Responsible Innovativeness Under Sustainability Pressures: Causal Machine Learning Evidence on the Green Innovation Premium and Stakeholder Trust

Résumé fourni par la source

ABSTRACT Innovation is widely recognized as a source of competitive advantage, yet firms increasingly need to show that innovation is responsible, credible, and environmentally accountable. This study introduces the green innovation premium as its central contribution: the additional trust return to customer‐perceived innovativeness when innovativeness is grounded in visible environmental innovation. Drawing on legitimacy‐based signaling, with institutional theory, dynamic capabilities, and paradox theory specifying the mechanism's boundaries, we examine 9365 customer evaluations of 43 firms across eight industries. Double Machine Learning (DML) is used to estimate flexible, covariate‐adjusted relationships under explicit conditional‐independence and overlap assumptions; because the data are cross‐sectional and same‐source, the estimates are interpreted cautiously rather than as design‐based proof of causality. Customer‐perceived innovativeness is positively related to trust and loyalty, and environmental innovation strengthens the innovativeness‐trust relationship, particularly in service and experience‐based industries. Digital sustainability orientation, a constructed digital‐environmental integration index, does not independently predict trust after visible innovation signals are considered; a non‐overlapping digital‐capability sensitivity check reaches the same substantive conclusion. Digital‐green balance moderates the direct DSO‐trust relationship, but the conditional indirect effects on loyalty are effectively unchanged across balance levels; H10 is therefore not supported. Stakeholder trust remains the central relational pathway linking innovativeness to loyalty. The findings extend signaling theory by showing that environmental innovation changes the marginal trust return of innovativeness rather than merely adding a favorable CSR cue.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.