logo ajeb vi

Giỏ hàngCart

    MUACheckout
    Gửi bài viết Submit Article
    Asian Journal of Applied Financial Econometrics (AJAFE)
    Volume 1_Number 2 , Tháng 8/2026, Trang 55-65

    Nonlinear associations between esg uncertainty and financial performance: Explainable machine learning evidence from listed real estate firms in Vietnam

    Pham Thuy Tu, Mai Ngoc Dung, Duong Hong Thin, Tran Thi Huong Thao, Tran Huu Thuan
    DOI: 10.63065/ajafe.2026.2.004
    Ngày nhận bài: 17/09/2026
    Ngày nhận bài sửa: 15/08/2026
    Ngày duyệt đăng: 09/09/2026
    Email: tupt@hub.edu.vn
    Đơn vị công tác:
    Lượt xem: 5
    Downloads: 0
    Các định dạng trích dẫn

    Tóm tắt

    Purpose – This study investigates how ESG-related informational uncertainty is associated with the financial performance of listed real estate firms in Vietnam. It examines whether this relationship is nonlinear and heterogeneous while evaluating the usefulness of explainable machine learning in an emerging market.
    Design/methodology/approach – A balanced monthly panel of 28 listed real estate firms covering 2013-2024 is analyzed. Eight machine-learning algorithms are compared, with Particle Swarm Optimization used for hyperparameter tuning. Model interpretability is achieved using SHAP, complemented by partial dependence plots and accumulated local effects to identify nonlinear and heterogeneous associations between ESG uncertainty and return on equity (ROE).
    Findings – XGBoost delivers the strongest out-of-sample predictive performance, indicating that the ESG uncertainty-performance relationship is inherently nonlinear. Explainable AI reveals a state-dependent pattern: ESG uncertainty is negatively associated with ROE at low levels, becomes broadly neutral as firms adapt, and is associated with more favorable outcomes only among firms with stronger adaptive capacity under high uncertainty. Large firms are more sensitive to ESG uncertainty, medium-sized firms display more stable adaptive responses, and small firms exhibit weaker responses. These findings represent model-implied associations rather than causal effects.
    Originality/value – This study shifts the ESG-performance literature from static ESG levels to ESG-related informational uncertainty and integrates explainable machine learning with SHAP to uncover nonlinear and heterogeneous effects, providing new evidence from Vietnam's listed real estate sector.

    Từ khóa

    Apparent randomness, chaos, chaotic dynamics, endoge- neous deterministic dynamics, exogeneous random shocks, Lyapunov expo- nent, SDIC

    Tài liệu tham khảo