Please Note: Blog posts are not selected, edited or screened by Seeking Alpha editors.

Does A Socially Responsible Portfolio Outperform A Conventional Portfolio? (The Case Of The Jantzi Social Index)

Summary

Socially responsible investment takes financial return as well as environmental, social, and governance issues into consideration.

There is a controversy among investors regarding the performance of socially responsible portfolios compared to conventional ones.

In this article, the performance of a portfolio based on members of Jantzi Social Index is compared with a comparable conventional portfolio.

Socially responsible investment (NYSE:SRI) not only considers the financial return, but also takes into consideration the non-financial concerns of investors such as environmental, social, and governance issues known as ESG factors. ESG factors are activities companies engage in to reduce or eliminate the adverse effects of their operations on environment, society, and stakeholders. A socially responsible (NYSE:SR) company is one that engages in corporate social responsibility (NYSE:CSR) activities, of which the most important are the ESG factors. 

SRI is growing fast in terms of the amount of investments. For Europe, we see an increase in the capitalization of the SRI market from $336 billion in 2003, to $13608 billion in 2014, a growth of 3950%. For the US, this growth is 204%, from $2164 billion in 2003 to $6572 billion in 2014. As SRI is growing in financial markets, more SR indices and funds are introduced to the financial world. An example of such indices is the Domini 400 Social Index. It consists of 400 SR companies in US. This index is designed to help SR investors incorporate ESG factors in their investment choices. An example of SR fund is the Domini Impact Equity Fund, a diversified portfolio of SR companies designed to have expected long-term total return, while protecting social and environmental concerns of investors. 

In this article, our focus is on the performance of a SR portfolio constructed from the members of the Jantzi Scoail Index (JSI). JSI is managed by Sustainalytics, a global investment research firm specializing in ESG research and analysis. Sustainalytics defines JSI as a socially screened, market capitalization-weighted common stock index modeled on the S&P/TSX 60 which consists of 50 Canadian companies (formerly 60 companies) that pass a set of broadly based ESG criteria, and are listed on the Toronto Stock Exchange (TSX). The underlying pool of companies used by Sustainalytics to choose the constituents of JSI is S&P/TSX Composite Index. Sustainalytics uses two exclusionary criteria to form JSI, "product involvement" and "major negative ESG impact." Sustainalytics ranks corporate controversies on a scale of 1 to 5 with 1 being least controversial and 5 being most controversial. Companies with Category 4 or Category 5 controversies cannot be members of JSI. The constituents of JSI are updated on a yearly basis in March. 

As SRI grows in financial markets, the number of studies comparing its performance with that of conventional investments is increasing. There are two main views on this relative performance: that of supporters and that of opponents. Supporters of SRI argue that the performance of SR portfolios is either superior or similar to the performance of conventional portfolios, while opponents of SRI believe that SR portfolios underperform conventional portfolios. 

Based on Modern Portfolio Theory (MPT) limiting the constituents of a portfolio to only SR companies, decreases the size of investment pool and results in a less diversified portfolio. Based on the premise of MPT, a SR portfolio is expected to underperform a more diversified portfolio on a risk-adjusted basis. This is the main argument of the opponents of SRI. They believe that SR portfolios underperform conventional ones since their pool of investment is limited to SR companies. On the other hand, supporters of SRI believe that the performance of SR portfolios is similar to that of conventional ones, or can even be better thanks to the superior financial performance of the constituents of a SR portfolio. Many studies have been conducted to analyze the debate between supporters and opponents of SRI. For example, as shown in the figure below, JSI is compared with two conventional indices, S&P/TSX Composite and S&P/TSX 60. As we can see, JSI almost moves in line with these two indices. Moreover, from inception to December 2016, JSI achieved an annualized and cumulative return of 6.53% and 193.14% respectively. In the same period, S&P/TSX Composite and S&P/TSX 60 achieved annualized return of 6.14% and 6.09%, and cumulative returns of 175.34% and 173.39% respectively.  

