|M.Sc Student||Hani Mansour|
|Subject||An Empirical Characterization of R and D Processes in a|
Model of Sequential Search
|Department||Department of Industrial Engineering and Management||Supervisor||Dr. Bental Benjamin|
This thesis estimates an R&D model that treats R&D as a sequential search process. The R&D process precedes production and is conducted sequentially. Untried technologies are represented by a distribution. Firms sample technologies from this distribution, where each sample entails a cost. Sampling out of the technology distribution is interpreted as R&D. Every time a technology is sampled, firms must decide whether the result is satisfactory or not. In the latter case, resources are invested in an additional sample. In the former, the best technology available is used, the remaining resources are invested in capital, labor is hired and production starts.
The decision problem of whether to continue sampling or to start producing yields an R&D policy that is characterized by a stopping rule. This rule maps available levels of capital and technology into a threshold technology. The estimation procedure is based on the stopping rule that depends on some underlying parameters. A maximum likelihood procedure may be used to estimate these parameters, given data that includes the R&D expenditure of firms. Two databases are used for this purpose: An NBER data set and Bronwyn Hall’s data set. The estimation is carried out for firms belonging to five sectors in 1986 and 1988.
The estimation results show that the data rejects the basic formulation of the search setup. Nevertheless, the thesis provides some estimates that are used to carry out policy experiments. It also suggests ways to reformulate the problem.