What Do Good Integrated Reports Tell Us?: An Empirical Study of Japanese Companies Using Text-Mining

In recent years, ESG investment has been attracting a lot of attention both practice and academia. In the midst of these ESG investment trends, corporate non-financial information has become more important, and integrated reports that include both financial and non-financial information have been attracting attention. However, unlike financial statements, there are no standard forms in integrated reports. The content and amount of them vary, and the ingenuity of each company can be seen. In addition, while financial information can be shown as numerical values and graphs, non-financial information tends to be qualitative information. Since it is difficult to read and compare multiple integrated reports consisting of tens to hundreds of pages, the need to read more important non-financial information efficiently and objectively is increasing. Therefore, in this study, we analyze these integrated reports quantitatively by using natural language processing techniques to tell the differences between "good" and "not good" integrated reports. Here, as "good" integrated reports, we use a list of excellent integrated reports and significant improved integrated reports from the GPIF website. For the good integrated reports, we found that (1) they describe more sustainable words that create long-term value, (2) describe more about their services and customers, and (3) are proactive in sustainability and corporate governance.