How formal control fosters relational behavior for innovation performance in infrastructure megaprojects

Megaproject innovation has become an increasingly prominent research focus in recent years. Prior studies have proposed that relational behavior—defined as the desired actions facilitating collaboration among participant organizations (POs)—may serve as a critical antecedent to innovation in megaprojects. However, their relationship remains underexplored, with limited insights regarding the ability of formal control in promoting relational behavior. Grounded in agency theory, the present study investigated the influence of the project owner’s formal control mechanisms on the relational behavior of POs and the subsequent impact on megaproject innovation performance. In addition, prior cooperation experience—whether POs have collaborated in previous megaprojects—was introduced as a moderating variable affecting the relationship between formal control and relational behavior. Data were collected from 152 Chinese megaprojects using a snowball sampling strategy and analyzed through partial least-squares structural equation modeling (PLS-SEM). The results revealed a significant positive association between relational behavior and megaproject innovation performance. Furthermore, behavior control positively affects relational behavior, whereas outcome control does not demonstrate a statistically significant influence. Contrary to expectations, prior cooperation experience negatively moderates the relationship between behavior control and relational behavior. These findings establish relational behavior as a key antecedent of megaproject innovation performance and contribute to a more detailed understanding of the interaction between formal control mechanisms and relational dynamics by differentiating the effects of behavior and outcome control. The insights from this study offer practical guidance for megaproject practitioners aiming to foster innovation through strategic management of interorganizational relationships.