Early-Warning Signals of Reform: How Political Value Priorities Predict Energy Policy Change

Do shifts in political value priorities predict energy policy change? Focusing on Sweden’s energy sector—a cornerstone of the green transition—we examine how evolving political beliefs and problem framings shape policy stability and reform over time. We assembled and analyzed a new, longitudinal corpus of political texts spanning 1990–2024, combining parliamentary debates, party programs, and formal policy consultations. Texts were sentence-tokenized and lemmatized; an energy-specific lexicon seeded and expanded feature extraction. We integrated natural language processing and large language models (LLMs) to map actors, belief-based advocacy coalitions, and their dynamics. Beliefs were operationalized via problem definitions, policy goals, and preferred instruments. Predictive modeling linked these features to subsequent policy changes using time-to-event analyses. Discernible shifts in value priorities, particularly the reweighting of environmental stewardship, energy security, and industrial competitiveness, emerge as leading indicators of major energy policy changes. Periods of intra-coalition coherence combined with cross-coalition convergence are also consistently followed by legislative or regulatory change. The study advances theories of policy change by demonstrating that beliefs, measured as values, frames, and instrument preferences, are predictive, not merely explanatory. Substantively, it provides actionable insights to anticipate and guide energy policy reforms critical for the green transition; methodologically, it shows how text-as-data and LLMs can identify early signals of reform from routine political communication.