The KLD Broad Market and Large Cap Social Indexes at Five Years: A Risk and Return Analysis

KLD Indexes Welcomes Chris McKnett

The KLD Broad Market and Large Cap Social Indexes
at Five Years: A Risk and Return Analysis

         On January 1, 2001, KLD launched the Broad Market Social Index (BMSI) and Large Cap Social Index (LCSI). The risk and return analysis in this article offers new information about the KLD BMSI and LCSI approach to social investing for the first five years of operation.

         The Indexes are, respectively, subsets of the Russell 3000® Index and Russell 1000® Indexes. The Russell 3000® Index consists of approximately 98% of total U.S. equity market capitalization and the Russell 1000® Index represents approximately 92%.

         To construct the Indexes, KLD applies a two-step screening process to the Russell universe. KLD first excludes from consideration companies involved in alcohol, tobacco, firearms, gambling, nuclear power and military-weapons. KLD excludes companies that fail these screens regardless of any other positive social and environmental attributes.

         After that cut, KLD evaluates all remaining companies for performance on environmental, social, and governance factors, relative to industry peers and in relation to the larger index universe. Since no company is perfect, KLD selects companies with above average performance on environmental, social, and governance issues.

         Mirroring the composition of the Russell Indexes, all LCSI holdings are also BMSI holdings. Both the BMSI and LCSI are reconstituted annually on June 30 which is when Russell reconstitutes its Indexes. Prior to annual reconstitution, companies may be removed from the KLD Indexes due to corporate actions such as mergers, acquisitions, or bankruptcies, or infrequently for social or environmental concerns. As of December 31, 2005, the BMSI had 2,211 constituents and the LCSI had 669 constituents.

Table 1: Comparison of Cumulative Total Returns
(January 2001 – December 2005)

Source: Russell Indexes and FactSet Research Systems

         Since its inception in January 2001, the BMSI underperformed the Russell 3000, and the LCSI underperformed the Russell 1000, on a total return basis. The graph above shows the Cumulative Total Return of the BMSI (4.25%), LCSI (0.97%), Russell 3000 (8.16%) and Russell 1000 (5.48%) for the past five years.

Table 2: Comparison of Multi-Period Annualized Returns (January 2001 – December 2005)

Source: Russell Indexes and FactSet Research Systems

         Annualized Returns are the average return gained each year over one or more years. The multi-period annualized returns chart above shows the annualized returns for different time periods throughout the five-year history, beginning with the most recent year. Annualized returns over the past five years show that the BMSI underperformed the Russell 3000 by an average of 0.74% (0.84% versus 1.58%), and the LCSI underperformed the Russell 1000 by an average of 0.88% (0.19% versus 1.07%).

Table 3: Comparison of Upside/Downside Capture
(January 2001 – December 2005)

Source: Russell Indexes and FactSet Research Systems

         Upside and Downside Capture Ratios measure the difference in return of the portfolio and the underlying benchmark during an up and down market, respectively. Upside capture is defined as the ratio of the cumulative monthly portfolio return to the cumulative monthly benchmark return when the market rises; downside capture is similarly defined as the ratio of the cumulative monthly portfolio return to the cumulative monthly benchmark return when the market declines.

         The BMSI upside ratio of 111.01% indicates outperformance of its benchmark, the Russell 3000, during market rises. The BMSI downside ratio of 105.94% indicates slight underperformance of its benchmark during market declines.

         The LCSI upside ratio of 110.83% indicates outperformance of its benchmark, the Russell 1000, during market rises. The LCSI downside ratio of 106.27% indicates slight underperformance of its benchmark during market declines.

         Both the BMSI and LCSI outperform their respective benchmarks during market rises and underperform during market declines.

Table 4: Comparison of Performance Statistics
(January 2001 – December 2005)

Source: Russell Indexes and FactSet Research Systems

         Annualized Returns are the average return gained each year over one or more years. Annualized returns over the past five years show that the BMSI underperformed the Russell 3000 by an average of 0.74% (0.84% versus 1.58%), and the LCSI underperformed the Russell 1000 by an average of 0.88% per year (0.19% versus 1.07%).

         Annualized Standard Deviation measures the average deviations of historical annual returns from the mean of the data set. It is often used as one measure of risk, as a larger standard deviation implies larger volatility of portfolio returns. The difference between the BMSI standard deviation of 16.78 and the Russell 3000 standard deviation of 15.00 is 1.78, indicating the higher risk associated with the BMSI. Similarly, the difference between the LCSI standard deviation of 16.74 and the Russell 1000 standard deviation of 14.88 is 1.86.

