In the investment management industry, there are two general schools of thought as to what style of equity selection offers the highest return over time. The two schools are commonly referred to as "value" and "growth," and the debate over which side offers a higher return on investment is rooted in the seminal Security Analysis of Graham and Dodd. Their writings were the first to distinguish value-oriented approaches to equity investing from "growth" or "glamour" strategies.
Moreover, Graham and Dodd were the first to argue for a return premium attributable to value investing, which was eventually coined as the value premium. Fama and French affirmed the value premium, as did Lakonishok, Shleifer and Vishny, and Lakonishok and Chan. However, Black and Mackinlay suggested the value premium was sample specific, and possibly the consequence of "data snooping." This question, whether or not a value premium is evidenced by equity-return data, holds considerable importance in the field of money management. Investment professionals largely form their equity purchasing decisions, such as buying value or growth stocks, around the theories and models created by financial economists (Lakonishok, 2004).
While Graham and Dodd pioneered the theory of a value premium, its empirical foundation is generally traced back to the studies of Fama and French (1992, 1995). In their analyses, Fama and French classify companies with high multiples of book-to-market equity (B/M) or earnings-to-price (E/P) as value stocks. Contrarily, those firms with relatively low B/M or E/P multiples are categorized as growth stocks. B is defined as the book value of common equity plus balance sheet deferred taxes; E is defined as earnings (income before extraordinary items, plus income statement deferred taxes, minus preferred dividends). Fama and French sort stocks on the New York Stock Exchange (NYSE) into ten portfolios based on these ratios, and in turn divide the top and bottom decile portfolios in half.
The two portfolios comprised of companies with the highest B/M and E/P ratios are considered "value portfolios," and at the opposite extreme are "growth portfolios." At the end of each year, the portfolios are reformed on the basis of B/M and E/P rankings. Fama and French measure the average monthly returns of these portfolios from 1968 to 1990, and find strong positive correlation between book-to-market equity and earnings-to-price and average return. Specifically, value stocks outperform growth stocks by an average of 1.53 percent per month over the period. In essence, the value premium is confirmed over their sample period.
Lakonishok, Shleifer and Vishny (1994) conduct a similar study of value versus growth portfolios over the 1968 to 1990 period, but differentiate their analysis in four significant ways. First, they construct the portfolios based only on the book-to-market (B/M) ratio of each company, rather than a combination of B/M and E/P. Second, they report results only for the largest fifty percent of stocks on the NYSE, as measured by market capitalization, to mitigate the effects of survivorship bias.
Third, they measure portfolio returns on an average annual basis, as opposed to monthly. Lastly, the portfolios are rebalanced every five years instead of annually, in an attempt to better mimic a long-term buy-and-hold strategy (long-term is commonly understood to mean five years in institutional investment management). Their findings show value stocks outperform growth stocks by an average 10.5 percent annually over the 1968 to 1990 sample period.
Lakonishok and Chan (2004) use the methodology of Lakonishok, Shleifer and Vishny (1994) for the 1979 to 2001 period, and reaffirm the existence of a value premium. Over the period, value outperforms growth by an average of 5.7 percent annually. Lakonishok and Chan note that, due in large part to the findings of Fama and French and Lakonishok, Shleifer and Vishny, money managers now widely recognize value and growth as distinctive specializations. They also remark that the academic community generally agrees that value strategies outperform growth investment strategies. However, in their concluding statements, Lakonishok and Chan point out the lack of research conducted outside of the 1968 to 1990 period. In fact, they articulate a concern that the value premium may be the result of collective data snooping and sample bias.
Black (1993) is the first to argue that the value premium is sample specific. Although he does not conduct a formal study of equity returns, he suspects the outperformance of value strategies will vanish over time, and attributes its historical precedence to data-mining. His views are supported by MacKinlay (1995), who insists that the findings of a value premium have been the consequence of analyzing within-sample fits through data-snooping.
Like Black, MacKinlay does not formalize a study of equity returns, and rather focuses on the deviations of value and growth returns from the Capital Asset Pricing Model (CAPM). Regardless, Black and MacKinlay, and to an extent Lakonishok and Chan, challenge the merit of a value premium and suggest it may be sample specific.
The most recent literature is that of Fama and French in their working paper for the Chicago Booth School of Business. In this paper, Fama and French essentially update their research for the 1991 through 2010 period. They find the value premium to be pervasive domestically and internationally.
I have begun an empirical analysis to test the argument of Black and MacKinlay that the value premium is sample biased. I have aggregated equity-return data from Capital IQ to determine whether a value premium exists in publicly traded companies in the United States between 1991 and 2012. Capital IQ provides an extensive database of publicly traded securities listed on the S&P 500, which is a broad-market index of the largest 500 publicly traded companies in the United States. It offers daily closing prices for each company as far back as 1985, and also lists biannual accounting metrics for each firm, namely inverted earnings-to-price and book-to-market-equity ratios.
I have employed Capital IQ accounting data to sort the top 250 component securities (by market capitalization) of the S&P 500 by book-to-market equity, and each decile of securities constitutes a portfolio. The portfolio with the highest B/M (lowest price-to-book) ratios is the value portfolio, and at the opposite extreme is the growth portfolio.
