Ecological costs generate a hidden equity return premium

Citizen science data shows facility openings reduce local bird abundance by 6% to 9%

The study constructs a Biodiversity Pressure Index by matching 30 years of eBird citizen science observations to the precise locations of every facility reporting to the US Toxics Release Inventory. Bird counts function as a measurable proxy for broader local ecosystem health and higher index values capture a larger ecological footprint relative to firm assets. The pressure index measures the ecological context of industrial activity, distinct from the firm’s actual volume of toxic emissions.

A staggered difference-in-differences design comparing inner rings around new facilities to outer control rings shows that bird abundance falls by 6% to 9% within 17 kilometres of a new opening, with effects monotonically attenuating beyond 20 kilometres. Species richness also declines by 3% to 5%. Within-facility regressions confirm a dose-response relationship, where more chemical releases in a given year reduce local abundance further. The ecological signal is real, localised and sensitive to operating intensity.

Investors mistake stable production costs for operational decline

Operating in ecologically rich areas carries a persistent cost. Firms with high biodiversity pressure report cost of goods sold 40 basis points higher as a share of sales, and the gap persists eight quarters out. Sales growth declines modestly. The channel runs through elevated production costs, while consumer demand remains largely unaffected.

These costs are persistent and location-bound. Investors observe the depressed margins but misattribute them to deteriorating operational efficiency, setting earnings expectations too low. When quarterly results arrive and show that costs have not in fact deteriorated, prices correct upward. The premium also stands apart from the cost of pollution itself. In head-to-head tests, the ecological footprint subsumes the predictive power of raw chemical releases.

40% of the small-firm premium materialises around earnings announcements

Decomposing the long-short return shows that 40% of the small-firm premium accrues in the 3 trading days surrounding quarterly earnings announcements. Firms with the largest ecological footprints systematically beat their own seasonal earnings benchmark, while analyst long-term growth forecasts do not adjust at all.

The pattern reproduces the classic forecast-error fingerprint identified by La Porta (1997) for value stocks, where pessimistic expectations are corrected at earnings releases, with a slow drift to follow. Among small-cap firms, prices continue to drift upward for a further 60 trading days, reflecting how slowly markets absorb the news that a recurring cost was not a fundamental problem.

ESG raters assign the lowest exposure scores to firms with the largest ecological footprints

The premium persists because the information infrastructure for biodiversity does not exist. Of more than 30,000 corporate 10-K filings examined, only 1.8% mention biodiversity, and firms with worse exposure are no more likely to disclose. MSCI biodiversity scores invert the underlying signal, assigning the lowest exposure ratings to firms with the largest ecological footprints. Because ecological sensitivity depends on spatial geography rather than a standardised metric like carbon equivalents, it falls outside the simplified models used by financial analysts and ESG-raters alike.

When EPA enforcement reveals previously hidden liabilities through a monetary penalty, high-BPI firms underperform their low-footprint peers by 7.78% over the following 60 trading days. The equity-value loss equates to roughly 25 million dollars for the median firm. Enforcement is the channel through which the market eventually impounds information that disclosure infrastructure leaves on the table.

Atreya Dey, Research Associate (Finance) at Cambridge Centre for Finance (CCFin) and Cambridge Endowment for Research in Finance (CERF), says: “Carbon emissions have a standardised metric, an analyst forecasting infrastructure and a steady flow of disclosure events. Biodiversity has none of these. Without that infrastructure, ecological costs do not enter analyst forecasts, and the forecast errors that follow show up in returns.”

Methodology

The analysis covers 1,873 US-listed rms and roughly 235,000 rm-months between May 1989 and December 2024. Bird abundance and species richness are constructed from eBird citizen science observations matched to TRI facility coordinates. Ecological causal estimates use the Callaway and Sant’Anna staggered difference-in-differences estimator with facility-clustered standard errors.