Abstract
Narrative economics argues that stories move asset prices, yet most empirical work flattens complex narratives into simple positive or negative sentiment or broad topics. To address this, we introduce the vector of needs (VoN), a five-dimensional framework that measures the specific human motives a story activates: security, status, control, empathy, and transcendence. We operationalize this framework by applying natural language processing to 4.5 million US news headlines from 2007 to 2022. We then apply the VoN to analyze the cross-sectional structure of market risk. While the VIX measures the level of risk investors expect, the VoN reveals how that risk is distributed across assets. We find that narrative composition significantly predicts future market comovement. Status narratives concentrate comovement, drawing assets together, while transcendence narratives decouple industry returns. Conditional on the VIX and standard predictors, our narrative factors forecast next month's correlation structure out of sample, adding 6.8% incremental R^2 for factor concentration and 6.4% for average pairwise correlation. Ultimately, we demonstrate that different psychological states in the public sphere produce distinctly different shapes of financial risk.