Modelling Experience as Signal Accumulation

Abstract

Experience gained in a workplace characterised by decision-making and learning-by-doing is modelled via a process of signal accumulation under several different frameworks. We initially look at the probability of success based on uninterrupted signal accumulation, then consider the impact of rapid labour turnover under two alternative regimes. The first allows new workers to gain some of their predecessor�s experience through Bayesian inference on reported earlier actions. The means of information transfer between workers is therefore similar to observational learning in herding or informational cascade models. The second regime considers all experience to be lost when a worker is replaced. We see that although with valuable experience the first regime appears a much better outcome for firms, transferring some knowledge to future workers carries with it the risk of excess inertia in decision-making.