Optimal R and D Processes and Competition

The result that competition in innovation can lead to the dissipation of rents and excessive investment in R and D is now well known in the economic literature. When the social and private benefits from discovery are equal, over-investment in R and D is a consequence of rent dissipation because resources are allocated to R and D until the average product of R and D investment, rather than the marginal product, equals the marginal resource cost. In most models, the marginal and average products diverge because of statistical duplication: when more than one research avenue or experiment can potentially lead to discovery, the marginal productivity of additional experiments declines as experimentation increases, even though no firm replicates exactly the experiments of other firms.' Hence, when experiments are independent and equal in cost and probability of success, decreasing returns from investment result. To achieve optimality, the social planner (or monopolist) must restrict investment in R and D.2 In this paper we analyse two neglected sources of potentially large gains from socially coordinated R and D relative to the competitive process: coordination of research efforts over potential experiments, and over time. By coordinating the avenues or experiments undertaken by research units, resources from an actual replication of experiments can be saved. Such replication is particularly likely during the development stage of the R and D process; at this stage, the number of promising research avenues is relatively small and well known, in contrast to basic research stage, where a great deal of uncertainty prevails. Thus, the probability that competing firms wastefully replicate experiments of their rivals increases in the later development stage. Coordination of research efforts over time ensures the efficient use of time as factor in the generation of technological information. Learning takes time, and, because of the patent race, competition prevents researchers from taking the time required to learn. These two features of the R and D process generally have been neglected in previous models because, in order to focus primarily on the patent race, the innovation process is characterized by independent research efforts and memoryless probability distributions.3 We adopt the simplest model that illustrates these aspects of the R and D process, while abstracting from those features that have been analysed extensively elsewhere. We model the R and D process as a series of experiments or avenues, culminating in successful development of an innovation, in which the experiments are drawn, without replacement, from a population of known potential experiments. Only a subset of the avenues leads to success.4 Where sequential research is feasible, an optimal research policy is derived in which a subset of avenues is covered during each period of time. If discovery is made, further experimentation ceases. Upon failure to discover, the unsuccessful avenues are discarded from the future choice set and experiments continue

Biology · Competition (biology) · Economics · Ecology · Firm Innovation and Growth · Innovation Diffusion and Forecasting · Intellectual Property and Patents

• The Poverty of Science

  Open Access•David Edgerton, Kirsty Hughes•Public Administration•1989

  Incluir item

• The economics of technological change

  Edwin Mansfield•The economics of technological…•1969

  Incluir item

• Economic Welfare and the Allocation of Resources for Invention

  Kenneth J Arrow•Rate and Direction of Inventive…•1962

  Incluir item

• Industrial Structure and the Nature of Innovative Activity

  Partha Dasgupta, Joseph E Stiglitz et al.•The Economic Journal•1980

  Incluir item