Tracking the public perception of heat pumps: A sentiment analysis of German news articles

Heat pumps are a key technology for decarbonizing residential heating, yet their diffusion remains limited in many countries, partly due to challenges of public acceptance and legitimacy. This study tracks how heat pumps are portrayed in German public discourse by analyzing sentiment in online news coverage. We compile a corpus of 33,131 German heat pump–related news articles published between 2018 and 2023, drawing on the German CommonCrawl news dataset. Relevant articles are retrieved using semantic search in a vector database with pretrained text embeddings and refined using a cross-encoder reranker. We then measure article-level sentiment with a pretrained Natural Language Inference (NLI)–based zero-shot sentiment classifier and identify dominant themes using BERTopic. Finally, we link topics and sentiment using regression models that control for article length and readability as well as media outlet and time fixed effects. Results show that heat pump articles are, on average, framed more positively than the broader news landscape, but sentiment varies substantially by topic. Coverage related to heat pump adoption and district heating planning is associated with comparatively positive sentiment, whereas articles on rising heating costs, grid strain management, heat pump thefts, and especially the debate around Germany’s Buildings Energy Act (GEG) exhibit more negative sentiment. Longitudinally, we observe a sharp decline in sentiment in early 2023 during the heating law controversy, followed by a gradual recovery thereafter. Overall, the findings highlight the thematic heterogeneity and volatility of acceptance-relevant discourse around clean-heating technologies and demonstrate how scalable text-as-data methods can monitor shifts in technology legitimacy over time.