Causal Lessons From the Field: What Really Moves Markets
From carbon to diversity to regulation — what holds up under causal inference and what doesn’t.
Markets are full of bold claims about sustainability. Carbon emissions, water use, gender diversity, regulatory alignment — all are said to drive financial performance. But how often do these stories hold up when we put them under causal scrutiny?
Over the past few weeks and months, I’ve run a series of case studies using real company data, testing whether these sustainability factors truly move share prices and profits. Some effects held up with remarkable clarity; others collapsed the moment we stress-tested them. Both outcomes matter: proof gives investors conviction, while refutation saves them from costly mirages.
This piece pulls together those lessons, offering a birds-eye view of what we’ve found so far.
Carbon Emissions
Carbon is the most obvious sustainability signal in markets — and also the most overanalyzed. Banks run climate-risk stress tests, asset managers build carbon-hedging strategies, academics publish paper after paper. Yet almost none of them prove causality.
When we looked at ArcelorMittal’s carbon footprint, the raw correlation with share price was barely noticeable: –0.076. Easy to dismiss. But correlation isn’t causation. So we ran a causal model using DoWhy, with confounders like steel production volume and workforce size, and a mediator for market volatility.
The result surprised even me: higher carbon emissions do causally depress ArcelorMittal’s stock price. Each one percent cut in emissions translated into roughly a half-percent boost in share price — a material effect for a company of this size.
Key lesson: causality matters. Correlations alone obscure the signal, but causal inference turns faint noise into investable evidence.
Water Stress (Groundwater Extraction)
Not every hypothesis survives the causal gauntlet. Groundwater looked promising: years with higher reliance on groundwater extraction correlated with lower profits at ArcelorMittal (–0.26). From a systems view, it made sense — over-extraction can signal inefficiency, environmental stress, or looming regulatory pressure.
But when we tested it causally, the relationship collapsed. Placebo tests outperformed the original model, meaning our strong-looking effect was likely spurious.
That non-result is valuable. It tells us that while water risks may be real, they don’t yet show up cleanly in this company’s profit data. It also saves investors from chasing false positives — those correlations that look enticing in a scatterplot but vanish under proper scrutiny.
Key lesson: even plausible stories fail under causal testing. Refuting them is just as useful as confirming them, because it clears the ground for stronger signals.
Diversity
Diversity is often framed in moral terms, but markets care about stability. In our test on ArcelorMittal’s management composition, we asked: does adding more women to management roles lift the stock price?
The raw correlation was small (~0.18) — nothing you’d trade on. But when modeled causally, the effect was significant. A +1 percentage point increase in women in management corresponded to roughly a +€2.5 to +€3.5 rise in share price. At current trading levels, a 10 percentage point shift could move the stock by ~€25–€35 — essentially doubling its value.
The mechanism matters. Diverse leadership doesn’t just add new perspectives; it reduces volatility. Teams with varied experience and temperament make fewer brittle decisions, and that stability shows up in valuation.
Key lesson: diversity isn’t just PR. It’s a structural stabilizer, and markets pay for stability.
Regulation (EU Taxonomy)
ESG ratings have always been noisy, inconsistent, and prone to greenwashing. But the EU Taxonomy changed the game by tying “sustainability” to hard numbers: revenues from specific, legally defined activities.
Empirical evidence shows investors are already rewarding firms with higher shares of Taxonomy-aligned revenues — even before full regulation took effect. One study found that a one–standard deviation increase in alignment translated into ~30 basis points of extra monthly returns. On the day the Taxonomy was officially published, aligned firms saw +0.66% abnormal returns, while the least aligned lost –0.67%.
In other words: the market reaction looked like an earnings surprise. That’s how powerful verifiable definitions are.
Key lesson: markets don’t reward vague ESG scores. They reward measurable, regulated sustainability signals — and they’ve already started doing so.
The Narrative Premium
Data by itself rarely moves capital. A scatterplot showing “X correlates with Y” is interesting, but it doesn’t trigger portfolio reallocation. What investors remember, repeat, and act on is the story that frames those numbers.
This is the essence of the narrative premium: the extra lift a company earns not from raw statistics but from the meaning attached to them. A causal study showing women in management improve share price is one thing; the story that “diverse teams make calmer, better decisions under pressure” is what spreads across boardrooms and drives capital flows.
Narratives also cut through complexity. ESG disclosures, transition risks, and regulatory shifts are overwhelming on their own. But frame them as “this company is a transition leader” or “this sector risks becoming stranded,” and suddenly investors can act.
The risk, of course, is empty storytelling — hype that detaches from evidence. The dot-com bubble, carbon offset fads, and glossy ESG reports all show how dangerous untethered narratives can be. The opportunity, however, is clear: when grounded in causal evidence, stories can transform analysis into conviction.
Key lesson: numbers open the door, but stories make people walk through it. The most effective analysts will weave causal rigor with narrative clarity — turning statistical insight into market-moving conviction.
The Bottom Line: It Remains Complicated
The bigger picture is simple: not every sustainability metric translates into financial value. Some, like carbon emissions or taxonomy alignment, show causal bite. Others, like groundwater extraction in our first trial, fail to validate. That’s not failure; it’s progress.
The real lesson is twofold. First, causality matters: correlations alone are too noisy to guide capital. Second, narrative matters: evidence without a story won’t move markets. Analysts who can combine the rigor of causal methods with the clarity of narrative framing will be the ones who shape tomorrow’s capital flows.
In other words, sustainability doesn’t just need good data. It needs good stories, grounded in proof.