Quantile factor models (QFM) represent a new class of factor models for high‐dimensional panel data. Unlike approximate factor models (AFM), which only extract mean factors, QFM also allow unobserved factors to shift other relevant parts of the distributions of observables. We propose a quantile regression approach, labeled Quantile Factor Analysis (QFA), to consistently estimate all the quantile‐dependent factors and loadings. Their asymptotic distributions are established using a kernel‐smoothed version of the QFA estimators. Two consistent model selection criteria, based on information criteria and rank minimization, are developed to determine the number of factors at each quantile. QFA estimation remains valid even when the idiosyncratic errors exhibit heavy‐tailed distributions. An empirical application illustrates the usefulness of QFA by highlighting the role of extra factors in the forecasts of U.S. GDP growth and inflation rates using a large set of predictors.

Unlike traditional asset categories (e.g., industry classifications) that are generally defined clearly, some groups of stocks are tied to certain loosely defined “concepts” (e.g., e-commerce). When investors find it difficult to analyze ambiguous concept-oriented information, information diffuses slowly, creating “concept momentum”. Based on unique concept data in the Chinese stock market, this study constructs a concept-momentum strategy that involves buying stocks from past winning concepts and selling stocks from past losing concepts, which can generate pronounced abnormal returns. Neither risk factors, firm-level momentum, nor industry-level momentum can explain concept momentum. Furthermore, we find that both the underreaction and cross-stock lead-lag effect channels can cause slow information diffusion and drive concept momentum. Moreover, the concept momentum effect is stronger for relatively ambiguous concepts, for concepts that attract less investor attention, and following high-sentiment periods.