Performance Attribution: A Look at KLD Indexes in 2005

By: Chris McKnett, Index Project Manager

         The family of KLD Indexes has two stories to tell:

• A social story based on the screens employed to construct and maintain the indexes, and
  

• A performance story that is expressed in both absolute and relative terms.

         Both aspects help investors interpret the financial performance of socially screened indexes.

         KLD constructs social indexes using screens that reflect the issues and interests of social investors. As a result, the indexes represent the portfolios of social investors in two ways: as performance benchmarks and as universes of companies appropriate for investment. As benchmarks, they help investors define and analyze the differences between screened and traditional market benchmarks.

         In this article, we look at the 2005 performance of the Domini 400 Social^SM Index (DS 400), KLD Broad Market Social^SM Index (BMS), KLD Large Cap Social^SM Index (LCS) and the KLD Select Social^SM Index (SS). We compare each KLD index to an unscreened market benchmark, such as the S&P 500 or the Russell 3000. The findings illustrate some general characteristics of social indexes. In addition, they shed light on differences between KLD indexes, showing the variance in their returns and the sources of this divergence.

Performance

         While this analysis focuses on 2005, performance in any given period should be framed in a longer, more representative context. Table 1 presents one-, three- and five-year performance of KLD indexes and their respective unscreened benchmarks.

Table 1: Index Performance¹

Source: Russell Indexes, Dow Jones & Co, and Standard & Poor's

Table 2 shows the relative performance of KLD Indexes by subtracting the returns of the benchmark from those of the KLD index. As demonstrated in the table, each KLD index underperformed its benchmark in 2005.

Table 2: 2005 Performance Differences

         This difference is captured in Graph 1, which illustrates how each KLD Index performed relative to its benchmark. The x-axis represents the benchmark baseline of 0.00% and each line captures the KLD Index performance on the y-axis relative to the benchmark. The performance variance for each index is explained by a combination of unique factors, yet nonetheless some common themes emerge.

Graph 1:
Comparison of KLD Indexes Versus Benchmark - 2005

Source: Russell Indexes and FactSet Research Systems

Sector Weightings and Stock Selection

         The impact of variance in sector weights introduced by social screens is an important element of index performance. Sector weightings relative to each index benchmark provides insights into both the social and performance story. For the DS 400, BMS and LCS, the largest contributor to underperformance was relative underweight positions in the Energy sector. Other sector and company weights contributed to performance variance, but when taken as a group they effectively canceled each other out. The net result was that the Energy sector impact on performance accounted for nearly 100% of BMS and LCS underperformance and for roughly 60% of DS 400 underperformance.

         The energy companies in the DS 400 outperformed those in the S&P 500 by 52.73% to 31.10%, a positive difference of 21.63%. However, the significant difference in weighting caused the DS 400 to underperform in energy.

         Stock selection within sectors is another important attribution factor. The LCS and BMS outperformed the DS 400 in 2005. The principal reason resides in better performance of holdings in the Health Care and Financials sectors during the year.

         The DS 400, BMS and LCS are market capitalization weighted, meaning that holdings are weighted in proportion to their size. In contrast, the SS selection and weighting process uses optimization to maximize exposure to KLD’s ESG scores. The SS concentrates positions in companies with higher KLD scores, resulting in larger company bets than the other KLD indexes. In 2005, SS performance was hurt most by stock selection within the Health Care and Financials sectors.

         Table 3 shows the performance of every sector relative to its market benchmark for each KLD Index. It is important to understand that these figures capture the effects of both sector weights and sector performance. Positive contributions are shown in blue and negative contributions are shown in red.

Table 3:
Performance Attribution by Sector in Basis Points – 2005

Source: FactSet Research SystemS, Inc.

         The table shows that all KLD Indexes outperformed the benchmarks in the Information Technology, Consumer Discretionary and Materials sectors and underperformed the benchmarks in the Industrials, Utilities, and Energy sectors.

         A common thread to outperformance was that within the Information Technology, Consumer Discretionary and Materials sectors, KLD Index stock selection was superior to the benchmark. In addition, KLD Indexes were overweighted in both the Information Technology and the Consumer Discretionary sector, which further contributed to relative outperformance. The DS 400, BMS, and LCS underweighted Materials, while the SS overweighted the sector.

         Relative underperformance for all four indexes in the Industrials, Utilities, and Energy sectors was attributable to underweighting. Holdings in these sectors tend to be limited by the social and environmental screens, and this characteristic dampens index performance during markets favorable to those sectors. In addition, stock selection in the Industrials and Utilities sector contributed to sector underperformance.

         Screening (used in the DS 400, BMS and LCS) and scoring (used in the SS) introduce similar but not identical biases. In a general sense, Information Technology firms tend to have a small environmental footprint while Energy firms tend to have a substantial environmental footprint. The performance data suggests that environmental factors, in aggregate, stand out as the most dominant screen.

Conclusion

         As we have shown, social screening influences performance of social indexes and portfolios. This influence may have a positive or negative effect depending on market conditions.

         In 2005, the market was broadly affected by higher energy prices, driven by a rise in the price of oil to more than $70 per barrel. The environment screens applied by KLD lead to a substantial underweight in the Energy sector. The DS 400, BMS and LCS hold no integrated oil companies, although they do hold smaller oil companies as well as oil and gas exploration firms. The largest negative impact on the performance of these indexes came from underexposure to energy. By contrast, the SS, while underweight, was closer to the market benchmark weight in energy, and its performance in this area was better than the other KLD indexes.

         In 2005, the DS 400, BMS and LCS were hurt more by their sector weight variations than they were helped by their stock selection. The SS was hurt by both its sector weightings and stock selection -with stock selection having the largest performance impact.

         Attribution analysis depicts correlations between KLD Indexes. For example, consistent with their benchmarks, the weightings in the BMS and LCS are substantially similar. Given this fact, one could hypothesize that contributors to relative performance for each index would substantially overlap. Attribution analysis confirms this hypothesis.

         Attribution can also be utilized to illustrate differentiation among KLD Indexes. For example, the DS 400 was the top performer in the Consumer Staples sector and the bottom performer in the Energy sector. Further examination of performance attribution data would reveal whether this was attributable to sector weighting, stock selection, or both.

         A review of Table 3 reveals the uniqueness of the SS. The social story of the index is differentiated from that of other KLD indexes and informs the performance story. The SS offers investors a risk-controlled ESG strategy and adds diversity to the KLD Index family.

         While the social and performance stories of KLD Indexes are interwoven, attribution analysis is a useful analytical tool to identify what drives the performance story, which, in turn, sheds light on the impact of the social construct of each index on performance.

Russell 3000® Index and Russell 1000® Index are trademarks of Frank Russell Company. S&P 500® Index is a trademark of Standard & Poor’s. Domini 400 Social^SM Index, KLD Broad Market Social^SM Index, KLD Large Cap Social^SM Index and KLD Select Social^SM Index are service marks of KLD Research & Analytics, Inc.

¹ The KLD Global Climate 100^SM Index is excluded from this analysis due to a July 1, 2005 launch date.