This paper overcomes an important objection against the empirical rele
vance of the Benhabib-Farmer model as a potential account of actual bu
siness cycle fluctuations. This is attributable to an elasticity effec
t and a returns-to-scale effect of capacity utilization. These effects
are closely related to the empirical puzzles that capital appears to
play an insignificant role in explaining cyclical movements in output
and that the estimated labor elasticity appears to be larger than labo
r's share. Due to these effects, multiple equilibria and persistent fl
uctuations can easily occur in a growth model for externalities mild e
nough so that the aggregate-labor demand curve is downward sloping. An
alyses show that the propagation mechanism generated by capacity utili
zation under mild increasing returns is capable of explaining the peri
odic patterns of U.S. business cycles documented by Watson. (C) 1998 A
cademic Press.