Responsible Investing: Delivering competitive performance
- Growing interest in Responsible Investing raises an important question: Does pursuing social goals — limiting the range of potential investment opportunities — require sacrificing performance?
- A Nuveen analysis of leading RI equity indexes over the long term found no statistical difference in returns compared to broad market benchmarks, suggesting the absence of any systematic performance penalty.
- Moreover, incorporating environmental, social and governance (ESG) criteria in security selection did not entail additional risk. RI indexes and their broad market counterparts had similar risk profiles, based on Sharpe Ratio and standard deviation measures.
- Although return patterns were similar over the long term, there were significant return and tracking error differences between RI indexes and broad market benchmarks over shorter periods. By narrowing the range of eligible investments, the RI process introduced biases that caused short-term index performance to deviate from broad market benchmarks, resulting in tracking error.
- RI index construction methodology is an important determinant of tracking error. Investors should consider specific ESG methodology and the relevant market benchmark when selecting an RI strategy.
Growing interest in RI, but performance questions persist
Interest in Responsible Investing (RI) is growing rapidly. From 2012 to 2016, RI assets in the U.S. more than doubled to $8.72 trillion,1 according to the Forum for Sustainable and Responsible Investment (US SIF Foundation). This represents more than 20% of assets under professional investment management in the U.S., as tracked by Cerulli Associates.
RI strategies apply various environmental, social and governance (ESG) criteria in selecting public companies for inclusion in a portfolio. The process of incorporating nonfinancial criteria restricts the range of investment opportunities, potentially limiting returns. On the other hand, companies that wisely manage ESG risks and opportunities may also improve financial measures, potentially enhancing stock performance.
The key question for investors: Does investing in an RI strategy require sacrificing performance or taking on additional risk, compared to a broad market index?
Many studies on the performance of RI mutual funds versus non-RI funds have attempted to answer this question.2 However, the range, variety and diversity of RI fund management strategies make apples-to-apples comparisons difficult. Instead, Nuveen sought answers through a simpler comparison, analyzing the performance of several leading RI indexes versus broad market benchmarks. We focused on equity strategies because indexes with longer-term track records are readily available — and represent the majority of RI assets. It is important to note that RI indexes themselves are not perfectly comparable, due to differences in index construction and ESG evaluation processes. However, they provide a close proxy for RI as a strategy versus the broad market.
Our analysis found no statistical differences in RI index returns compared to the two broad market benchmarks.
The result: our analysis found no statistical difference in RI index returns compared to the two broad market benchmarks. In other words, RI can achieve comparable performance over the long term without additional risk, despite using a smaller universe of securities meeting ESG criteria. Exhibit 1 illustrates the similarity of cumulative returns for RI indexes and broad market benchmarks over the long term.
RI returns were comparable to broad market indexes
Returns for the RI indexes were similar to each other, and compared to the broad market. Ten-year average annual performance for the five U.S. RI indexes ranged from 7.67% to 9.16% versus 8.47% to 8.57% for the S&P 500 and Russell 3000 indexes, respectively. The gap between best and worst average annual performance spanned 149 basis points (Exhibit 2).
More importantly, statistical analysis showed no meaningful difference in returns when comparing RI indexes with relevant broad market indexes.5 Any return variations appeared to be random and not systematic. For the analysis, performance was measured from the period when weekly returns first became available for each index. Track records ranged from 15 years for the FTSE4Good US Index, to 27 years for the MSCI KLD 400 Social Index. Time periods were long enough to ensure that results were statistically valid.
RI index returns were similar to each other and compared to the broad market. Any return variations appeared to be random and not systematic.
Volatility and risk-adjusted measures were also comparable
Standard deviations for the RI indexes clustered fairly closely together, and were similar to the S&P 500 and Russell 3000 indexes:
- Average annualized standard deviations for the RI indexes ranged from 15.99% to 17.31% over the past 10 years, compared to 16.67% and 17.18% for the S&P 500 and Russell 3000, respectively (Exhibit 3).
- The spreads between standard deviations for RI indexes and benchmarks averaged only 26 basis points for the 10-year period.
- Even though some standard deviations topped 40% during the 2008–2009 market collapse, the maximum spread between RI indexes and their benchmarks averaged only 1.78% for the 10-year period.
Meanwhile, risk-adjusted returns also showed little variation from broad market indexes. Sharpe Ratios, or returns per unit of risk, also tracked fairly closely over various time periods, with average RI index Sharpe Ratios mirroring the underlying market or lagging only slightly.
- For the 10-year period, RI index average annual Sharpe Ratios ranged between 0.81 and 0.92, compared with 0.88 and 0.86 for the S&P 500 and Russell 3000, respectively (Exhibit 3).