Equity-return data on Capital IQ begins in 1985, but the value and growth portfolios are measured from 1991 through 2012. This is done because return data is available for only a few companies before 1991. Beginning in 1991, portfolio performance is measured by matching component accounting metrics, namely book-to-market (B/M) ratios, at fiscal year-end in calendar year t - 1 (1991 - 2011) with return data for year t. Returns for each company are measured as the difference of calendar-year-end closing price and calendar-year-end closing price at time t - 1.
To determine whether value investments outperform growth over the 1991 to 2012 period, portfolio returns are measured annually on a five-year rebalancing basis as an equal-weighted index (Lakonishok, Shleifer and Vishny, 1994). For each portfolio of N securities, the individual return of security i accounts for i/N of the total portfolio return. The portfolios are reformed after five years on the basis of B/M rankings for each company, which better reflects the purchasing behavior of long-term investors than the annual rebalancing methodology of Fama and French.
Over the 1991 to 2012 period, the growth portfolio outperformed its value counterpart by an average of 470 basis points per year. These results are surprising because they contradict the widely regarded economic literature which argues for the pervasiveness of a value premium in equity selection. The findings are also significant because they provide deeper transparency into the importance of methodology in index-construction. My results showed growth outperformed value over the same period where Fama and French found the opposite, and the reason for this disparity is methodology (they rebalanced annually; I rebalanced every five years).
In the existing literature, only nominal performance is measured to compare value and growth portfolios. I measure the value and growth portfolios' average Sharpe ratios to provide transparency into the riskiness associated with the returns of each. The Sharpe ratio - pioneered by Nobel Laureate William Sharpe - measures returns on a risk adjusted basis by subtracting the risk-free rate from the underlying security's returns, then dividing the difference by the standard deviation of these returns. The result shows the marginal return of the underlying per unit of volatility. Quite simply, higher Sharpe ratios are more desirable.
The data show that between 1991 and 2012, the value portfolio had an average Sharpe ratio of .22, while growth averaged .14. In other words, a higher price was paid for higher return.
My analysis suggests that, over time, equity selection with a growth-style emphasis may lead to higher returns, though earlier sample periods convey the opposite. As such, below is a table of stocks in the rebalanced growth portfolio started in 2012, as constructed by price-to-book rankings.
The practical costs associated with forming the above index are material, and one might consider conducting a more fundamental, bottom-up analysis of the individual securities. The following are the companies in list-form (descending P/B):
DIRECTV (DTV), Philip Morris International (PM), Western Union Company (WU), Colgate-Palmolive (CL), Yum! Brandes (YUM), The Hershey Company (HSY), Salesforce.com (CRM), Boeing Company (BA), priceline.com (PCLN), Ford (F), Amazon.com (AMZN), Limited Brands (LTD), Altria Group (MO), Accenture (ACN), Coach (COH), Express Scripts (ESRX), International Business Machines (IBM), United Parcel Service (UPS), Kellogg Company (K), Estee Lauder (EL), FMC Technologies (FTI), Lockheed Martin Corporation (LMT), Fastenal (FAST), Mastercard (MA), Simon Property Group (SPG), and Starbucks (SBUX).
For reference, below is the list of companies in the value portfolio started in 2012 (descending P/B):
Chubb (NYSE:CB), Edison International (NYSE:EIX), Time Warner (NYSE:TWX), State Street (NYSE:STT), Duke Energy (NYSE:DUK), WellPoint (WLP), BB&T (NYSE:BBT), Goldman Sachs Group (NYSE:GS), PNC Financial Services Group (NYSE:PNC), Fifth Third Bancorp (NASDAQ:FITB), The Travelers Companies (NYSE:TRV), Bank of New York Mellon (NYSE:BK), Ace Limited (NYSE:ACE), CME Group (NASDAQ:CME), Valero Energy (NYSE:VLO), JP Morgan Chase (NYSE:JPM), Loews (NYSE:L), Prudential Financial (NYSE:PRU), Allstate (NYSE:ALL), MetLife (NYSE:MET), Capital One Financial (NYSE:COF), Morgan Stanley (NYSE:MS), Citigroup (NYSE:C), American International Group (NYSE:AIG), and Bank of America (NYSE:BAC).
It is worth emphasizing that the price-to-book ratio measures the price paid for the underlying stock in the market for one dollar of that company's equity value on its books. For example, investors pay 53 dollars for each dollar of equity on the books of DIRECTV. The reason why such high multiples are sustained in the market is because investors price in rapid growth estimates for the companies. In other words, investors pay a premium for anticipated growth.
I personally subscribe to a value-oriented investment philosophy and purchase companies whose market prices seem substantially below their intrinsic values. However, the data unbiasedly speak to the underperformance of such a strategy over the last 20 years.
In my next article, I will offer the results of this study with value and growth portfolios sorted by price-to-earnings ratios, rather than price-to-book. I will then couple these analyses with dividend-yield filters, as a key determinant of whether a company is "more value" or "more growth" is its use of cash; if a company is priced to have rapid growth opportunities, then one would expect management to reinvest cash in the business, rather than pay dividends.
Disclosure: I have no positions in any stocks mentioned, and no plans to initiate any positions within the next 72 hours.