Simultaneously Discovering and Quantifying Risk Types from Textual Risk Disclosures

Abstract

Managers and researchers alike have long recognized the importance of corporate textual risk disclosures. Yet it is a nontrivial task to discover and quantify variables of interest from unstructured text. In this paper, we develop a variation of the latent Dirichlet allocation topic model and its learning algorithm for simultaneously discovering and quantifying risk types from textual risk disclosures. We conduct comprehensive evaluations in terms of both conventional statistical fit and substantive fit with respect to the quality of discovered information. Experimental results show that our proposed method outperforms all competing methods, and could find more meaningful topics (risk types). By taking advantage of our proposed method for measuring risk types from textual data, we study how risk disclosures in 10-K forms affect the risk perceptions of investors. Different from prior studies, our results provide support for all three competing arguments regarding whether and how risk disclosures affect the risk perceptions of investors, depending on the specific risk types disclosed. We find that around two-thirds of risk types lack informativeness and have no significant influence. Moreover, we find that the informative risk types do not necessarily increase the risk perceptions of investors—the disclosure of three types of systematic and liquidity risks will increase the risk perceptions of investors, whereas the other five types of unsystematic risks will decrease them. Data, as supplemental material, are available at http://dx.doi.org/10.1287/mnsc.2014.1930 . This paper was accepted by Alok Gupta, special issue on business analytics.

Improving Financial Statement Footnotes: Evidence from Derivative and Hedging Disclosures

“…The FASB is currently re-examining certain aspects of hedge accounting, including disclosures (Burkholder, 2014 In addition to studying derivative and hedging disclosure, prior research also 7 Another stream of research studies the effects of risk-related disclosures not specific to derivatives and hedging. These papers find that risk disclosures are useful sources of information and are associated with investor perceptions of firm risk (Kravet and Muslu, 2013;Campbell et al, 2014;Bao and Datta, 2014;Hope et al, 2015;Filzen, 2015).…”

A Domain Oriented LDA Model for Mining Product Defects from Online Customer Reviews

“…Previous studies [2], [22], [17] present different kinds of method of automatic content analysis for extracting valuable knowledge from large amounts of unstructured data. For example, dictionary methods are using key words or terms to summarize documents [2].…”

“…Later works are based on some dictionary methods or supervised learning techniques to learn aspects and their ratings [21]. However, most of current studies [22] are based on unsupervised topic models or LDA. Latent aspect-based opinion mining is mainly helpful for the customers to make decisions but not from product producers' perspective.…”

Customers’ Risk Factor Disclosures and Suppliers’ Investment Efficiency

“…To test this argument, we identify whether suppliers express concerns about the volatility of future demand in their own RFDs of the 10-Ks. We use the measure from Bao and Datta (2014), who categorize and quantify the types of risk disclosed in Item 1A of the 10-K. 22 We separate supplier firms that 22. Bao and Datta (2014) employ the latent Dirichlet allocation topic model and its learning algorithm to quantify and classify the risk factors disclosed in Item 1A into 30 risk types.…”