«Equity Style Investing RONG, WU How to cite: Equity Style Investing, Durham theses, Durham University. RONG, WU (2013) Available at Durham E-Theses ...»
Recent research of style investing has shifted from providing empirical evidence on the existence of relative style returns to the investigations of various components and theory-based interpretations of relative style performances. While the benefit of style investing is less controversial, it remains an ongoing debate why some stock groups can generate higher average returns than others in a given period of time. Rational asset pricing theory argues that stock markets are efficient and the outperformance of one style over another is not abnormal but rather represents compensation for higher nondiversifiable systematic risks. Chan et al. (1985) and Huberman et al.
(1987) show that the relative returns of small and large stocks are due to their different sensitivities to the risk factors important for pricing assets. Fama and French (1993) document that value premium is related to a distinct distress factor proxied by firm leverage or the book-to-market ratio. As a result the outperformance of value stocks would suggest they are fundamentally riskier than growth stocks. In contrast to the traditional rational-based explanations, behavioural finance links the divergence of style returns to the mispricing of some asset groups caused by investors’ cognitive biases unrelated to economic fundamentals. Lakonishok et al. (1994), for example, argue that value stocks and growth stocks are not properly priced in stock markets. The outperformance of value stocks is driven by investors’ systematic judgement errors to believe that the past growth rate for growth stocks would persist far in the future. Value and growth returns reverse when investors subsequently receive surprises regarding the financial results for the two styles. Hence the reason for value premium is driven by investor’s cognitive biases rather than due to the compensation for higher systematic risks. Apart from the rational and behavioural frameworks for the size and value premiums, papers such as Daniel and Titman (1997) propose other school of characteristic-based interpretation. They contend that the crosssectional variations in expected returns between stocks with different characteristics are not due to there being risk factors associated with Fama and French (1993) three factors, but rather from characteristics themselves. Hence the size and value premiums are caused by their underlying firm-specific characteristics instead of different loadings on the risk factors underpinning the asset pricing dynamics.
A growing number of empirical studies demonstrate that the observed variations on returns across some equity styles are related to the dispersions of cross-sectional expected returns. Conrad and Kaul (1998) and Berk et al. (1999) argue that stocks with high (low) expected returns tend to achieve high (low) realised returns. These studies have highlighted the importance of the macroeconomy in determining such cross-sectional variations in expected stock returns.
There are strong a priori grounds to relate stock returns to the business cycle conditions. Finance theory provides a suggestive correlation between stock price and economic states. For example, the dividend discount valuation model suggests that the present value of a stock equals to the aggregate discounted expected future dividends received. There are 4 parameters involved when evaluating the value of a stock, namely, the expected future cash flows, the market risk premium, the market risk exposure and the term structure of interest rates. Dahlquist and Harvey (2001) point out that these variables share a common component, the business cycles. Indeed, a firm’s ability to generate cash flows and its risk exposure often differs in different phases of the economic cycles. The market risk premium is low when the economy peaks and high when it troughs. The term structure of interest rates (the yield curve) is the leading indicator of business cycle volatility that determines a firm’s cost of capital. Bolten and Weigand (1998) demonstrate how the underlying parameters in a basic dividend discount valuation model vary and are affected by different states of the economy.
Chan and Chen (1991) and Fama and French (1993) propose that the returns of distressed stocks are especially sensitive to economic states and are driven by many of the same macroeconomic factors such as variations over time in bankruptcy costs and the accessibility to credit markets. Bernanke and Gertler (1989), Gertler and Gilchrist (1994), Kiyotaki and Moore (1997), and Hahn and Lee (2006) show that changing credit market conditions can exert different effects on risks and expected returns across styles. Berk et al. (1999) provide a theoretical model in which the value of a firm is the sum of its existing assets that generate cash flows and the value of an option that makes positive net present value investment in the future. Their model suggests that the expected return of a firm is jointly determined by the current interest rate, the firm’s systematic risks of its existing assets and the number of active projects. Thus expected returns vary across firms with changes in interest rate and the number of old projects that are dead and replaced. Consistent with these studies, authors such as Perez-Quiros and Timmermann (2000) document asymmetries in the variation of small and large firms' risk characteristics over the economic regimes. Vassalou and Xing (2004) find that the size and value premiums are intimately related to the default risk, which is related to macroeconomic factors and varies with the business cycles (c.f. Denis and Denis (1995)). More recently, Zhang (2005) suggests that value and growth firms have different ability in investing (disinvesting) in good (bad) times and therefore the dispersion of risk between value and growth stocks is high in bad times, while the risk differential is low or even negative in good times. Black and McMillan (2005) also show that value and growth stocks exhibit asymmetric responses to the shocks in macroeconomy across the business cycles.
Value stocks tend to be more responsive to changes in macroeconomic conditions than growth stocks, and such responsiveness increases during economic contractions.
In a recent paper, Chordia and Shivakumar (2002) investigate the influence of time-variations in risk premia on the momentum effect of Jegadeesh and Titman (1993). The momentum effect suggests that if stocks are classified by their past performance, the winner group continues to earn higher returns than the loser group in medium term.
Using a parsimonious set of macroeconomic variables in a multifactor business cycle model framework, Chordia and Shivakumar (2002) find that momentum profits can be attributed to the higher conditional expected returns predicted by business cycle model. Thus the relative return differentials of the two asset classes can be interpreted as the compensation for bearing the business cycle risks rather than the diversifiable firm-specific risks. Griffin et al. (2003) also study whether global momentum profits could be attributed to macroeconomic risks.
