Every Little Helps? ESG News and Stock Market Reaction

The Coming Technological Singularity, 1993.

The academic literature on the impact of ESG news is larger, but most of the papers assess the market reaction following a specific event, based on small hand-collected samples without much details. Consequently, they provide little evidence concerning the determinants of the reaction or comparison between different events.

Recently, Elayan et al. (2014) have also used data from Covalence EthicalQuote, but they only consider aggregated information. However, their results suggest that these pieces of information convey useful information to shareholders.

Empirical studies on the relationship between CSP (corporate social performance) and CFP (corporate financial performance) are of three types. A first piece of literature examines whether portfolios composed of firms with a high level of CSP outperform the market (e.g., Barnett and Salomon 2006; Capelle-Blancard and Monjon 2014). A second strand considers the long-term relationship between CSP and accounting-based measures of CFP (e.g., King and Lenox 2001; Konar and Cohen 2001; Guenster et al. 2011). In this paper, we focus on the third type, namely event studies (see “Hypotheses Development” section).

Extensive media coverage might also be explained by the fact that ESG news are “blue-compatible”. Empirical evidence suggests that Democrats (the “blues”), in contrast to Republicans (“the reds”), are more apt to support causes such as environmental and labor protection, and they are more supportive of the stakeholder theory. For instance, Hong and Kostovetsky (2012) find that mutual fund managers who make campaign donations to Democrats hold less of their portfolios in industries that are deemed socially irresponsible (based on KLD ratings). Moreover, news media seems to be biased toward liberal ideas; see Groseclose and Milyo (2005) for empirical evidence from the USA. All together, this is consistent with the heavy weight news media give to ESG issues.

Interestingly, Zyglidopoulos et al. (2012) find that more media attention leads to an increase in CSR strengths, but does not drive any significant change in CSR weaknesses. Yet, this is still consistent with our hypothesis suggesting that it is more expensive for firms to improve their weaknesses, than to enhance their strengths.

For a formal model of greenwashing in presence of “soft” and “hard” (verifiable) information, see Bazillier and Vauday (2009).

Alternative specifications (with or without the sectoral index or with the CAPM) lead to very similar results. The most serious concern is about confounding events. While there is an extensive literature on the event study methodology, the contamination of the estimation period (Aktas et al. 2007) or the event period (Nelson et al. 2008) has attracted little attention. The standard approach consists in using a case-by-case analysis to neutralize all the confounding events, but this approach is impractical for large samples. In our case, we know precisely the dates of the possible confounding events (not all of them, but the ones related to ESG news). Then, we provide a robustness check in order to tackle the problem of confounding events. We consider another estimation window (−130, −11) where confounding events included in the Covalence EthicalQuote database are neutralized. In other words, for each firm, the days surrounding ESG news (−1, 0, +1) are dropped from the estimation window. The results are qualitatively the same (see in “Appendix” section). But the contamination by non-ESG events of the estimation period or the event period remains a possible limitation of this study.

See “Appendix” section for some examples of ESG news collected by Covalence EthicalQuote.

These lists have been manually check. The complete list (with more than 1000 sources) is available on request.

For example, the percentage of bad ESG news published by the media on environmental issues is equal to 7% for the banking industry and to 35% for the chemical industry. Conversely, the percentage of bad ESG news published by the media on corporate governance issues is equal to 24% for the banking industry and to 11% for the chemical industry (see Capelle-Blancard and Petit 2017).

An alternative would be to use the Volatility Index (VIX) computed by the CBOE and based on the implied volatility of the S&P 500 index options. Often referred to as the fear index, it shows a very similar pattern.

We also used the Fortune Magazine’s annual ranking of the “world’s most admired companies.” A problem, however, is that it does not cover all the firms included in our sample. Moreover, according to McGuire et al. (1988), Fortune Magazine’s ranking is linked to prior financial performance.

We also used a 6-month and a 2-year time frame for robustness tests, but it does not change the results.

We only consider news in English.

R&D spendings (divided the total assets) and percentage of floating shares have been used also, but the availability of the data (moreover, only annual) lowers the size of the sample without changing meaningfully the results.

Note that this first set of results holds when we consider alternative specifications: with a restricted sample that excludes events for which abnormal returns are likely to be misspecified (i.e., when parameters of the market model are jointly not significant), with or without sectoral indexes, with a longer estimation window cleaned of contaminating events, or with the Fama-French 4 factors. Those results are available on request.

Table 10 in “Appendix” section only considers cumulative abnormal returns significant at the 1% level. Table 11 uses a larger estimation window \([-130;-11]\) which is “decontaminated” (the days when a firm experiments ESG events were removed from the estimation windows).

Similarly, we also tested whether upcoming weekend (the so-called Friday effect) or holidays distract shareholders from ESG issues (DellaVigna and Pollet 2009), but we did not find any impact as well. We have also controlled for the total amount of news reported in the Dow Jones Factiva database (therefore, including non-ESG news), which can be considered as a proxy of limited attention of shareholders (Hirshleifer et al. 2009); the results were not significant.

See Coval and Moskowitz (2001), Grinblatt and Keloharju (2001), Huberman (2001) or Ivkovic and Weisbenner (2005).

In a previous version, we tested whether the impact of ESG news depends on the country where the event took place ; in particular, we distinguished developed and developing countries. On the one hand, the impact should be higher for negative events if they occur in developed countries compared to developing ones, because of environmental liability rules or the stringency of regulation. On the other hand, due to the difference of media coverage, small events are more likely to be reported in developed countries. We did not observe any significant differences.