Does the disappearance of the size effect signal that it was spurious (i.e. firm size was never a risk factor and its premium observed prior to 1980s happened purely by chance) or does it signal a near-efficient market in which practitioners quickly exploit any anomaly?