Ambiguity in the Cross-Section of Expected Returns: An Empirical Assessment