This paper presents a model in which public R&D stock is included as a quasi-fixed input in a variable cost function. Its price affects the long run desired level, while its shadow price indicates whether under (over) investment occurs in the short run.