In the context of achieving carbon peak and carbon neutrality goals and the opening of a capital market in China, an emerging country, the relationship between an ESG rating and northbound capital shareholding preferences (NCSP) is a topic worthy of discussion. In this research, we selected CSI 300-listed companies from 2015 to 2020 as the research object and examined the influence and mechanism of the ESG rating on the NCSP. Our findings showed that the ESG rating is significantly correlated with NCSP, that the ESG rating can dramatically enhance corporate accounting conservatism, and that accounting conservatism has a partial mediating effect between an ESG rating and the NCSP. Furthermore, we noticed that the positive effect of ESG ratings on NCSP among non-state (non-SOE) corporations is more pronounced. The most prominent of the three perspectives of ESG ratings was governance (G). We found that the ESG rating had a stronger impact on the NCSP during the post-COVID-19 period than in the pre-COVID-19 period. In this paper, based on the perspective of accounting conservatism, we enrich the study of ESG ratings in the capital market, provide empirical evidence for the theoretical study of NCSP, and offer a reference for the optimization of the ESG concept and its positioning in corporations. In future studies, expanding the sample range may lead to different interesting findings.
AI-driven personalization (AIP) has become a core mechanism of digital commerce platforms, yet its psychological consequences remain theoretically fragmented. Drawing on the Stimulus–Organism–Response (SOR) framework and Psychological Reactance Theory (PRT), this study proposes a Threat-Substitution Mechanism (TSM) to explain how AIP shapes continuance intention in high-involvement online travel decisions. Using survey data from 488 Generation Y and Z users of Chinese online travel agencies and analyzing the model via PLS-SEM, results show that AIP significantly increases usage intention (UI) and reduces psychological reactance. Psychological reactance partially mediates the relationship between AIP and UI, indicating the presence of underlying psychological friction alongside dominant utilitarian benefits. Furthermore, privacy concerns amplify the negative relationship between AIP and reactance, suggesting that privacy-sensitive users exhibit heightened appraisal sensitivity rather than uniform resistance to personalization. By reconceptualizing the personalization paradox as a context-contingent threat appraisal process, this study advances electronic commerce research beyond parallel dual-effect models and clarifies the boundary conditions under which AIP enhances or constrains user continuance. Practical implications highlight the importance of algorithmic precision and autonomy-supportive design in AI-enabled commerce platforms.