Figure 1: The performance of JSI compared to S&P/TSX Composite and S&P/TSX 60

In comparing these numbers, we have to be cautious. Apart from the fact that these numbers are not adjusted for risk, the industry, size, and risk profile of the members of JSI, S&P/TSX Composite, and S&P/TSX 60 might be different which can have significant effect on the performance comparison. 

To perform a proper comparison, we create a SR portfolio consisting of companies included in JSI, using the same weights they carry in JSI. To have a comparable benchmark for this portfolio, we create a hypothetical conventional portfolio from conventional (non-SR) companies that match the industry and size of companies in our SR portfolio with exactly the same weights. To match for industry, we use the North American Industry Qualification System (NAICS) codes. To match for size, we use the total assets of companies. The pool that we choose the companies from is TSX. The period of this study is between March 2003 and December 2015. Within this period, the members of the matched portfolio is updated in those months that changes happen in the members of JSI. 

After finding the matched portfolio, we compare it with our SR portfolio based on several measures. In Table 1, the results of this comparison is shown. As we can see in this table, the SR portfolio outperforms the matched portfolio based on total return, total risk, and Sharpe ratio, the differences being significant at 1% level. Looking at the market risk of the the two portfolios, we can see that although our SR portfolio has lower market risk, the difference is not statistically significant. However, when we look at the Treynor ratio, we can see that our SR portfolio significantly (at 1% level) outperforms the matched portfolio. Another important measure for performance comparison is Abnormal Return (alpha). Two very common methods to find abnormal return is Capital Asset Pricing Model (CAPM) and Fama-French three factor model. Based on the abnormal return (both CAPM and Fama-French), the SR portfolio outperforms the matched portfolio, the differences are significant at 1% level. 

Apart from the portfolio performance measures, we have to also look at other important measures, such as Tobin's Q. Tobin's Q is the ratio of the "market value" of a company to the "replacement cost" of that company. As shown in Table 1, since the SR portfolio has higher Tobin's Q compared to the matched portfolio, we can conclude that SRI creates value in a portfolio. Another important measure is leverage ratio. Leverage ratio can be calculated as "total debt" divided by "total assets." As we can see in Table 1, our SR portfolio has higher leverage ratio compared to the matched portfolio, which indicates that, on average, the constituents of our SR portfolio have higher level of debt in their capital structure in comparison to the members of the matched portfolio. We can also argue that our SR portfolio's higher Tobin's Q is partly due to its higher leverage. However, since the difference in leverage ratios is only a few percent, economically, such a small difference cannot have a significant effect on the valuation. As a result, we can almost be confident that the higher Tobin's Q of the SR portfolio is due to the ability of SRI in creating value. 

As a result, our study indicates that investing in socially responsible portfolios not only satisfies the ethical concerns of investors, but also leads to a superior risk-adjusted performance. However, we must be cautious about the magnitude of the outperformance. It could be partly due to our matching technique that does not allow any of our socially responsible companies be presented in the matched portfolio. As a result, the matched portfolio could be deprived from the good performance of some of the top-performing companies in our SR portfolio. 

Table 1: Results of the performance comparison of SR and matched portfolio 

Measure SR Portfolio Matched Portfolio t-Test
Total Return 1.272% 0.711% 2.62***
Total Risk 6.689% 8.552% -6.20***
Sharpe Ratio 0.178 0.059 3.97***
Market Risk (Beta) 0.840 0.910 -0.47
Treynor Ratio 1.249 0.604 2.87***
Abnormal Return (CAPM) 0.567% 0.100% 2.61***
Abnormal Return (Fama-French) 0.572% 0.101% 2.61***
Tobin's Q 1.519 1.363 26.01***
Leverage 22.516% 21.530% 11.57***

t-Test is used to investigate if the differences between measures are statistically significant. *, **, *** shows significance at 10%, 5%, and 1% level respectively. 

