Group vs. Individual Performance in a Price-Searching Experiment

Abstract We report results of an experiment evaluating performance of small groups relative to individuals in an iterated monopoly-pricing problem. Profits generated by 10 three-person groups and 18 individuals in six nominal groups are compared. Half of the sessions were conducted under a “coarse-grid” treatment in which price choices were restricted to 25-cent increments. The remaining sessions were conducted under a more complicated “fine-grid” treatment in which price choices were allowed in 5-cent increments. The problem is well suited to an investigation of learning by groups relative to individuals: solutions are not obvious and substantial learning occurs in both environments. Although in the coarse-grid treatment groups perform no better than individuals, in the fine-grid treatment groups respond better to feedback than individuals and generate higher profits in later periods. We conjecture that superior learning by groups is a consequence of error-checking by groups, unhampered by any source of bias that would induce common errors across individuals.