An Incremental Incentive Scheme (IIS) encourages some activity by rewa
rding an agent for overachieving a base level determined by past perfo
rmance but not penalizing underachievement. We examine an IIS R&D subs
idy in a dynamic model due to Grossman and Shapiro (1986). We show tha
t the firm's optimal R&D path either cycles around the no-subsidy path
or follows a ''ratchet'' pattern of small increases in R&D relative t
o the no-subsidy path. A simple condition determines which type of beh
avior occurs. Furthermore, we show that an IIS may be an inefficient m
ethod of encouraging R&D compared to a flat-rate subsidy.