Project 4: Measuring and modelling climate policy uncertainty
Work Package 1:
WP1 develops novel news-based indices of climate policy uncertainty for Europe using advanced NLP techniques such as BERT and sentence embedding methods. We construct a European Climate Policy Uncertainty (ECPU) index based on 19 million newspaper articles from 2005–2020, covering all EU ETS phases. A new Google Universal Sentence Encoder Dissimilarity Index captures disagreement and sentiment variation across texts. The ECPU index is then validated through econometric models linking it to CO₂ price volatility in the EU ETS and to stock market volatility and variance risk premia via GARCH-MIDAS-X models. This enables an empirical assessment of how policy uncertainty influences long-term volatility and risk pricing.
A NPL augmented measure of climate policy uncertainty and its empirical impact on realized volatility of asset returns
Work Package 2:
WP2 extends production-based asset pricing models to incorporate regulatory uncertainty and transition risk in the move to a carbon-neutral economy. We model taxes, subsidies, emission caps, and stochastic policy decisions as sources of uncertainty affecting “green” and “brown” sectors. The framework builds on Pástor and Veronesi (2012) and introduces political costs as drivers of random policy changes. Using recursive preferences and equilibrium modeling, we analyze how transition policies influence productivity, consumption, and risk premia. Numerical methods such as Chebyshev polynomials and neural networks are employed to solve the system of nonlinear equations, linking political uncertainty to asset prices.
Extension of asset pricing models to include climate policy uncertainty
Work Package 3:
WP3 studies how financial intermediaries—particularly mutual funds—mediate the effects of climate policy uncertainty. We hypothesize that delegated investment in green assets has a non-linear, hump-shaped relationship with both uncertainty and investor participation. Using U.S. mutual fund data from the SEC’s EDGAR database and CRSP, we quantify direct versus delegated green investment and combine it with ESG ratings from Bloomberg and MSCI. The econometric analysis employs panel regressions and nonlinear models to test how climate policy uncertainty affects green investment through intermediation. This work identifies whether intermediaries mitigate or amplify regulatory uncertainty and how they influence the allocation of capital in the transition to a low-carbon economy.
Work Package 4:
Financial intermediation as a channel to mitigate climate policy uncertainty
Empirical asset pricing consequences of climate policy uncertainty
WP4 integrates the new ECPU index (WP1), theoretical pricing implications (WP2), and intermediation data (WP3) to test whether climate policy uncertainty is a priced risk factor. Using Fama–MacBeth regressions for European and U.S. markets, we estimate the sensitivity of asset returns to climate policy uncertainty, controlling for standard risk factors (e.g., Fama–French, Carhart, and climate risk indices). A second set of tests explores the interaction between direct and delegated investment—examining whether assets held mainly by households respond differently to CPU than those held by intermediaries. These tests assess how uncertainty affects cross-sectional returns, revealing its implications for financial stability and transition finance.
This project investigates how climate policy uncertainty (CPU) affects financial markets, asset pricing, and the behavior of financial intermediaries. While carbon pricing mechanisms such as the EU Emissions Trading System (EU ETS) are central to the transition to a carbon-neutral economy, the uncertainty about future regulatory measures, their timing, and political feasibility remains substantial. We aim to measure, model, and price this uncertainty.
The project combines novel machine-learning and text-mining methods with theoretical and empirical asset pricing approaches. We develop new natural language processing (NLP)-based indices of CPU from European news data, link them to asset price volatility, and integrate them into dynamic equilibrium models with transition risk. Further, we study how financial intermediaries channel, mitigate, or amplify the effects of CPU and examine how uncertainty is priced across assets.
The research is structured into four Work Packages:
