Nowcasting Green Investment Sentiment from Google Trends: An Environmental Economic Analysis Using Machine Learning

Abstract

This study investigates global public sentiment toward green investment by analyzing five years (2020–2024) of Google Trends data using machine learning. A Green Investment Sentiment Index (GISI) was constructed from five key search terms: green finance, green investment, climate finance, ESG investing, and sustainable investing. Results show that green finance consistently dominated search interest, with a mean index score of 38.5 and a maximum of 100. GISI rose significantly from 35 in 2020 to 68 in 2024—an increase of 94%—with major spikes during COP26 (2021) and COP28 (2023). The LSTM model outperformed XGBoost and Random Forest in forecasting sentiment patterns, achieving the lowest RMSE (6.8), MAE (5.0), and MAPE (9.8%), with the highest R² (0.93), indicating strong predictive performance. The model captured both seasonal peaks (Q4 increases of 20–25%) and event-driven surges (+42% in climate finance during COP26). This research contributes a novel real-time approach to environmental sentiment monitoring, offering scalable tools for aligning policy interventions with public awareness. It provides actionable insights for policymakers, ESG investors, and sustainability communicators. Future research should expand by integrating social media and news data and adopting interpretable AI for transparent forecasting.