Forest Lands: Evaluating Forest Managers’ Decision-Making

Current forest land valuation research faces substantial challenges that limit our understanding of investment decision-making in this critical sector. Despite forests covering 39% of EU land area and generating €27.9 billion annually, existing studies on land valuation remain fragmented across different regions without systematic synthesis of their findings. Traditional hedonic pricing models, while widely used, struggle to capture the complex non-linear relationships between forest characteristics and prices. Furthermore, as private investors increasingly enter forest markets alongside traditional forest managers, we lack empirical evidence about how these different stakeholders value forest land. Perhaps most critically, behavioral factors such as risk attitudes and time preferences, known to influence forestry decisions due to long production cycles, have been largely ignored in forest land valuation research.
This project addresses these gaps through an integrated three-year research program combining systematic literature review, machine learning analysis of over 13,000 transaction records from HessenForst (1997-2023), and discrete choice experiments with 800 participants including forest managers, private investors, and forestry students. The methodological innovation lies in integrating advanced machine learning techniques like XGBoost and Random Forest with traditional hedonic pricing models, providing both superior predictive accuracy and economic interpretability through SHAP values. Furthermore, this represents the first comprehensive cross-stakeholder comparison of forest investment decisions. By incorporating incentivized measurements of risk attitudes and time preferences through Holt-Laury- and Coller-Williams-tasks, the study will reveal how behavioral traits influence long-term natural resource investments. Lastly, this project aims to quantify hypothetical bias by directly comparing experimental willingness-to-pay estimates with actual transaction prices from HessenForst data analysis.
Expected outcomes include quantified determinants of forest land values, validated methodology bridging computational and econometric approaches, and actionable policy insights as forest ownership diversifies under mounting climate pressures and EU Forest Strategy 2030 implementation demands.
Keywords: forest land valuation, systematic literature review, machine learning, discrete choice experiments