It is generally believed that models of endogenous growth with expandi
ng product variety typically yield a too low level of research. We sho
w here that this result is due to the implicit choice of a particular
value for a crucial parameter. Once this parameter is set free, one fi
nds that too much research can occur as well.