Why I Built a Greenwashing Risk Model
Read the full methodology here.
Greenwashing isn't a PR problem. It's a risk management problem.
Over the past decade, sustainability has become impossible to ignore. Companies publish net-zero pledges, tout recycled packaging, and fill earnings calls with ESG buzzwords. Investors get hundreds of pages of sustainability reports. Regulators demand disclosures. In theory, all this transparency should help us figure out who's actually doing the work and who's just talking about it. But in practice, we've built a system that rewards good storytelling over good behavior.
Greenwashing gets written off as a reputational risk: embarrassing when you get caught, but not something you'd model alongside credit default or operational failure. Most ESG frameworks care about how much companies disclose, not whether those disclosures are actually true. A company with a polished sustainability narrative and questionable claims can outscore a quieter competitor making real progress but fewer promises.
That's a problem. Misleading claims don't just damage reputations—they distort capital allocation, weaken accountability, and slow decarbonization by letting firms look climate-aligned without paying the costs of actually getting there.
I think greenwashing deserves better than a footnote in someone's ESG report. So I built a mathematical model to quantify it.
How I got here?
This framework started as my capstone project for the Sustainability Analysis Committee of UC San Diego's Sustainable Investment Group (SIG). I was supposed to build a simple tool to help evaluate greenwashing risk for equity analysis and stock pitches (something practical).
The first version was intentionally straightforward: identify incidents, categorize them, flag patterns. But as soon as we started applying it, the gaps became obvious. How do you distinguish a one-off mistake from systemic misconduct? Should a company that keeps making the same misleading claim be penalized more than one with diverse but isolated incidents? How do you account for severity, evidence quality, and financial exposure without just making subjective calls and calling it analysis?
Those questions, born out of trying to actually use the tool in real investment contexts, pushed the project beyond its original scope. What began as a theoretical exercise evolved into a formal mathematical framework: a quantitative model that assigns numerical risk scores to companies based on documented greenwashing incidents, making it possible to measure, compare, and track greenwashing risk with the same rigor we apply to other material risks.
How the model works: turning incidents into numbers.
At its core, the model converts documented greenwashing incidents into a quantitative risk score through a structured three-stage process.
Stage 1: Scoring individual incidents
Each incident gets scored using a multiplicative formula that combines base values with contextual risk factors:
IncidentScore = T × S × C × R × D × M × I × G × K × A × F
Where the components represent:
Base severity factors (establishing the core magnitude):
T = Greenwashing type weight (1-5 scale): Different types of misleading claims—from vague language to false data to certification misuse—carry different baseline risk levels
S = Real-world severity (1-5 scale): Observable consequences ranging from minor language issues to material legal penalties
Risk amplification multipliers (refining the context):
C = Evidence confidence (0.80-1.00): Source credibility, with tier-based scoring for investigative journalism, regulatory findings, and established NGOs versus lower-quality sources
R = Repeat offense multiplier (1.0-1.6): Escalating penalty for firms making the same type of misleading claim multiple times
D = Temporal decay (exponential): Incidents lose influence over time using a 5-year half-life, balancing accountability with improvement
M = Mitigation factor (0.70-1.00): Credit for substantive corrective action, from public corrections to structural governance changes
I = Industry materiality (1.1-1.5): Sector-specific weighting reflecting how much certain claims matter in different industries
G = Evidence gap (1.0-1.65): Captures both factual contradiction (claims provably false) and opacity (claims unverifiable)
K = Claim scope/revenue proximity (1.0-1.4): Whether claims are abstract corporate statements or embedded in core revenue drivers
A = Amplification factor (1.0-1.5): Distribution channel reach, from ESG reports to mass media advertising
F = Financial exposure (1.0-1.3): Links risk to the share of company revenue affected by the claim
This multiplicative structure ensures that severe, repeated, high-confidence incidents tied to core revenue compound in risk, while isolated, minor, or well-mitigated incidents remain bounded.
Stage 2: Aggregating to company level
Individual incident scores get aggregated into a company-level risk score:
RawGreenwashingRisk = PB × E × (Average of all IncidentScores)
Where:
PB = Pattern bonus (1.0-1.3): Distinguishes isolated incidents from systemic behavioral patterns across different greenwashing types and time periods
E = Sector exposure factor (1.15-1.5): Baseline industry risk reflecting structural differences in sustainability risk and regulatory scrutiny
This is where the model separates companies that made a mistake from those with structural credibility problems. Two firms might have similar incident counts but very different risk profiles once you account for patterns, persistence, and industry context.
Stage 3: Normalization and comparison
The raw company scores get transformed into comparable metrics depending on the use case:
Sector-relative scores (0-100 scale): Normalized within industry using 5th and 95th percentiles for peer comparisons
Cross-sector percentile ranks: Market-wide positioning to identify outliers
Robust z-scores: Standardized using median and median absolute deviation to handle skewed distributions
Time-series tracking: Period-over-period changes, rolling averages, and slope-based trend analysis to monitor whether risk is improving, deteriorating, or persistent
The output is a numerical risk score that can be integrated into equity models, ESG analysis, or screening tools the same way you'd use a credit rating or governance score—but specifically measuring the credibility gap between what companies claim about sustainability and what they actually do.
Why credibility matters more than ambition.
Greenwashing isn't just about bad optics. It's a coordination failure. When misleading claims go undetected, firms investing in real decarbonization end up at a competitive disadvantage. Capital flows toward narrative sophistication instead of operational substance.
Quantifying greenwashing risk helps restore that signal. This model doesn't judge environmental impact directly; it measures whether sustainability claims are credible. That distinction matters. Credibility is the prerequisite that allows you to meaningfully assess ambition, performance, and impact in the first place.
More broadly, this approach aligns sustainability analysis with how other risks are already managed. Credit failures, compliance violations, operational incidents: we track those as events, weight them by severity, adjust for persistence, monitor them over time, and express them as quantitative scores. Applying the same mathematical logic to greenwashing brings ESG analysis closer to established risk management practice without relying on opaque ratings or inflated moral claims.
Building systems that demand more than good stories!!
The goal here isn't to produce a perfect score. It's to make greenwashing risk measurable and comparable: to give investors, analysts, and regulators a quantitative tool that treats credibility gaps with the same analytical rigor we apply to balance sheet risk.
In a climate-constrained world, sustainability narratives need to be more than compelling—they need to be trustworthy. Overcoming greenwashing is a necessary step toward aligning corporate behavior, regulatory oversight, and capital allocation with the scale and urgency of the climate challenge.
This model is one way to start: by treating credibility not as an afterthought, but as a quantifiable risk that can be measured, tracked, and incorporated into the decisions that actually move capital.
