英文摘要 |
This study employs text mining, as a substitute for the traditional method of counting the number and words of key audit matters (KAMs), to explore the effect of specific keywords in KAMs on the cost of equity capital for listed and over-the-counter companies in Taiwan. The research sample includes companies listed between 2016 and 2018. Owing to the global spread of COVID-19 in 2020 and the resultant turbulence in Taiwan, the KAMs may contain more noise, potentially leading to measurement bias in variables; hence, the sample was only tested up until 2019. Utilizing the 2020 Jieba word segmentation tool and term frequency–inverse document frequency calculations, this study assesses the readability of audit report KAMs to observe their effect on the following year’s cost of equity capital. Through pointwise mutual information, it further calculates the dependency among keywords of KAMs across different industries to identify crucial keywords. The results indicate that when the keywords in KAMs issued by auditors belong to Group 1 (revenue), Group 3 (impairment and valuation), and Groups 5, 6, and 7 (non-operational categories), they increase the cost of equity capital. However, keywords related to inventory, provisions, amounts, and realizations in Group 2 (impairment and valuation) can lead to both increases and decreases in the cost of equity capital. Keywords categorized under Group 4 (assets, impairment, and valuation) decrease the cost of equity capital. |