Measures of Uncalculated Risk? Simulating investment value at risk under uncertainty with fuzzy cognitive maps based on non-probabilistic narratives.
Mainstream asset valuation models typically neglect idiosyncratic (i.e. firm-specific) risk. In contrast, recent research finds that real-life investors do care about asset-specific risks and shows that idiosyncratic risk factors do influence asset prices in many securities markets. At the same time, other contemporary researchers have succesfully applied textual analysis techniques to company disclosures and other news sources to estimate asset risk levels. This project intends to apply those proven textual analysis techniques to derive and test measures of expected idiosyncratic (return and) risk from corporate disclosures. It will evaluate if and how much these measures can improve the predictive power of mainstream asset valuation models.