Any result achieved from comparing the performance of our SR and matched portfolio could potentially be an artifact of the time, the models used, or the data set. To test for this possibility, we perform an event study on the companies that enter or exit JSI to investigate the response of the market to SRI. Event study is a statistical method used to investigate the effects of an event on the return of a company. We perform an event study to find out how the inclusion in or exclusion from JSI can affect the return of companies. Based on the efficient market hypothesis, the price of a company at any point in time represent all the information related to that company. This price can only change if new information about the company becomes available to the market. New information can be the result of an event. In this research, the events that may affect the return of JSI members are the inclusion in or exclusion from JSI.

The main step in an event study is to find the difference between the actual return of a company and the return predicted by an asset pricing model. The asset pricing model that we use here is the market model, which states that the return of an equity has a linear relationship with the return of the market.

In any event study, we are faced with three periods: event window, estimation window, and post-event window. Event window is a period that includes the event day and some days before and after it; in this study 20 days before and 20 days after the event day form the event window. Estimation window is the period based on which we find the parameters of the market model; we assume a period of 36 months prior to the start of the event window to be the estimation window. Estimation window and event window should remain separate. Post-event window is the period after the last day of the event window used to investigate the post-event effects of an event on the return of a company.

The difference between the actual return in the event period, and the return estimated by the market model is the abnormal return (NYSE:AR). If the event affects the company in a positive way, the abnormal returns should be positive and significance. If it doesn't have any effect on the company, the abnormal returns should not differ from zero. If the event affects the company in a negative way, the abnormal returns should be negative and significant.

A main difficulty in event studies is due to information leakage, which happens when new information becomes available to some investors prior to the event day. Information leakage causes stock prices to change before the whole market becomes aware of the information. As a result, the abnormal return on the event day may not reflect the whole effect of the event on the return of the company. One way to capture effects of information leakage is to calculate abnormal returns not only for the event day, but also for some days prior to that. In this study, we capture the effects of information leakage by calculating abnormal returns up to 20 days prior to the event day.

In order to draw overall inferences, abnormal returns must be aggregated over time. In other words, we must calculate the cumulative abnormal return (NASDAQ:CAR). CAR is equal to sum of the abnormal returns from the first day of the event window to the day that we want to find the CAR. In order to eliminate idiosyncrasies in measurement, CAR is averaged across companies. The result is cumulative average abnormal return (CAAR). By analyzing the movements and statistical significance of CAAR in the event window, we can investigate how the returns of a group of companies are affected by a particular event.

In Table 2, the results of the event study are shown for 3 days before and after the event day. For the companies included to JSI, CAAR for none of the days is significant. This can suggest that inclusion to JSI is a neutral event, and market response to this event is insignificant. These results may not be in accord with our expectations. Since becoming a member of a SR index seems to be a positive event, we expect to see a positive response from the market. One possible explanation for the observed neutral reaction of the market to these included companies might be that good news leaks much earlier. While companies usually try to hide any upcoming negative event until the latest possible day, they tend to take an opposite approach when it comes to positive events. Contrary to included companies, we can clearly see that the reaction of the market to the excluded companies is significantly negative. As we can see in the table below, CAAR for exclusions on days -1 to +3 are significantly negative, which suggests that exiting JSI is a negative event from the point of view of the investors. 

Table 2: Results of the event study for included and excluded members of JSI

Day CAAR for Included Members CAAR for Excluded Members
-3 -0.386 (-0.37) -1.758 (-1.11)
-2 -0.451 (-0.44) -2.010 (-1.64)
-1 -0.613 (-0.61) -3.193 (-2.16)**
0 -1.066 (-1.02) -3.343 (-2.13)**
+1 -1.705 (-1.37) -3.512 (-2.04)**
+2 -1.609 (-1.16) -3.263 (-1.96)*
+3 -1.043 (-0.75) -3.357 (-1.89)*

Numbers in parenthesis shows the t statistic. *, **, and *** shows significance at 10%, 5%, and 1% level respectively.