Implementing theoretical models in the laboratory, and what this can and cannot achieve

We investigate the methodology used in a significant genre of experimental economics, in which experiments are designed to test theoretical models by implementing them in the laboratory. Using two case studies, we argue that such an experiment is a test, not of what the model says about its target domain, but of generic theoretical components used in the model. The properties that make a model interesting as a putative explanation of phenomena in its target domain are not necessarily appropriate for such tests. We consider how this research strategy has been legitimised within the community of experimental economists.