Abstract
Purpose
The usefulness of risk disclosures (RDs) to support equity investors’ investment decisions is highly discussed. As prior research criticizes the extensive aggregation of risk information in existing empirical research, this paper aims to provide an attempt to identify disaggregated risk information associated with cumulative abnormal stock returns (CARs).

Design/methodology/approach
The sample consists of 2,558 RDs of companies listed in the S&P 500 index. The RDs were filed within 10 K filings between 2011 and 2017. First, this study automatically extracted 35,685 key phrases that occurred in a maximum of 1.5% of the RDs. Second, this study performed stepwise regressions of these key phrases and identified 67 (78) key phrases that show positive (negative) associations with CARs.

Findings
The paper finds that investors seem to value most the more common key phrases just below the 1.5% rarest key phrase threshold and business-related key phrases from RDs. Furthermore, investors seem to perceive key phrases that contain words indicating uncertainty (impacts) as a negative (positive) rather than a positive (negative) signal.

Research limitations/implications
The research approach faces limitations mainly due to the selection of the included key phrases, the focus on CARs and the methodological choice of the stepwise regression analysis.

Originality/value
The study reveals the potential for companies to increase the information value of their RDs for equity investors by providing tailored information within RDs instead of universal phrases. In addition, the research indicates that the tailored RDs encouraged by the SEC contain relevant information for investors. Furthermore, the results may guide the attention of equity investors to relevant text passages whose deeper analysis might be useful with regard to investors’ capital market decisions.