% spread to the market returns, while perfect foresight timing between value and growth stocks could achieve 24%-34% higher return than market average. Other relevant studies including Fan (1995), Sorensen and Lazzara (1995), Avramov (2002), Bauer and Molenaar (2002) and Amenc et al. (2003), they also have documented evidence of predictability in style returns and the corresponding style rotation strategies.
In the U.K. market, Levis and Liodakis (1999) examine the style rotation strategies based on size and value-growth dimensions for the period of 1968-1997. They demonstrate that a hypothesised investor who could perfectly identify the size premium turning points would generate average annual return of 34%. An accuracy of 60%for the investor’s forecasting ability would be sufficient to beat the small size long only investing or buy-and-hold passive investing.
Similarly, with a perfect foresight to identify value and growth style turning points, the value-growth rotation strategy would have earned annual returns of 29%. More recently, Clare et al. (2010) investigate the UK momentum-based multi-style rotation strategy. They argue that simple momentum style rotation strategy could outperform the complicated quantitative multi-style rotation strategy based on set of forecasting variables. Overall, these studies and many others generally conclude that since expected returns on leading market segments present predictable time-varying components over the business cycles, rotation strategies across equity styles could offer a substantial opportunity to outperform the market averages.
2.9 Optimal style allocation incorporating return predictability
Empirical finance documents the evidence of time-varying expected returns with predictable components across styles. The important implication of such return predictability is that active investors may wish to engage style rotation strategies to enhance returns. To model expected returns, traditional finance generally links expected returns with the condition risk premium by previous observable information set. One of the popular approaches to model the time-varying expected return patterns is to allow the information set to contain some economic pervasive variables that have been identified as return predictors by previous research6. Campbell and Viceira (2005) argue that the stock return predictability can have a strong impact on the variance and covariance structures of asset returns which is relevant for buy-and-hold investors with fixed investment horizons.
Brant (2010) observes that following the recent empirical evidence of such predictable time-varying return distributions, optimal portfolio selection problems has once again been in the forefront of financial research. For example, Kandel and Stambaugh (1996) show that from an ex ante perspective variables predicting the distributions of the moments of stock return exert significant impact on a tactical portfolio allocation. Brennan and Schwartz (1996), Brennan et al.
(1997) and Barberis (2000) examine the impact of predictability to the myopic versus dynamic portfolio choice problems. Ferson and Siege (2001) derive the optimal portfolio weights for mean-variance
Solnik (1993) argues there are three approaches to model expected returns: the
first is to contain past returns in the information set. The second is to contain the first and second moments in the information set, and the third is to use economic variables like yld, def, term and div as discussed in previous sections. Studies such as Harvey (1991) show the strong explanatory power of such variables to both U.S. and none U.S. equity risk premia.
investors assuming that the moments of stock returns are known functions of state variables. More recently, Avramov and Chordia (2006a, 2006b) find that a real time optimising investor benefits from incorporating business cycle information to the asset allocation between stocks and cash, and investment strategies such as ‘fund of mutual funds’ can also benefit from capitalising on the predictable time-varying dynamics over the business cycles.
Asset allocation is the key factor in determining the performance of long-term investments. Brinson et al. (1986) show that the decision of how to allocate assets accounts for about 90% of the performance variations for large pension funds. Likewise, the prominent study of Sharpe (1992) suggests that 90% of the performance of equity funds is due to the overall style of the fund, while the remaining 10% is due to the individual characteristics of the specific securities hold.
From a money manager’s perspective, for a solid strategy to decide an appropriate asset allocation, it requires first to consider on which level, tactical or strategic.
There is fundamental difference between tactical asset allocation and strategic asset allocation framework. Strategic asset allocation is mainly driven by the long-term return-risk assumptions for various asset classes. It specifies the overall weight of various styles in a portfolio to satisfy investor’s risk-return preference in a lengthy investment period. However, the change of investor’s life style will eventually impact the underlying risk tolerance and in turn his strategic asset allocation decision. Hence the risk-return profile for strategic asset allocation should be evaluated periodically once the investment landscape experience fundamental change. Unlike strategic framework, tactical asset allocation takes into account the short-term market conditions and is therefore designed to identify the possibility to tilt strategic asset allocations according to the changes in the investment opportunity set. Hence the underlying drivers for tactical asset allocation are valuation, momentum or contrarian, investor’s sentiment and business cycle effect etc. Overall, the strategic asset allocation is the establishment of a long-term investment objective, while the tactical asset allocation determines how to adjust strategic asset allocation by exploiting inefficiencies in equilibrium values among asset classes. A solid investment strategy must highlight the role of both frameworks from the very beginning.
