Technology adoption and training practices as a constrained shortest path problem

摘要

Technology adoption is not a new venue for research. Much work of decision modeling, diffusion of new technology and statistical analysis of survey data has been done. Some studies focus on finding the optimal forms of technology to adopt within a complementarity framework, but there is no mention of finding an optimal path from a firm's current state to its optimal state. This represents a significant gap in the literature. The paper applies a constrained shortest path problem to training and technology adoption decisions by firms. Given the current set of training and technology adoption the method solves for what technology/practice should be adopted or removed from the complete set of combinations and in what order so as to maximize performance subject to budget constraints. To the authors' knowledge, this is the first application of the constrained shortest path problem to technology adoption decisions.A modified version of the Lagrangian relaxation with enumeration method is developed and tested using randomly generated constrained shortest path problems and compared to current leading algorithms. The modified Lagrangian relaxation method was shown to outperform some leading methods for the constructed test problems. We used workplace and employee level data from a linked employer-employee survey (1999-2004). We found that the best practices using profit are not the same as for labor productivity. We find that the path for labor productivity growth consistently includes computer adoption and on-the-job training. For profit growth, both classroom training and employer sponsored career development training are always present in the constrained shortest path. For static profit, adoption of computer-controlled/assisted technology is present throughout the constrained shortest path.