Skip to content

Board Gender Diversity and Corporate ESG ​

University of Edinburgh · Shareholder Value and ESG · July 2026

Across 2,270 US firms and 18,101 firm-years, greater board gender diversity is robustly associated with higher environmental and social (E&S) performance — but a California-law natural experiment finds no evidence the relationship is causal. The panel uses firm and year fixed effects (statistical controls that strip out anything constant within a firm or a year, isolating the year-to-year relationship) with clustered standard errors to isolate the association, then tests causality with a lead-lag design, propensity-score matching, and California SB 826's board-diversity mandate as a quasi-experiment.

Snapshot ​

2,270
Firms
US Refinitiv-rated
18,101
Firm-years
2012–2024 panel
+2.133***
E coefficient
per 1 SD diversity
+1.08***
S coefficient
per 1 SD diversity

Main result — coefficient forest ​

Firm + year FE · N = 18,101

Effect of lagged board gender diversity (1 SD)

E and S rise 2.133 and 1.08 points per 1 SD of diversity — the real outcomes; the still-larger G coefficient flags mechanical inflation, not signal.

Points = FE coefficients; whiskers ≈ 95% CI (1.96 × SE). Green = primary outcomes (E, S).

2012–2024 panel · Source: Refinitiv

Economic magnitude ​

Scale

Coefficient as % of pillar mean

E moves more relative to its mean than S; G is shown only for measurement context.

2012–2024 panel · Source: Refinitiv

Hypotheses ​

PredictionResult
H1Diversity ↑ → E, S ↑Supported (both ***)
H2G / composite inflated vs E, SSupported (G largest)
H3If causal, SB 826 raises E, SNot supported (diversity ↑; E/S RF n.s.)

Reading: robust conditional association on E and S; diversity is better read as a marker of stakeholder-oriented firms than a stand-alone causal lever. Lead–lag flags reverse causality; SB 826 first stage is strong while reduced form on E/S is insignificant.

Process ​

1
Measure
Primary = E & S (not mechanical G)
2
Build panel
Refinitiv + Compustat + CRSP/CCM
3
Estimate
Firm + year FE, clustered SEs
4
Identify
Lead–lag · PSM · SB 826 DiD

Design note ​

Board gender is an input to the governance pillar. Primary outcomes are E and S only. Composite and G are reported to make mechanical inflation visible — not as the main claim.

Stack ​

Python / Refinitiv Codebook · WRDS Compustat (SAS) · Stata reghdfe · Typst write-up

Competencies ​

ESG measurement · panel FE · quasi-experiments · research integrity · Refinitiv/Compustat


← All projects