Document Type : Original Research Paper

Authors

• Mohammad Ebrahim Raei Ezabadi ¹
• Babak Nabati ²

¹ Department of Finance and Accounting, ST.C, Islamic Azad University, Tehran, Iran

² Department of Finance and Accounting, ST.C, Islamic Azad University, Tehran, Iran.

Abstract

BACKGROUND AND OBJECTIVES: The insurance industry in Iran, with 42 active firms, accounts for a relatively limited share of the total capitalization of the Iranian capital market. Nevertheless, its economic and informational importance is greater than what its market share alone may suggest. Insurance companies operate in a highly risk-oriented environment, where underwriting decisions, claim obligations, technical reserves, investment activities, and solvency conditions are closely linked to uncertainty and future expectations. For this reason, financial and managerial disclosures issued by insurance firms can provide market participants with important signals about risk exposure, operational performance, managerial outlook, and the credibility of corporate reporting. In such a context, the timing of information disclosure becomes particularly important, because delays in releasing information may affect investors’ perceptions of transparency, information asymmetry, and reporting reliability. At the same time, the textual quality of disclosure reports may theoretically influence how investors interpret and process disclosed information, especially when the reports contain risk-related or forward-looking content. However, whether these textual characteristics are actually reflected in short-term capital market reactions remains an empirical question. Accordingly, this study aims to simultaneously analyze the timeliness of information disclosure and the textual quality of disclosure reports, and to examine their roles in shaping capital market reactions within Iran’s insurance industry.
METHODS: This study is applied in terms of purpose and quantitative in nature, and it was conducted within the positivist research paradigm. The empirical analysis is based on data from 19 insurance companies listed on the Tehran Stock Exchange over the period 2015–2024. In the first stage, the market reaction to information disclosure was measured using the event study methodology. For this purpose, logarithmic cumulative abnormal returns (CAR) were calculated for the shares of the sample firms across several event windows surrounding the disclosure dates. This approach made it possible to capture short-term abnormal stock return behavior before and after the release of disclosed information. In the next stage, textual disclosure quality indicators were extracted using natural language processing techniques. Specifically, contextual language representations generated by the ParsBERT model were used to measure semantic similarity in relation to two disclosure content dimensions: risk disclosure and forward-looking disclosure. These semantic similarity measures were employed as proxies for the textual quality of disclosure reports. Subsequently, the effects of both the occurrence of disclosure delay and the magnitude of disclosure delay, together with textual disclosure quality, on capital market reactions were analyzed. The empirical tests were conducted using linear regression models based on ordinary least squares (OLS), a logit model for examining the direction of market reaction, and gradient boosting–based machine learning algorithms as complementary tools for assessing nonlinear relationships and the robustness of the results.

FINDINGS: The event study results indicate that the capital market exhibits an immediate and predominantly negative reaction to delayed information disclosure, with this reaction being discontinuously concentrated on the disclosure day. Regression results show that the occurrence of disclosure delay has a statistically significant negative effect on logarithmic cumulative abnormal returns, whereas delay severity and textual disclosure quality measures do not exhibit a stable or significant impact on the magnitude of market reactions. The logit model results show that the occurrence of disclosure delay is associated with an increased likelihood of a positive market reaction. By contrast, delay severity reduces this likelihood. Disclosure quality has no significant effect on the direction of market reaction in the short-term window; however, in the medium-term window (-2, +2), its interaction with delay severity is positive and statistically significant, suggesting that the relationship between delay severity and the likelihood of a positive market reaction may vary depending on the level of disclosure quality. Furthermore, the weak performance of machine learning models in predicting the continuous value of logarithmic cumulative abnormal returns indicates the difficulty of predicting the intensity of market reactions over short-term horizons.
CONCLUSION: Market reaction to information disclosure is multidimensional. In the linear models, the occurrence of delay is associated with a reduction in cumulative abnormal returns, whereas in the logit model, the occurrence of delay is associated with an increased likelihood of a positive market reaction. Textual disclosure quality has no significant effect on the magnitude of CAR in the linear models; however, in the logit and machine learning models, it may contribute to explaining the direction of market reaction. This distinction is important because different empirical models capture different aspects of investors’ responses to disclosed information. Therefore, the effect of disclosure delay should be interpreted by distinguishing between the magnitude of returns and the direction of market reaction.

Keywords

• Disclosure Delay
• Disclosure Quality. Insurance Industry
• Logit Model
• Machine Learning

Main Subjects

• financial markets