This paper empirically examines contagion effects of bank failures by
analyzing the behavior of deposit flows in a sample of failed and heal
thy banks over the 1929-1933 period. We find evidence of contagion for
1930-1932, while none seems to have existed in 1929 or 1933. In addit
ion, the pace of contagion accelerated over 1930-1932. We find that ev
en during 1930-1932, failing-bank deposit outflows, exceeded those at
a matched control sample of nonfailing banks. This finding is consiste
nt with the presence of a significant number of informed depositors wh
o distinguished among ex ante failing and nonfailing banks. (C) 1996 A
cademic Press, Inc.