Entrepreneurial overconfidence and firm survival: an analysis using the Kauffman firm survey

ABSTRACT

We used the Kauffman Firm Survey of ventures founded in the year 2004 to identify overconfident (OC) entrepreneurs and found, contrary to the existing literature, that the hazard ratio for these entrepreneurs was lower than the corresponding value for their non-OC peers. We categorized an entrepreneur as OC if he or she believed that his or her firm enjoys a competitive advantage over its industry peers, while simultaneously underperforming relative to the industry in terms of average initial ROA. Specifically, we looked at the average ROA during the years 2004–2007 and compared it to the industry median. In addition to our hazard findings, we discovered that these OC entrepreneurs, while starting with lower initial ROA levels relative to their industry peers, may have enjoyed slightly better movement in ROA over the intervening years. Our results are explained in the context of the psychological literature on optimism.

KEYWORDS:

• Kauffman survey
• firm survival
• firm performance
• overconfidence
• optimism

JEL CLASSIFICATION:

• M13
• M21
• L25
• L26

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

¹ Overconfidence is stronger when subjects self-select into the skill-based sessions. Cain, Moore and Haran (2015) find that, rather than confidence in one’s own skill, confidence in one’s skill relative to that of others drives market entry.

² KFS is further described in Ballou et al. (2008) and Baek and Neymotin (2016).

³ Libby and Rennekamp (2012, 205) classify overconfidence into over-optimism (i.e. overestimation of mean) and miscalibration (i.e. underestimation of variance). Their over-optimism definition is consistent with our overconfidence measure.

⁴ In the initial year (2004), 1719 respondents answered ‘no’ to this question while 3139 said ‘yes.’ The remaining 70 out of the total 4928 individuals did not respond.

⁵ The difference is significant at the 10% level.

⁶ No response to the comparative advantage question is considered a missing observation. For testing the robustness of the sample, we also regarded the non-responses as having no comparative advantage. We also removed the observations with medium 20% ROA, and used 40-20-40 definition of overconfidence. Results are qualitatively the same. Consequently, this study contributes to the literature by shedding additional light on the field with such different specifications.