The optimal strategic and tactical asset allocations are perhaps most relevant for delegated asset management. As mentioned previously, institutional investors like pension funds and endowment funds act as fiduciaries and generally accept substantial responsibilities and assume significant liabilities. van Binsbergen et al. (2008) argue that the asset allocation of such investors are mainly structured around asset classes. As a result the fund’s Chief Investment Officer (CIO), who acts in the best interest of his beneficiaries, would pick asset manager who is specialised in a single style or delegates the portfolio decision to such specialists. Therefore the asset allocation decisions are made in two stages, namely CIO’s strategic allocation to different styles represented by different style managers and the individual style manager’s tactical allocation within his style7. The CIO usually has long-term investment horizon and his objective is to minimise the utility cost from the misalignments of incentives induced by the above two-step allocations by optimising the investment weights to
The reason why the CIO in the asset management firm should hire such multistrong>
style managers can be justified by Sharp (1981) who argues that the decision to employ different managers is to exploit their specialisation or to diversify among managers (i.e. style diversifications).
different style managers in a mean-variance framework. In contrast, the individual style manager, however, is motived to maximise his remuneration on a relatively short horizons. van Binsbergen et al.
(2008) argue that if asset returns are predictable, the CIO’s optimal style manager choice problem depends on his investment horizon and requires being tactically optimised. This introduces the hedging demands from the difference between the strategic and tactical style portfolio weights in response to changes in the future investment opportunity set.
A variety of theoretical solutions have been explored in the literature to solve the optimal portfolio choice problem incorporating return predictability. Brandt and Santa-Clara (2006) point out that most techniques are out of reach for ordinary investors since close-form solutions are not always available. Over the years the mean-variance paradigm of Markowitz (1952) is the major workhorse of portfolio optimisation. When solving the optimal portfolio choice problem, prior studies generally first estimate the conditional moments with state variables and then apply traditional Markowitz approach. This methodology raise concerns such as rigid assumptions between moments of returns and state variables to safeguard covariance matrix and massive number of parameters be estimated. Michaud (1989) argues this will inevitably results in notoriously noisy and unstable test results. Recently, Brandt (1999) develops a framework to bypass the procedure of estimating the joint distributions of conditional stock return but directly estimate the optimal portfolio weights based on the state variables. Ait-Sahalia and Brandt (2001) argue that the predictability of expected returns and the covariance structure is difficult to be translated into portfolio selection advice because the two moments may be predicted by different variables.
Moreover, a variable may be both significant for predicting the variations of expected return and variance but such variations offset therefore it is not useful for determining optimal portfolio weights.
Based on that, Brandt and Santa-Clara (2006) propose an approximation to solve the CIO’s problem by introducing managed and timing portfolios in the asset space. This approach is easy to apply by investors in the traditional static Markowitz paradigm.
Chapter 3 Equity Style Drivers: Business Cycle Risk versus Firm-specific Characteristics
3.1 Introduction Over the past decades a large number of empirical studies provide evidence to show that certain firm characteristics can profitably differentiate among stocks. For example, Banz (1981) first reports the size premium that stocks with small market capitalisation can earn higher risk-adjusted returns than those with large market values.
Defined as having higher earnings-to-price ratios (E/P), Basu (1983) first documents that value stocks could generate higher absolute and risk-adjusted returns than growth stocks. The outperformance of value stocks (often called the value premium) is also found when value stocks are defined by different firm characteristics such as book-tomarket ratios (BM), price to cash-flow ratios (PC) or dividend-yield (DY) (c.f. Fama and French (1993, 1998); Lakonishok et al. (1994)). These results are robust across U.S. and international markets. Parallel to the findings of divergent return patterns across different equity groups, the concept of style-based investment strategy has evolved in the U.S.
markets. For instance, around 1980s, institutional investors such as pension funds start to engage in style investing by searching the best style managers to build portfolios that can capitalise on the relative style performance within the investment cycles. The premise of style investing is that investors allocate their asset along style level rather on the individual stock level. Since asset categorisation based on firm characteristics provides common structure in the complex investment environment, the idea of style investing has gained growing popularity in today’s financial markets because it simplifies money managers’ decision-making process and makes the investment process less intimidating (c.f. Mullainathan (2002), Barberis and Shleifer (2003)).