With standard deviations of returns and Sharpe Ratios comparable between RI indexes and benchmarks, this suggests that incorporating ESG criteria in investment decisions doesn’t require taking on additional risk relative to broad market benchmarks.
RI indexes: Significant short-term performance variations
Although our results showed that index returns patterns were similar, they were not the same. In particular, performance variations increased significantly over short time periods, compared to broad market indexes.
Short-term differences are to be expected because any strategy that does not replicate the index, such as the RI process, introduces portfolio biases causing performance to deviate from broad market indexes. Exhibit 4 shows an example of how variable short-term performance can be. Measured on a 3- and 12-month basis, the rolling active return of the MSCI USA IMI ESG Leaders Index reveals considerable volatility, relative to an equivalent broad market benchmark, the Russell 3000 Index. This multicap RI index outperformed by as much as 6%, and underperformed by up to 4% on a 12-month basis.
RI index tracking error rates varied measurably
All the RI indexes had performance that deviated from broad market indexes, as measured by their tracking error. An important question for investors is whether some RI indexes more closely matched the performance of broad market benchmarks than others. We performed statistical analysis to determine whether the tracking error rates were similar or different across the five RI indexes. Results showed that tracking error differences were statistically significant and, therefore, important for investors to consider.
We compared the MSCI USA IMI ESG Leaders Index and Calvert U.S. Large Cap Core Responsible Index to the Russell 3000 Index because these RI indexes include smaller-capitalization stocks. (MSCI actually benchmarks this RI index against its own MSCI USA IMI Index.) The remaining RI indexes were compared with the S&P 500 because they included primarily large-capitalization stocks. Among the indexes, the MSCI USA IMI ESG Leaders Index showed the lowest tracking error at 1.68%, and the DJSI U.S. had the highest at 3.27% (Exhibit 5). All tracking error rates were measured from the inception date for each index, or first availability of weekly data through 29 Dec 2017. Although time periods varied by index, all were sufficiently long to ensure statistical validity.
Some RI indexes tracked the performance of broad market benchmarks more closely than others — differences that investors should consider.
Average tracking error for the MSCI USA IMI ESG Leaders Index was meaningfully lower than for the Calvert U.S. Large Cap Core Responsible Index. For RI indexes benchmarked to the S&P 500, the MSCI KLD 400 Social tracking error was lower than the DJSI U.S. and the FTSE4Good US by a statistically significant margin. However, there appears to be no statistical difference between tracking errors for the DJSI U.S. and the FTSE4Good US.6
The consistency of returns versus a selected benchmark is an important consideration for investors in measuring performance and managing risk. Tracking error does not introduce absolute risk per se, but is a source of relative risk versus a benchmark. Low tracking error indicates the index’s performance and risk characteristics closely match the benchmark’s profile.
Investors considering RI strategies may be indifferent to the level of tracking error, provided that long-term performance is comparable to the broad market. However, they should be aware of tracking error variations and their causes. Institutional investors, for example, may be constrained by client mandates to limit tracking error within specific ranges and against specific benchmarks.
Index methodology drives short-term return variability
Variations in tracking error and short-term returns, relative to benchmark indexes, are by-products of the RI process. Some approaches for incorporating ESG criteria can eliminate or concentrate holdings in certain industries, resulting in portfolio characteristics that differ from the market.
The five RI indexes use explicit ESG criteria to select a smaller subset of stocks from a universe of eligible companies. A particular strategy can involve excluding certain industries (such as gambling, tobacco and firearms), favoring companies that are leaders among their sector peers in managing relevant ESG risks and opportunities, or a combination of both. Decisions about how stocks are rated, selected and managed differentiate RI indexes from each other and the broad market.
The ESG evaluation and rating process itself can vary, as indexes use different research approaches to select companies for inclusion in the index. Company assessments may differ depending on the ESG approach, the range of factors considered, and relative emphasis on the “E,” “S,” or “G” components. The potential impact on performance of different ESG research approaches was beyond the scope of this paper. However, an understanding of these differences may help investors select an RI index appropriate for their needs. (See Appendix 4 for details on ESG rating and index methodologies.) Differences in index construction drive tracking error and short-term return variability, including some of the following factors:Sector/industry weighting: Certain industries or companies may tend to be excluded or have lower ESG ratings due to the nature of their business, such as gambling, tobacco or firearms. Conversely, some industries, like technology, may tend to receive higher ratings because they may face fewer ESG challenges than other industries. These variations can impact performance, and alter the investment style versus the benchmark. Omitting mining and energy companies due to environmental concerns, for example, could potentially exclude some value-oriented companies, and introduce a growth style bias to an index. Heavily capitalized and highly concentrated industries often include some of the largest companies, so including or excluding them could also skew the average size of companies in the index.
