Is socially screened investment a "dumb" idea?
Our experience at Vantage Global Advisors indicates it is not "dumb" to be socially conscious investors; but one must look at how a manager implements the investment process, the same way one must look at managers who do not invest in a socially conscious style. As with any investment class, returns can vary dramatically across managers.
We refer to socially responsible investing as using social screens to exclude stocks from portfolios on the basis of specific criteria. Vantage uses screens provided by Kinder, Lydenberg, and Domini & Co.: military, nuclear power, product (alcohol, tobacco and gambling); and environment.
There is a growing literature in academic and professional investment journals that suggests socially responsible investing might produce higher risk-adjusted portfolio returns than merely using all available stocks in the equity universe.
When one considers the possible corporate expenses of fines and litigation, an investor might expect lower returns to companies that damage the natural environment; sell liquor and other alcoholic products; produce, design or use nuclear power; engage in gambling; or are large defense contractors.
The 1993 study by Sally Hamilton, Hoje Jo and Meir Statman found 17 socially responsible mutual funds established prior to 1985 outperformed the traditional mutual funds for similar risk for the 1986-1990 period. However, the relative monthly outperformance of seven basis points was not statistically different from zero. It also is not obvious what criteria were used to determine the socially responsible universe in that study.
Two recent studies by J. David Diltz of the University of Texas at Arlington found no statistically significant difference in returns for 28 stock portfolios generated from a universe of 150 securities during the 1989-1991 period. Mr. Diltz found favorably rated firms significantly outperformed unfavorably ranked firms using only the environmental and military screens.
It is the case that 24 socially screened mutual funds in the Morningstar universe have substantially underperformed the S&P 500 during the past five and 10 years. However, the difference between the average return on socially screened equity mutual funds and the 2,034 unscreened equity mutual funds drops from -417 basis points over the past five years to -105 basis points over the past 10 years, a less meaningful differential, particularly given the very small number of socially screened equity mutual funds with long-term track records. There are only six socially screened equity mutual funds with five-year track records in the Morningstar universe, and only Dreyfus Third Century and Parnassus have 10-year records.
On the other hand, the College Retirement Equities Fund Social Choice Account, a balanced account containing 62% socially screened equities and 38% debt, has produced an annualized equity return of 12.41% for the five years ended Dec. 31, whereas its unscreened Stock Fund produced a five-year equity return of 13.58%. It is very difficult to compare the CREF Social Choice Account equity performance with the CREF Stock Fund performance because the Stock Fund contains 20% growth companies and 20% international stocks. However, it does not appear the CREF Social Choice Account equity component has substantially underperformed the CREF Stock Fund equity component.
Our primary point is that an investor apparently has not suffered in the equity component Social Choice Account relative to the unscreened Stock Fund. The Social Choice Account exceeded $1.5 billion in assets by December.
Further evidence of reasonable socially responsible investment performance is found in the performance of the Domini Social Index, a market-capitalization index composed of approximately 400 companies meeting the KLD social criteria. The DSI produced a net return of 15.81% for the five years ended Dec. 31, exceeding the corresponding S&P 500 return of 15.22%.
The evidence is mixed on socially responsible investing, the CREF Social Account and DSI Index have equalled or exceeded their benchmarks whereas many socially screened mutual funds in the Morningstar universe have underperformed.
Vantage's experience is that a socially screened universe return is not significantly different from an unscreened universe return for the 1987-1994 period.
Vantage is the adviser to the Lincoln National Social Awareness Fund in its Multi Fund variable annuity family. The Vantage social composite, including the Social Awareness Fund and exceeding $700 million in assets, produced a five-year annualized net return of 16.51% for the five years ended Dec. 31. The socially responsible benchmark was 14.91% for the five years and the S&P 500 was 15.22%. The annualized 31 basis points of difference represents the cost of social investing.
In its research, Vantage examined the returns of an unscreened equity universe composed of 1,300 equity stocks and a socially screened universe of approximately 950 stocks and test if there are statistically significant differences in the average returns of the two universes. We find there is no significant difference between the average monthly returns of the screened and unscreened universes during the 1987-1994 period. Indeed, from January 1987 to December 1994, there is less than a 15 basis point differential in equally weighted annualized stock returns.
