The value premium within and across GICS industry sectors in a pre-financial collapse sample

Abstract

A portfolio manager employing a top-down/bottom-up method who seeks to capture the value premium long promised in academic literature would want to first determine whether the premium exists across industries and not just observed in firm-specific book-to-market (BE/ME) relationships. Next, the investor would want to know if BE/ME characteristics are stable across these defined homogeneous groups or whether there is considerable variation. Results show that certain industries appear to have a natural or structural tendency to reflect either a high or low BE/ME characteristic. Results also shows that growth-oriented industry BE/ME characteristics appear to be more stable than value-oriented industries over time. Moreover, stocks from growth-oriented industries tend to cluster at high rates in the lowest BE/ME quintile, while stocks from value-oriented industries appear more evenly distributed across middle BE/ME quintiles over time. Value stocks found in growth sectors outperform value stocks in value sectors, contrary to prior published results. The January premium exists both within and across Global Industry Classification Standard industry sectors, but the value premium is not subsumed by the January effect in either analysis.

Keywords:

• value premium
• portfolio management
• value stocks
• GICS
• industry groups

Public Interest Statement

A 25-year lineage of research exists that says stocks with high book-to-market accounting characteristics are “riskier” than those with low characteristics. Thus, investors should be able to capture this risk and improve their investment portfolio performance by purchasing stocks within industry groupings that exhibit such characteristics. This article suggests that the task is more complex, predictable in some instances and unpredictable in others.

Notes

1. Chan, Lakonishok, and Swaminathan (2007) find that both GICS and FF industry coding systems yield “sets of economically related stocks.” They compared GICS and FF systems with a mechanical industry clustering method, and find GICS and FF performs well in capturing out of sample return covariance as well as co-movement in fundamental characteristics such as sales growth.

2. The Standard & Poor’s Company uses GICS for its highly popular SPDR® exchange traded funds. S&P converted its ETF funds to the GICS system in June 2002. The giant Vanguard investment firm also uses GICS to classify stocks to their various sector ETFs. Internationally, several stock exchanges such as the Toronto, ASX in Australia, and Nordic exchanges use GICS for stock listing classifications. According to the sales literature produced by S&P, 8 of the top 10 sell-side investment firms and 9 of the 10 buy-side investment firms utilize the GICS system. The fact that Standard &Poor’s and Morgan Stanley own and manage the dominant S&P and MSCI global index products ensures that GICS will be heavily used by the practitioner community to construct any index-related industry sub-classifications. S&P announced that the conversion of their popular S&P/Citicorp equity growth and value indexes to GICS was completed in July 2005. Yet another popular classification system in wide use by practitioners is the Industry Classification Benchmark (ICB) by Dow Jones Indexes and FTSE. The popular Dow Jones I-Shares utilize the ICB classification system.

3. See Chiang (2002) for a comprehensive literature survey and analysis of the effect of statistical return weighting methods on the value effect.

4. Mutual fund portfolio holding data source: Morningstar.com.

5. When using the value-weighted return computation method employed by Fama and French, none of the intercepts are statistically significant. The average across-industry alpha of 0.30 for value-weighted returns is almost identical to the average absolute regression intercept of 0.28 across the 48 value-weighted industry portfolios in Fama and French (1997).

6. Prior criticisms of the three-factor model for persistent negative correlation between alpha and the HML factor loading is also not remedied by using a different industry coding system. The correlation (not shown) between industry intercepts and HML slopes in three-factor model regressions in Table 3 remains negative and statistically significant (ρ = −0.53, t = −2.87).

7. Chan et al. (2007) find that two-digit GICS codes provide lower differences between return correlations for stocks in a particular sector and correlations for all other stocks outside that sector when compared to the four, six, and eight digit codes. While not optimal, two-digit GICS sorted sectors still reflect a considerable range of BE/ME characteristics.

8. For robustness, a check was also performed using the Fama and French average 50th percentile ME breakpoint on size, and results (not shown) are not materially different.

9. The Hi-LO value premium was statistically significant at the 5% level for all sectors below the 50/50 ME breakpoint and once again not statistically different from zero for all sectors above the size breakpoint.

10. Average annual sector ROA sorted into five BE/ME quintiles are observed for the current sample period (not shown). Hi-Lo quintile ROA statistics are distinctly negative for growth-oriented sectors and positive for value-oriented sectors (ROA/BEME ρ = 0.72, t = 2.57). Therefore, results are not inconsistent with arguments by Banko and Conover (2006) that the value premium results from investor risk-pricing of distress.

11. The January premium may simply be the result of data snooping as generally suggested by Lo and MacKinlay (1990) and Fama (1998). Fama argues that most market anomalies disappear after certain tweaks in statistical methods.

12. Results shown in Table 5 represent stocks above and below the Fama and French 25th percentile average size breakpoint for the sample period. The January premium completely disappears in large stocks in sorts using an average 50th percentile (below 50th/above 50th) size breakpoint.

Additional information

Funding

Funding. The authors received no direct funding for this research.

Notes on contributors

Kenneth E. Scislaw

Kenneth E. Scislaw’s resume reflects a 35-year involvement with retail, institutional, buy-side, sell-side, investment, and academic research segments of the finance profession. He is currently an assistant professor of Finance at Drury University. He has taught finance at several universities around the world including Drexel University (USA), University College Dublin (IRE), and the University of St Andrews (UK). He worked for almost 20 years for major global investment firms including Merrill Lynch in New York and for the investment billionaire, Sir John Templeton.