PTDoes it pay more to do ESG or to tell what you do?
Companies everywhere invest in environmental, social and governance (ESG) practices. Does it pay off? The thesis reviewed 108 studies and analyzed 2,987 yearly records of companies from Brazil, China, India and South Africa. In these markets, investors rewarded firms that disclose well, not those with the best scores. Explained in 10 screens.
Height = effect of each score on total shareholder return (standardized coefficient): +0.035, not significant, and +0.165, significant. Source: Tables 14 and 15.
Does ESG pay? The studies don't agree
Some studies find gains, others no effect, others losses. To understand why, the author searched Web of Science and Scopus for articles that tested what changes this relationship. Of 610 articles found, 108, published from 2019 to 2023, made it into the analysis.
Circle area proportional to the number of articles. The 108 articles have 320 authors and appeared in 76 journals. Screened with the PRISMA protocol in Rayyan; search closed on 18 Dec 2023. Source: section 2.4.1.
Tap a year. Recent articles have had less time to be cited; the author also reads the drop as a sign of a field without consensus. Source: Table 2 and section 2.5.1.
The answer depends on who runs the company
Across the 108 studies, the factor most often tested as a moderator, something that can strengthen or weaken the effect of ESG on results, was the board of directors, followed by the characteristics of the firm itself. Tap a bar:
Of the 20 gaps the review points out, four guided the second study: more financial measures (including shareholder return), separating E, S and G, combining the "do" and "tell" scores and more robust methods with global data. Source: Table 8 and section 2.9.
Height = % of moderator occurrences in the 108 studies (the six most frequent). Source: Figure 9 and section 2.5.4.
% of studies by type of measure. Risk (volatility, credit risk) hardly appears. Source: Figure 10 and section 2.6.1.
Two scores for the same company: what it does and what it tells
Bloomberg, a database used by investors, gives every company two ESG scores. The author used both, overall and by pillar: environmental (E), social (S) and governance (G).
ESG performance
What the company delivers on each pillar, weighted by what matters financially. Scale of 0 to 10; sample mean: 2.77.
ESG disclosure
How much ESG data the company makes public. Scale of 0 to 100; sample mean: 45.8.
Means by pillar. In both scores, governance is the highest pillar. Source: Tables 11 and 12.
And financial results? The thesis uses four measures: two look at the market, two at the accounts. Tap to flip:
Sample means before standardization. Source: Tables 11 and 12.
2,987 yearly company records in four countries
Listed BRICS companies with ESG scores and financial data on Bloomberg, from 2016 to 2023. Each company in each year is one record. Russia was left out: after the sanctions that followed the invasion of Ukraine, its data were no longer available.
Records by sector. China accounts for two thirds of the sample (1,967 of 2,987). Source: Tables 9 and 10 and section 3.4.
More profitable companies may invest more in ESG, not the other way round. The two-stage method (2SLS) isolates the part of the score that does not reflect the result itself and compares companies in the same sector and year.
Standardized variables: the next screens show how much the result changes, in standard deviations, when the score rises by one standard deviation.
Doing, on its own, doesn't move the needle. Telling does.
With the overall performance score, no measure of return or profit improved. With the overall disclosure score, shareholder return, ROA and ROE went up. For the author, in markets with weak rules and little information, disclosure works as a signal to investors. Compare:
Standardized 2SLS coefficients: how much the measure changes, in standard deviations, when the score rises by one standard deviation (about 1.2 points out of 10 for performance; about 12 points out of 100 for disclosure). *** p < 0.01; ** p < 0.05. Source: Tables 14 and 15 (models 1.1 to 1.4 and 2.1 to 2.4).
Governance is the pillar that pays most consistently
Splitting the scores into environmental, social and governance, governance disclosure is the only one that improves all four measures, and governance never pulls results down. Social makes no difference in performance; in disclosure, it helps shareholder return but lowers market value. Explore:
View as a table
Height = size of the coefficient; color = direction. Tap a block. Source: Tables 14, 15, 17 and 18.
Disclosure helps up to a point
The relationship is not a straight line. For shareholder return, ROA and ROE, the disclosure curve rises, peaks and falls: disclosing too much stops paying. Market value does the opposite: it penalizes firms stuck in the middle. The author calls this the "too much of a good thing" effect. Move the score:
Curves recalculated by us from the coefficients in Table 16, with other variables at their mean; axis from −2 to +2 standard deviations. Scores out of 100 (or 10) approximated from the mean and standard deviation in Table 11. Dashed line: significant only at 10%. Source: Table 16 and Figures 19 and 20.
The gray zone: stopping halfway is the worst place to be
When the author put both scores in the same model, a gray zone appeared: with low-to-middle disclosure, predicted shareholder return, ROA and ROE fall below the average company. At high disclosure, the best shareholder return and ROE come with average performance, not maximum. Place a company:
Curves recalculated by us from Table 19 (models 6.5 to 6.8), with other variables at their mean; they reproduce Figure 21 of the thesis. Low, average and high performance = −1, 0 and +1 standard deviation. In this joint model, the disclosure curve has a different shape than on screen 08, where it appears alone: read the curves as tendencies, not fixed rules.
Disclose coherently, and don't stop halfway
For managers
Disclose ESG in an integrated way, covering all three pillars, and in line with what the company actually does. Disclosure without substance invites accusations of greenwashing. Avoid the lukewarm engagement of the gray zone.
Section 3.6.3For investors
Look at shareholder return together with ROA and ROE: the market may reward the narrative before profit shows up. Governance is the pillar that creates value most consistently.
Section 3.6.3 and chapter 4For regulators
Standardize ESG reports, require independent verification and curb purely symbolic disclosure, so that what is told reflects what is done.
Section 3.6.3What this research does not say
- Only listed companies, and no Russia. Private and mid-sized firms were left out. · according to the author
- A single source of scores. Bloomberg scores may miss qualitative aspects; comparing with MSCI or Refinitiv would help. · according to the author
- The pandemic was not isolated, and the review covered 5 years and two databases. · according to the author
- Two thirds of the records are from China, and the effects are small (tenths of a standard deviation). 2SLS reduces, but does not remove, doubts about cause. · our note
Credits
CARDILLO, Marcos Alexandre dos Reis. ESG/CSR measures impacting financial performance: an analysis of moderating variables. Doctoral thesis (Business Administration) — Mackenzie Presbyterian University, São Paulo, 2025. Advisor: Prof. Dr. Leonardo Fernando Cruz Basso.
Article 1 published: Cardillo & Basso (2025), Journal of Innovation & Knowledge, 10(1), 100648. doi.org/10.1016/j.jik.2024.100648
Thesis written in English, supported by CAPES (Brazil's federal agency for graduate education). Data from Bloomberg and the World Bank; 2SLS analysis in Stata. This version simplifies the work for a general audience.