The Vantage Global Advisors' unscreened 1,300 stock universe produced a 1.068% monthly average return during the January 1987-December 1994 period, such that a $1 investment grew to $2.77. A corresponding investment in the socially screened universe would have grown to $2.74, representing a 1.057% average monthly return. There is no statically significant difference in the respective return series, and more importantly, there is no economically meaningful difference in the return differential.
The variability of the two return series is almost equal during the 1987-1994 period. One can test for statistically significant differences in the two return series using the F-test, which examines the differences in series mean (returns) relative to the standard deviations of the series. When one applies the F-test, one finds series are not statistically different from one another.
As an example, let us examine the financial characteristics of the stocks in the unscreened and socially screened Vantage Global Advisors' universes as of December 1994. The unscreened VGA universe of 1,300 stocks had BARRA growth and book-to-price sensitivities of 0.185 and 0.306, whereas the socially screened VGA universe had corresponding BARRA growth and book-to-price sensitivities of 0.269 and 0.279, respectively. The unscreened universe had an average market capitalization of $3.433 billion in December 1994, whereas the socially screened universe had a mean capitalization of $2.796 billion. The average BARRA growth and book-to-price sensitivities of the excluded securities were -0.164 and 0.414, respectively, and the average market capitalization of the excluded stocks exceeded $6.1 billion. Thus, screened-out stocks had higher market capitalizations and were more value oriented than the unscreened universe, a condition noted by Lloyd Kurtz and Dan DiBartolomeo in a study published in the Journal of Investing (Fall 1996).
There was a statistically significant difference between the price-to-book ratios of the unscreened and screened VGA universes. Professors Eugene F. Fama and Kenneth R. French at the University of Chicago found smaller stocks with lower price-to-book ratios tended to outperform larger stocks with higher price-to-book ratios in the very long run. The higher price-to-book ratio of the screened universe represents a risk exposure to socially responsible investors. The screened universe is more sensitive to BARRA growth factor return than the unscreened universe and this exposure should help relative performance for socially responsible investors when the BARRA growth factor return outperforms the BARRA value factor (as measured by the book-to-price, basis point) return.
The higher growth sensitivity helped Christopher Luck and Nacy Pilotte in another Journal of Investing study (Fall, 1993) find that the Domini Social Index outperformed the S&P 500 Index during the May 1990-September 1992 period. Luck and Pilotte used the BARRA Performance Analysis package and found the 400 securities in the Domini Social Index produced an annualized active return of 233 basis points relative to the S&P 500 and specific asset selection composed 199 basis points of the active return. Luck and Pilotte noted the May 1990-September 1992 period was characterized by positive growth factor and size returns (smaller stocks outperformed larger-capitalized stocks as a rule during this period).
Superior asset selection might have been created as Kinder, Lydenberg, Domini created the DSI in May 1990 by including non-S&P 500 stocks with "good" records on corporate citizenship, product quality, board representation of women and minorities. KLD developed criteria to establish the records of socially responsible firms. For example, in March 1992, KLD produced a screen of 24 publicly traded firms that dealt in or used recycled material. A second screen of 20 companies known for quality products was developed by KLD, although one-third of these firms failed other screens. In August 1992, 12 firms were recognized by a KLD diversity screen that identified firms with four or more (or at least one-third of the members if the firm had fewer than 12 members) women or minority board seats. Additional KLD screens in August 1992 identified 10 firms with women and minority chief executive officers and 20 firms that possessed notable records on promoting women and minorities. KLD screens established criteria to substantiate good corporate citizenship. It is important to note these criteria did not "cost" the investor any meaningful average return during the 1987-1994 period and might have produced positive active (relative to the S&P 500) returns during some subperiods.
Although we have noted that much of the recent socially responsible literature has been concerned with the impact of screening on BARRA growth and value factors, in a forthcoming study in the Journal of Investing, we found no statistically significant differences in the information coefficients - the correlation between ranked composite scores and ranked subsequent total returns for a composite model - in the Vantage unscreened 1,300 stock universe and its corresponding socially screened universe. The regression-weighted model combined value and growth (a proprietary variable created from I/B/E/S data). The ICs of the unscreened and socially screened universes are shown in the chart.
The data show there has been no statistically significant difference between the average returns of a socially screened and unscreened universe during the 1987-1994 period. Socially conscious investing need not be a dumb idea, but one should be attentive when selecting a socially screened mutual fund or manager. Performance can vary dramatically across managers.