Small Business Research in a World of Skewed Returns

Abstract

This paper observes that outcomes from small business support programs are likely to have a small fraction of observations with extremely positive outcomes that may impact program evaluations and conclusions. Current evaluations of small business support programs generally estimate the efficacy of the program using an estimation of an average outcome or an average treatment effect. Because there are so few small business programs that have been designed for evaluation, researchers often report a simple average outcome based on a survey response of treated firms or, in very rare cases, report the effectiveness of the program as an average-treatment effect based on a quasi-experimental or experimental design. Unfortunately, presenting the small business programs'' impact as an average might be understating the magnitude of the impact of the program because the actual distribution of the outcome of the program is probably not normally distributed and is probably skewed in the positive direction. In light of the possibility that small business programs produce skewed outcomes, this paper recommends that the small business evaluations include discussions on the distribution of outcomes to better inform policy makers, administrators, and participations on the nature of the outcomes.