         Annualized Tracking Error measures the average amount by which the performance of the portfolio differs from that of the benchmark. It is defined as the standard deviation of returns relative to the benchmark. The BMSI tracking error signifies the likelihood that BMSI annual returns will be within + or – 2.80% of the Russell 3000. The LCSI exhibits similar characteristics with a tracking error of 2.96% benchmarked to the Russell 1000.

Table 5: Comparison of Risk Statistics
(January 2001 – December 2005)

Source: Russell Indexes and FactSet Research Systems

         Beta is a measure of portfolio volatility relative to the benchmark of the portfolio. The BMSI and LCSI both have a beta of 1.11 compared to their respective benchmarks, the Russell 3000 and Russell 1000. This indicates a high correlation with fluctuations in their benchmarks and indicates a risk level slightly higher than the benchmarks.

         Alpha measures the risk-adjusted performance relative to the benchmark of a portfolio, or the value added by the selection of stocks in the portfolio. The BMSI alpha of -0.06 and the LCSI alpha of -0.07 demonstrate a performance marginally less than would be predicted given the beta value for each index. This statistic quantifies the average annual risk-adjusted return of the BMSI and LCSI relative to the Russell 3000 and Russell 1000.

         Sharpe Ratio is another measure of risk-adjusted performance calculated by dividing the excess annualized return of a portfolio – measured by subtracting the risk-free rate from actual portfolio return – by its standard deviation. The BMSI Sharpe Ratio value of -0.08 is less than the Russell 3000 value of -0.04 and indicates lower annualized return per unit of risk. The LCSI, with a Sharpe Ratio of -0.12, also exhibits lower annualized risk-adjusted performance compared to the Russell 1000, with a Sharpe Ratio of -0.08.

         R-Square is a measure of correlation between a portfolio and the benchmark. The BMSI and LCSI both have values of 0.98, meaning that 98% of the variation in both indexes’ price changes could be attributed to changes in their benchmarks. The other 2% is attributable to a variety of factors.

Conclusion

         The first five years of the BMS and LCS indexes were marked by two distinct periods corresponding to phases in the market cycle: two years of downward trend followed by three years of upward trend.

         KLD launched the BMSI and LCSI in January 2001, nine months into the bear market that followed the end of the technology bubble. The market continued its downward movement for the next two years.

         The capture analysis suggests that, on average, the BMSI and LCSI can be expected to underperform when returns are negative and outperform when returns are positive. The first two years created a performance deficit that the indexes are overcoming during this later period of growth.

         Annualized returns for the three years ended December 2005 show very small performance differentials for both indexes relative to the benchmarks, with the BMSI underperforming by 0.01% and the LCSI underperforming by 0.05%. During this time period the indexes saw their greatest cumulative total returns, and given the upside capture ratio characteristics, each index narrowed the performance gap with its benchmark.

         The screening and index selection process, which eliminates approximately 21% of companies from the Russell 3000 and 28% of companies from the Russell 1000, introduces some sector bias into the BMS and LCS indexes relative to the Russell 3000 and Russell 1000. Although not analyzed here, the differences in sector weights contribute to the tracking error and performance differentials between the BMSI and the Russell 3000, and the LCSI and Russell 1000.

         KLD created the Broad Market Social Index and Large Cap Social Index in 2001 to give socially responsible investors the most comprehensive market benchmark and the largest possible investment universe to choose from. After five years, the indexes are accumulating a performance track record that will support the ability of investment managers, consultants, trustees and investors to understand the performance, risk, and fundamental characteristics of their socially screened portfolios.

The Russell 3000® Index and Russell 1000® Index are owned by the Frank Russell Company.

KLD Indexes welcomes Chris McKnett

Chris McKnett has replaced Priya Khetarpal as Index Project Manager for KLD Indexes. Chris supports various components of new index projects, including index operations, product creation and product marketing, in addition to ongoing index maintenance and client services responsibilities.

         Chris previously worked for American Century Investments, a large U.S. investment management firm. While at American Century, he supervised Registered Representatives and achieved the Accredited Asset Management Specialist (AAMS) and Chartered Mutual Fund Counselor (CMFC) certifications.

         Chris holds an MBA with a concentration in Values Based Leadership from the Daniels College of Business at the University of Denver. His thesis equivalent was titled “SRI Indices: A Comparative Analysis” and he was a contributing writer to “Integration: Critical Link for Corporate Citizenship”, a research report by The Center for Corporate Citizenship at Boston College.

         While pursuing his degree, Chris was President of the Net Impact chapter, recipient of the 2005 Wall Street Journal Student Achievement Award, and nominated for the Daniels Award for Excellence.