The case for inflow of the broad-line region of active galactic nuclei