Abstract
Financial reporting and disclosure are important means for management to communicate firm performance and governance to external stakeholders. Existing research provides limited insight into firms’ disclosure of information about their market-based assets such as brands, although brand is an important part of firms’ value proposition. Our research intends to address this research gap through (1) a theory-based conceptualization of brand-related information (BRI) and (2) an empirical analysis of BRI disclosure in corporate reports, more specifically, in US firms’ annual reports on Form 10-K. We incorporate automated text analysis to identify and measure BRI disclosure in 10-K reports. Automated text analysis allows us to process massive amounts of text efficiently and reliably. Extant studies using automated text analysis are primarily rely on word-level analysis, yet most concepts entail multiword terms or phrases to convey meaning. We address this limitation of existing methods by applying phrase-level analysis to BRI disclosure. First, we draw on the literature on customer-based brand equity and intellectual capital disclosure to develop the conceptualization of BRI. Building on this conceptualization, we explore the capacity of natural language processing to build a phrase-level BRI coding dictionary. Second, we use automated text analysis to construct a BRI disclosure index that captures the BRI disclosure level in firms’ 10-K reports. Unlike prior studies that focus on word-level analysis for construct measurement, our method utilizes proximity search to match phrases in the BRI coding dictionary with texts and incorporate a proximity-weighing mechanism to conduct a phrase-level analysis of BRI disclosures. The contribution of this study is twofold. First, this study fills the research gap in the field of information disclosure concerning BRI by conceptualizing a BRI framework and providing evidence on whether and how firms disclose BRI in financial reporting. Second, our method broadens the possibility of utilizing automated text analysis in constructing research-specific coding dictionaries as well as conducting phrase-level text analysis. While current research has made some efforts to provide a structured process for phrase-level text analysis, there is limited guidance regarding the operationalization of phrase match with proximity search. © 2022, The Author(s), under exclusive license to Springer Nature Switzerland AG.