Responsible workplace data governance and organizational performance: evidence from Chinese listed firms
TL;DR
The findings show that governance structures for data-intensive work can be measured at scale with employee-informed weights and behave as procedural-justice-relevant work-system conditions: they are associated with coordination-sensitive performance, strengthened by learning capability, and weakened by technical complexity, while the disclosure-based measure and observational design limit causal and individual-level psychological claims.
Abstract
Background Listed firms increasingly embed analytics, automated processing, algorithmic recommendations, and data flows into work, so data governance shapes how employees and managers experience authority, accountability, privacy, and fairness. This creates an urgent tension: data infrastructures can improve coordination and monitoring while obscuring how data are collected, interpreted, challenged, and audited, raising risks for trust, perceived fairness, voice, and cross-functional cooperation. Objective Building on this tension and prior research that emphasizes responsible AI principles more than organizational evidence, this study examines whether employee-informed and disclosure-visible responsible workplace data governance is associated with organizational performance and whether governance friction, organizational learning, and technical complexity shape this association. Methods Using 519 employee questionnaires from listed firms, five dimensions, and 19 indicators, this study builds a governance index and applies annual-report scoring to Chinese A-share non-financial firms from 2011 to 2023. Results Two-way fixed effects estimates show that higher visible responsible workplace data governance is positively associated with ROA; a one-standard-deviation increase in the index corresponds to a 0.49 percentage point higher ROA, equal to 13.2% of the sample mean. An exploratory mediation check attributes 10.8% of this association to lower administrative compliance cost; organizational learning strengthens the association, technical complexity attenuates it, and evidence for Tobin's Q is positive but less stable across timing structures. Conclusion For organizational psychology, the findings show that governance structures for data-intensive work can be measured at scale with employee-informed weights and behave as procedural-justice-relevant work-system conditions: they are associated with coordination-sensitive performance, strengthened by learning capability, and weakened by technical complexity, while the disclosure-based measure and observational design limit causal and individual-level psychological claims.