Probabilistic Forecasting of Climate Policy Uncertainty: The Role of Macro-financial Variables and Google Search Data

TL;DR

This study addresses forecasting Climate Policy Uncertainty ($CPU$) by integrating a broad set of macro-financial predictors with public attention data from Google Trends within a Bayesian Structural Time Series ($BSTS$) framework. It identifies a parsimonious, economically grounded predictor set via four screening methods and a spike-and-slab prior, and evaluates probabilistic forecasts for US and Global CPU across 3-, 6-, 12-, and 24-month horizons. Impulse-response analysis via local projections reveals clear channels through housing, credit, labor, and sentiment that drive $CPU$, with housing and financial conditions generally reducing uncertainty while leverage and overvaluation amplify it. The results show that BSTS with time-invariant coefficients and included Google Trends signals delivers superior long-horizon accuracy, offering actionable guidance for adaptive climate policy and risk-aware decision-making under economic cycles.

Abstract

Accurately forecasting Climate Policy Uncertainty (CPU) is essential for designing climate strategies that balance economic growth with environmental objectives. Elevated CPU levels can delay regulatory implementation, hinder investment in green technologies, and amplify public resistance to policy reforms, particularly during periods of economic stress. Despite the growing literature documenting the economic relevance of CPU, forecasting its evolution and understanding the role of macro-financial drivers in shaping its fluctuations have not been explored. This study addresses this gap by presenting the first effort to forecast CPU and identify its key drivers. We employ various statistical tools to identify macro-financial exogenous drivers, alongside Google search data to capture early public attention to climate policy. Local projection impulse response analysis quantifies the dynamic effects of these variables, revealing that household financial vulnerability, housing market activity, business confidence, credit conditions, and financial market sentiment exert the most substantial impacts. These predictors are incorporated into a Bayesian Structural Time Series (BSTS) framework to produce probabilistic forecasts for both US and Global CPU indices. Extensive experiments and statistical validation demonstrate that BSTS with time-invariant regression coefficients achieves superior forecasting performance. We demonstrate that this performance stems from its variable selection mechanism, which identifies exogenous predictors that are empirically significant and theoretically grounded, as confirmed by the feature importance analysis. From a policy perspective, the findings underscore the importance of adaptive climate policies that remain effective across shifting economic conditions while supporting long-term environmental and growth objectives.

Figures (8)

• Figure 1: Global and US interest in the Google search terms (a) "Environmental Policy" and (b) "Carbon Credits" from January 2004 to June 2021. The series display similar temporal patterns, highlighting the close alignment between US and worldwide search behavior. Data sourced from Google Trends: https://trends.google.com/trends?geo=US&hl=en-GB.
• Figure 2: Interest in the worldwide search terms "Climate Policy" and "Climate Risk" from January 2004 to June 2021, plotted alongside the primary US CPU index. The three series exhibit similar temporal patterns.
• Figure 3: Multiple comparisons with the best (MCB) plot comparing model performance across the two covariate sets and the three CPU indices. BSTS-X$_\text{MG}$-2.50, for instance, signifies that the average ranking of the BSTS-X$_\text{MG}$ model, according to the RMSE metric, is 2.50. This same interpretation holds for other models.
• Figure 4: Murphy diagrams of BSTS-X$_\text{MG}$ with baselines (BSTS-X$_\text{M}$ (top) and ARIMA-X$_\text{MG}$ (bottom)) for the 24-month-ahead forecasts for the (a) primary US, (b) alternative US, and (c) Global CPU indices. The parameter $\theta$ represents the shape parameter as defined in Eqn. \ref{['Eq:Theta']}. Lower scores indicate better performance.
• Figure 5: Impulse response functions generated via the local projections method, quantifying the CPU index's dynamic reactions to one-standard-deviation innovations in macroeconomic and financial cycle variables. The reaction is represented by the solid black line, and the 95% confidence intervals are represented by the gray shaded regions. The dashed black line displays the zero line. The sample period runs from April 1987 to June 2021.

Theorems & Definitions (1)

• Definition 1