Growing investor demand for consistent and reliable sustainability disclosures necessitates effective corporate social responsibility (CSR) reporting. While many companies adopt CSR reporting, a disconnect often emerges between CSR disclosures and actual practices, known as CSR decoupling. This inconsistency diminishes information quality and potentially harms firm value and reputation. The Chinese stock market's volatility, characterized by frequent fluctuations and sharp price movements, further emphasizes the importance of understanding factors influencing stock price crash risk (SPCR). Existing research on the relationship between CSR and SPCR yields mixed results, with some studies suggesting a positive effect (mitigating crash risk) and others showing a negative effect (exacerbating crash risk). This inconsistency is partly attributed to the neglect of the CSR decoupling phenomenon. Previous studies often used only CSR disclosure or performance scores as proxies for CSR, overlooking the potential discrepancy between the two. This paper aims to clarify the mixed evidence by directly examining the impact of CSR decoupling on SPCR in the context of the Chinese stock market.

The literature on CSR reveals conflicting views on its impact. Some studies emphasize CSR's positive effects on various stakeholders, including enhanced financial performance, improved investor relations, and increased transparency. Others argue that CSR can be a tool for self-serving impression management or even a cover-up for unfavorable corporate news. The concept of CSR decoupling, the inconsistency between CSR disclosures and actual practices, has emerged as a crucial area of research. While several studies explore the determinants of CSR decoupling (environmental factors, corporate characteristics, manager traits), research on its consequences is relatively scarce. Prior research on stock price crash risk (SPCR) has primarily examined factors related to managerial rationality (both perfect and bounded). The present study bridges this gap by investigating the economic consequence of CSR decoupling on SPCR and examining information asymmetry as a potential mechanism linking the two.

This study uses data from Chinese A-share listed companies from 2010 to 2019. The sample selection process involves several steps to ensure data quality, including excluding companies in the financial sector and removing observations with missing data. The primary independent variable, CSR decoupling, is measured using textual analysis of CSR reports to gauge the 'Optimistic Tone' (a proxy for disclosure quality) and CSR performance ratings from an external database (Hexun). CSR decoupling is defined as the difference between the Optimistic Tone and CSR Performance. The dependent variable, SPCR, is measured using two indicators: negative conditional return skewness (NCSKEW) and down-to-up volatility (DUVOL). The analysis employs regression models, controlling for factors such as market volatility, firm size, leverage, and information opacity. To address potential endogeneity concerns, the study uses the Heckman two-stage model and incorporates firm-fixed effects. Further robustness checks involve alternative measurements of SPCR and CSR decoupling. Mechanism tests are conducted using subsample analyses based on different levels of agency risks, considering factors such as executives' experience (overseas and academic), institutional investor ownership, media attention, and the legal and market environment.

The baseline regression results reveal a significantly positive association between CSR decoupling and SPCR, as measured by both NCSKEW and DUVOL. This finding remains robust across various robustness checks, including the Heckman two-stage model and firm-fixed effects. The economic significance suggests a substantial increase in crash risk with higher levels of CSR decoupling. Subsample analysis shows that the relationship between CSR decoupling and SPCR is stronger in companies with higher agency risks. Specifically, the effect is more pronounced in companies with executives lacking overseas or academic experience, lower institutional ownership, less media attention, and a weaker legal and market environment. These results strongly support the hypothesis that CSR decoupling exacerbates SPCR, particularly in contexts with higher information asymmetry.

The findings indicate that CSR decoupling, by creating information asymmetry, increases the likelihood of stock price crashes. Managers may strategically use optimistic CSR disclosures to mask negative information and delay the market's recognition of adverse news. The stronger effect observed in contexts with higher agency risks underlines the importance of information asymmetry in this relationship. The results reconcile the mixed findings in existing research by explicitly considering the discrepancy between CSR disclosure and actual performance. The study's implications extend beyond the Chinese context, suggesting that the negative consequences of CSR decoupling may be a broader phenomenon.

This study provides robust evidence that CSR decoupling exacerbates stock price crash risk, particularly in firms with higher agency risks. The findings highlight the importance of aligning CSR disclosures with actual practices to enhance information transparency and mitigate financial market instability. Future research could explore the role of specific CSR initiatives in decoupling, use more nuanced measures of CSR disclosure, and extend the analysis to other market contexts. It would be beneficial to investigate if the impact varies across different types of CSR activities.

The study's reliance on textual analysis for measuring CSR disclosure and external ratings for CSR performance might limit the precision of the CSR decoupling measurement. The sample is limited to Chinese A-share listed companies, potentially affecting the generalizability of the findings. The study does not include data from 2020 and beyond due to the disruptions caused by the COVID-19 pandemic. Future studies could investigate the impact of the pandemic.