Regulatory Signals Move Green vs. Brown Stock Markets
This paper by Victor DeMiguel, Javier Gil-Bazo, and Milind Goel investigates whether climate-related tweets by U.S. Congress members contain material information that affects the relative valuation of green (low-emission) versus brown (high-emission) stocks, with a focus on the regulatory channel of green transition policy.
Data & identification strategy:
The authors employ a high-frequency event study around legislator tweets to isolate causal effects. They downloaded 2.8 million tweets from 489 U.S. Congress members (2014–2023) and identified 38,055 climate-related tweets using FinBERT. Using ChatGPT-3.5, they measured each tweet’s sentiment toward the green transition on a -100 to +100 scale. For the 1,000 largest U.S. stocks, they obtained minute-level prices and classified stocks daily into green, gray, or brown terciles based on carbon emissions intensity (Scope 1 + 2 emissions per revenue).
Event window & regression:
The core analysis examines stock returns in the one-to-ten-minute window around each tweet, from 1 minute before to 10 minutes after. This tight window minimizes confounding by other news. They estimated a panel regression where stock returns are regressed on tweet sentiment and interactions with green/brown dummies, controlling for firm characteristics, risk factors, and fixed effects (firm, legislator, time).
Regulatory channel test:
To verify the mechanism, they identified 40,821 tweets mentioning specific environmental regulations and analyzed how the green-brown return spread evolves relative to congressional votes—increasing in the weeks before votes and declining after—as well as varying with vote closeness and congressional control.
Findings
- Pro-transition tweets trigger immediate green outperformance
Within 1–10 minutes of a pro-transition tweet, green stocks significantly outperform brown stocks by 0.154 basis points per standard deviation of tweet sentiment. Cumulative returns over the sample period show a 6.9% spread between pro- and anti-transition tweet days. - Brown firms bear the regulatory burden
Pro-transition tweets reduce all firms’ value, but brown firms experience the largest declines because they face higher compliance costs. Green firms are insulated by comparison. - Markets identify adaptive brown firms
High-emission firms with stronger positive returns around pro-transition tweets subsequently reduce their carbon intensity more, showing that markets correctly reward firms capable of transitioning. - Effects strengthen over time
The return spread grows from 0.154 bps (10-minute window) to 1.533 bps at the one-day horizon, indicating prices fully incorporate tweet information within one day. - Regulation is the primary channel
The green-brown spread increases in weeks leading to congressional votes and vanishes afterward. The effect is larger before close votes and when Congress is split, confirming markets respond to regulatory probability, not just climate sentiment.
Legislator Tweets About the Green Transition and the Returns of Green versus Brown Stocks
Authors: Victor DeMiguel, Javier Gil-Bazo, Milind Goel
From: London Business School, Universitat Pompeu Fabra
The cost of waiting: How climate policy uncertainty freezes green investment
This paper by Piero Basaglia, Clara Berestycki, Stefano Carattini, Antoine Dechezleprêtre, and Tobias Kruse, examines how uncertainty about climate policies, distinct from actual policy changes, impacts firm investment, innovation, and stock market outcomes. The authors construct a novel Climate Policy Uncertainty (CPU) index using newspaper-based keyword searches from three major U.S. newspapers (New York Times, Wall Street Journal, Washington Post) from 1990–2019. The CPU index identifies articles containing terms related to: (1) climate (e.g., “CO2,” “global warming”), (2) policy (e.g., “regulation,” “tax”), and (3) uncertainty (e.g., “uncertain”). Sub-indices (CPU+ and CPU-) distinguish between uncertainty pointing to policy strengthening vs. weakening.
Using firm-level data from Compustat, patents, and stock prices, the authors interact the CPU index with sector-level CO2 intensity (from EPA Environmental Extended Input-Output tables) as a proxy for exposure to climate policy. The key assumption is that firms in more polluting sectors are more exposed to climate policy uncertainty. The specification includes firm fixed effects and sector-year fixed effects to control for sector-specific shocks.
Findings
Climate policy uncertainty reduces firm investment and employment
A 20-point increase in CPU (average year-over-year change) reduces in industrial sectors (service firms show no response):
- R&D spending by -7.3% for mineral manufacturing, -3.5% for chemicals
- Capital expenditure by -0.7% to -1.2% depending on sector
- Employment by -0.3% to -0.9% depending on sector.
Innovation is particularly vulnerable
Higher CPU reduces clean patent filings by 2.6% and total patents by 2.4% in highly exposed sectors. Dirty patents are unaffected, indicating climate policy uncertainty doesn’t reduce investment in carbon-intensive innovation.
Stock market reacts to policy uncertainty
Higher CPU increases stock volatility and reduces stock returns for exposed firms. Effects are small but statistically significant: a 31-point month-over-month CPU increase raises volatility by ~0.06% for median CO2-intensity firms.
Direction of uncertainty matters asymmetrically
This suggests firms hesitate to invest when policy could roll back, but don’t aggressively capitalize on expected tightening:
- CPU- (uncertainty about policy weakening) drives most results: reduces R&D by ~2× more than CPU+
- CPU+ (uncertainty about strengthening) has smaller, inconsistent effects
Effect is distinct from economic policy uncertainty and actual policy changes
- Results hold after controlling for the Baker et al. (2016) Economic Policy Uncertainty index.
- CPU’s negative impact persists after controlling for climate policy salience and stringency.
Climate Policy Uncertainty and Firms’ and Investors’ Behavior
Authors: Piero Basaglia, Clara Berestycki, Stefano Carattini, Antoine Dechezleprêtre, Tobias Kruse
From: Bordeaux School of Economics, Columbia University, Georgia State University, OECD