As an example, Exhibit 6 shows sector over/under weights for the MSCI USA IMI ESG Leaders Index, compared to MSCI’s own broad market index over a five-year period, and how these levels changed over time.
Number of holdings: In general, the more stocks in an RI index, the greater the opportunity to diversify risk, and match the performance of a broad market benchmark. An index with fewer stocks increases the potential for individual names or sectors to impact performance. The market capitalization target, such as large-cap vs. all-cap, can influence the number of stocks in the index.
The MSCI USA IMI ESG Leaders Index broadly targets stocks of all capitalizations with higher ESG ratings, starting with a universe of over 2,400 securities. In contrast, the DJSI U.S. begins with the 600 largest-cap U.S. companies in the Dow Jones Sustainability North America Index and selects the most highly rated 20%. As a result, the MSCI index held 1,110 mostly large- and mid-cap names as of 29 December 2017, while the Dow Jones index held only 133 large-cap names. The Calvert U.S. Large Cap Core Responsible Index had 725 holdings, the MSCI KLD 400 Social Index held 405 issues, and the FTSE4Good US Index had 229 holdings.
Overall, we found that RI indexes with a larger number of stocks tended to have lower tracking error. The MSCI USA IMI ESG Leaders Index and the MSCI KLD 400 Social Index had the largest number of holdings among indexes tracking the Russell 3000 and S&P 500, respectively, and the lowest tracking error. The DJSI U.S. had the fewest names and the highest tracking error.Efforts to address tracking error: RI indexes have various procedures for adjusting position size and weights to help the index more closely and consistently track a broad market benchmark. These may include optimizing sector/industry weightings, limiting the size of individual holdings, periodic rebalancing, and using buffers and ranges to limit turnover when making constituent changes.
For example, MSCI ESG Leaders indexes specifically target sector weights to match MSCI’s own underlying benchmarks. Calvert caps individual positions based on economic sector weights, but does not match a benchmark per se. The DJSI U.S. weights its holdings by market capitalization and relative industry weights, with individual constituents capped at 10%. The FTSE4Good US Index weights individual constituents based on their adjusted market cap, but does not adjust sector weights.
Tracking error considerations for investors
Since long-term RI index performance is comparable to the broad market, tracking error may matter most to institutional investors subject to specific limits over shorter time periods. It’s important to note, however, that RI indexes with lower tracking error are more likely to provide performance more consistent with a broad market benchmark.Investors should carefully consider specific RI index construction, the ESG evaluation process, and the underlying market benchmark when selecting a strategy. Index methodology drives tracking error, resulting in meaningful differences among RI indexes.
- RI indexes achieved long-term performance similar to broad market benchmarks, while pursuing social goals.
- Incorporating ESG criteria did not result in higher risk levels, measured by Sharpe Ratio and standard deviation. By constraining their investment universes, RI indexes introduce tracking error and greater short-term return variability, although the magnitude depended on how the index is constructed.
- RI indexes differed significantly in how closely they tracked broad market indexes. Understanding index methodology is critical to properly evaluating and selecting a specific index. In particular, investors should consider an index’s breadth of holdings, market-cap and benchmark exposure, and steps taken to reduce tracking error and help improve consistency, relative to its broad market benchmark.
2 See Appendix 1 in the Download.
3 See Appendix 2 in the Download for study methodology.
4 The Russell 3000 and S&P 500 indexes were selected as the most appropriate proxies against which most investors might measure RI performance. They have not necessarily been constructed for these comparisons, and do not necessarily represent what would be an appropriate comparison as a parent index.
5 See Appendix 2 in the Download for the statistical analysis of RI index returns vs. broad market indexes.
6 Statistical comparison is based on the shorter time period common to both indexes, based on the 10 Jan 2003 inception of the FTSE4Good US Index. The average tracking error for the DJSI U.S. Index shown in Exhibit 5 is based on the index’s 08 Jan 1999 inception date. See Appendix 3 in the Download for tracking error statistical analysis.
Risks and other important considerations
Investments in Responsible Investing strategies and products are subject to the risk that because social criteria exclude securities of certain issuers for non-financial reasons, investors may forgo some market opportunities available to those that don’t use these criteria. Investment products may be subject to market and other risk factors. See applicable information regarding responsible investing for details.
These views are presented for informational purposes only and may change in response to changing economic and market conditions.
The investment advisory services, strategies and expertise of TIAA Investments, a division of Nuveen, are provided by Teachers Advisors, LLC and TIAA-CREF Investment Management, LLC. Securities offered through Nuveen Securities, LLC.