What are the differences between socially responsible investment (SRI) indexes, both in terms of structure and performance? A roundtable panel at the Green Mountain Summit on Investor Responsibility that convened in Stowe, Vermont late last month addressed these issues, focusing on SRI indexes covering the US market. Particularly interesting was a study of five US SRI indexes presented by Dan diBartolomeo, founding president of Northfield Information Services. Northfield is a Boston-based firm that provides quantitative analysis to investment managers.
Mr. diBartolomeo prepared this study for the conference concurrent to KLD Research and Analytics contracting Northfield to do a ten-year back test of its recently launched Select Social Index. KLD uses Northfield software to optimize the weighting of companies in the index.
Mr. diBartolomeo found that while the structures of SRI indexes do indeed differ significantly, there is no statistically significant difference in performance. In other words, social investors, especially institutional investors who typically construct broad, benchmark-tracking portfolios, enjoy a relative diversity of SRI index styles to choose from, all of which generate performance in the same ballpark.
The study compares the FTSE4Good US Index, the US component of the Dow Jones Sustainability Index (DJSI), the Calvert Social Index (CALVIN), KLD’s Domini 400 Social Index (DSI), and the KLD Social Select Index. The time under consideration runs from December 1, 2001 to November 31, 2003, the period for which data was available for all five indices (results for the KLD Select Social Index were based on the back test)
"Despite some seemingly large differences in cumulative returns, none of the differences in monthly returns amongst the five funds could be considered statistically significant over the two-year period studied," writes Mr. diBartolomeo, who serves on the judges’ panel for the Moskowitz Prize, awarded yearly to the best quantitative research paper on SRI by the Social Investment Forum (SIF), the SRI industry organization. "The differences in performance could have arisen from coincidences, rather than anything actually being superior about one index compared to another."
The study found that the indexes produced the following 2-year cumulative returns: the FTSE4Good US Index, -5.92 percent; the DJSI, -4.00 percent; the CALVIN, -3.93 percent; the DSI, -1.23 percent; and the KLD Select Social Index 2.41 percent.
"To the extent that there have been differences in performance, these differences tended to be in keeping with observable differences in the characteristics of the particular index portfolios," Mr. diBartolomeo adds.
For example, the CALVIN was the most heavily weighted in technology stocks (28 percent), which expectedly led to its having the lowest dividend yield. This overweight also contributed to the CALVIN having the highest estimated beta at 1.10, according to the study. (Beta is a measure of volatility, or systematic risk, compared to the market as a whole–a beta above 1 predicts increased volatility, and below 1 predicts decreased risk.)
The study also computed the monthly tracking error of each index against the other four, with lower values representing closer tracking and higher values indicating more distinct behavior. The DSI and FTSE indexes tracked each other closest, with a cross-tracking error of 0.47. The highest cross-tracking error counts went to the DJSI.
"The DJSI (US subset) had the most distinct behavior, as is understandable given that it is really just the US domestic portion of a global index," explains Mr. diBartolomeo.
He refers to the fact that the study extracts the US-domiciled constituents from the DJSI in order to create a domestic index for comparison purposes, as DJSI does not publish an explicit US index. This process unavoidably skews the index, as is reflected in the resulting portfolio containing significantly fewer stocks than the other indexes, and the average market capitalization being about one third larger than the other indexes.
"Investors who are considering using one of these indices should be aware of the meaningful differences in sector allocation, systematic risk, and growth/value orientation that are embedded in the various index construction schemes," Mr. diBartolomeo concluded.