They employ the model of Chen et al. (1986) to regress the momentum returns on contemporaneous macroeconomic variables but fail to find a direct relation between macroeconomic risks and momentum profits.
More recently, Avramov and Chordia (2006a) develop a framework extending that of Brennan et al. (1998) to test whether asset pricing models can explain size, value and momentum effects. In their paper, the factor loadings of a given asset pricing model change with characteristics such as firm size and BM ratios as well as with business cycle conditions. Avramov and Chordia (2006a) show that when beta is allowed to vary with size, BM and macroeconomic variables, the size and value premiums are often explained and the momentum effect can be captured by model mispricing that varies with macroeconomic variables, suggesting the risk-based explanation for size and value premiums and a potential business cycle related explanation for the impact of momentum on the cross-section of stock returns. Overall, the majority of recent studies generally suggest that economic exogenous forces dominate in affecting equity style return dynamics over time, and the reason why some stocks offer average higher returns than others is because they bear higher time-varying macroeconomic risks. Hence stock price evolves over time, reflecting the cyclical and structural changes in the aggregate economy.
While rational, behavioural and characteristic-based theories are able to explain the divergent equity style return patterns, the relative importance of such theories has not been carefully studied in the extant literature. From an investor’s perspective, the observed timevarying relative equity style returns is of obvious importance as it introduces opportunities for active portfolio manager to tactically invest in some specific asset classes in certain periods of investment cycles. However, to successfully implement such equity style rotation strategy, one must be able to not only identify the underlying driving forces that determine the relative style returns, but also to capture the mechanisms through which those forces work. Understanding the relative importance of such competing theories is important since different interpretations would suggest different driving forces that underlie style return dynamics and consequently provide different practical guidance for active portfolio management.
This chapter contributes to the literature by empirically investigating the relative importance of common risk factors versus the firm-specific information as driving sources of equity style return differentials. The objective of this chapter is to answer a central research question: what is the dominant factor that affects size and value premiums, common risk factors or the firm-specific information? Answers to this question tells rational and behavioural theories apart because a common structure to the divergent style return could point towards a rational risk-based interpretation, while the firm-specific based finding is more likely to be within the behavioural framework.
To pursue this research question, Chapter 3 builds some simple style trading strategies and examines the underlying sources determining the profitability of such style investing in the U.K. stock market. The study of the U.K. market is motivated by the fact that despite being one of the most influential financial markets, the U.K. experience of style investing has lagged considerably behind that of the U.S.
(Williams (2004)) and therefore needs careful research. Although style investing develops from and still dominates in the U.S. stock markets, given the fact that such investing is based on sound and observable characteristics that are theoretically as relevant as they are in the U.S.
context, the fundamental rule of style investing is arguably the same in the U.K with different economic and institutional environment.
This chapter develops and employs the methodology used in Chordia and Shivakumar (2002) to investigate the relative importance of common risk factors versus firm-specific information as sources of size and value premiums in the U.K. stock market. Over a sample period of January 1980 to December 2004, all U.K. stocks are categorised into size and value-growth groups according to firm characteristics such as market value (MV), market-to-book ratios (MTBV), price to cash flow ratios (PC) and dividend yields (DY). Based on asset classification, simple long-short style investing strategies are tested and their return dynamics over the business cycles are examined. Using firm characteristics to categorise stocks is pervasive in the financial market. Empirical research consistently finds robust cross-sectional relation between average stock returns and equity characteristics. More importantly, it is found that stocks with similar characteristics tend to move together. Huberman et al. (1987) find that returns of stocks within the same size range tend to comove and respond to risk factors in similar ways. Berk et al. (1999) argue that firms with same characteristics are affected by the same state variables relating to systematic risks and expected returns. Hence firms share similar characteristic tend to have the same underlying pervasive forces affecting stock returns. These studies point to the rationale of simple asset allocation strategies focusing on specific asset classes that share similar characteristics.
In response to the popularity in recent studies to link macroeconomic effects with the observed cross-sectional variation on stock returns, Chapter 3 also models expected stock returns conditional on shocks originating in a set of pervasive economic variables that relate to the business cycles. To examine whether business cycle risks contribute to the realised return differentials, style investing strategies are tested based on both the predicted and unpredicted part of the business cycle
model. Specifically, 2 hypothecations are tested:
1. If business cycle risk is the major driving force to the crosssectional variations on stock returns, style spreads should be substantially decreased after controlling for the exposures to the predicted macroeconomic risk premias;
2. If mispricing (firm-specific information) is the major source that underlies the relative style returns, controlling for the business cycle effect would not cause material changes for the observed style spreads. Rather, simple style investing strategies based on business risk adjusted returns would generate significant profits.
Since equity characteristics under consideration explain significant cross-sectional variation in average stock returns, rational pricing theory would argue that such firm characteristics are proxy for risk factors or the information of mispricing, or alternatively they are cross-sectionally correlated with the underlying factor loadings. In order to better understand the mechanism that explains the crosssectional variation in mispricing of the business cycle model, the contemporaneous relations between equity characteristics, common risk factors and the mispricing from the business cycle model are also cross-sectionally examined using model pricing errors as dependent variable on equity characteristics augmented with estimated loadings on asset pricing models such as CAPM and Fama and Fench (1993) three-factor model.