It is argued that a "laboratory" data set would greatly facilitate the teaching of graduate-level production economics. Development of a process model and an optimal experimental design for generating pseudo data are outlined. A translog, multiproduct profit function is estimated and the resulting net and gross elasticities are discussed.