Patience pays off – corporate social responsibility and long-term stock returns

ABSTRACT

This paper presents new evidence on the implications of corporate social responsibility (CSR) on stock returns. By implementing a long-term focus as well as using subdivided measures for CSR, we cater to the intangible nature and the heterogeneity of CSR activities. We use a novel classification of these activities into nine areas, each belonging to one of the standard environment, social, and governance (ESG) dimensions. Using cross-sectional return regressions and buy-and-hold abnormal returns, we find that firms with strong CSR significantly outperform firms with weak CSR in the mid and long run in certain areas. Firm returns increase up to 3.8% with respect to a one-standard-deviation increase of the CSR rating. In a two-stage least squares (2SLS) approach we verify that the main economic channel for the appreciation of strong CSR stocks is unexpected additional cash flows. The results are relevant for assessing the efficiency of CSR, and have broader implications for asset managers who can expect abnormal returns by investing in firms that exhibit a high CSR in the respective scores and holding the stocks for a longer period.

KEYWORDS:

• Asset pricing
• corporate social responsibility
• market efficiency
• buy-and-Hold abnormal return
• earnings surprises

JEL:

• G14
• G30
• G11
• Q56

Disclosure statement

No potential conflict of interest was reported by the authors.

ORCID

Gregor Dorfleitner http://orcid.org/0000-0003-0270-0660

Sebastian Utz http://orcid.org/0000-0003-1570-752X

Maximilian Wimmer http://orcid.org/0000-0002-6283-3169

Notes

1 For instance, reputation enhancement through good corporate citizenship is priced by stock markets efficiently (Brammer, Brooks, and Pavelin 2009).

2 We control the book-to-market portfolios for industry effects, as proposed by Wermers (2004).

3 This contrasts with many papers in financial economics that rely on the benchmark portfolio data for US stocks from Russ Wermer's homepage which rebalance the benchmark portfolios annually.

4 As a consequence, return on assets, which is another classical predictor, is omitted due to a high correlation with profitability.

5 As we have a relatively short panel and a broad cross section, this approach appears to be more suitable than classical panel regressions (Cameron and Trivedi 2010).

6 The standard deviations of the scores used for this sample are 30.91 (E), 28.54 (S), and 16.67 (G), which only deviate marginally from the figures reported in Table 1.

7 Note that we do not instrumentalize the ESG scores. From an investors perspective, we wish to investigate whether ESG scores can be used to forecast a persistent long-term abnormal return pattern, independent of the causal relation between ESG scores and abnormal returns.

8 To further investigate the long-term impact of the VS score, we compute cross-correlations of lagged VS scores with differentiated scores of all other areas. Indeed, lagged VS scores are positively correlated with all area scores in the environment and social dimension, respectively. Therefore, a high VS score is positively related to high E and S area scores in the medium term. These area scores may drive the observation.

9 We downloaded the respective risk factors from Kenneth French's website.

10 As a robustness check, we also computed the calender-time regression alphas with pure returns as endogenous variables. While the statistical interference is lower, the results are consistent with the results from Panel (a).