The patent system is mainly a renewal system: the patent life is chosen by
the patentee in return for fees. I ask whether such a system cart be justif
ied by asymmetric information on costs and benefits of research. In such a
model I show that renewal mechanisms (possibly with subsidies) are equivale
nt to direct revelation mechanisms and therefore cannot be improved on, reg
ardless of the objective function. Under plausible circumstances, patents s
hould have a uniform life, rather than varying in length, as typically occu
rs under a renewal system.