Experts adjusting model-based forecasts and the law of small numbers

This paper conjectures that the behaviour of experts who adjust statistical-model-based forecasts obeys the Law of Small Numbers [LSN]. To put this hypothesis to an empirical test, I propose a simple but effective methodology. It is applied to a database containing information on many experts and their adjustments to forecasts of monthly sales of a variety of products. I find strong evidence in support of the LSN. Implications of this finding are discussed.