Managerial Overconfidence, Self-Attribution Bias, and Downwardly Sticky Investment: Evidence from Korea

ABSTRACT

The extant literature on behavioral corporate finance has explored the effects of overconfidence on investment–cash flow sensitivity (ICS) to explain overinvestment, yet it has overlooked the asymmetric behavior of investments in relation to changes in cash flow levels. This study examines whether investments behave asymmetrically responding to changes in cash flows and, if so, how managerial overconfidence affects asymmetric ICS. Using a sample of KOSPI and KOSDAQ firms in Korea, we find the incidence of downwardly sticky ICS in unconstrained firms. We then find that overconfident managers encourage ICS to be stickier than their rational peers do in unconstrained firms. Finally, we find that managerial overconfidence intensified by self-attribution bias induces ICS to get even stickier, suggesting more explicit evidence of corporate investment distortions. The results of alternative tests using the asymmetric models of Homburg and Nasev (2008) are qualitatively consistent with prior results. Overall, our findings imply a higher incidence of excessive investment commitments driven by overconfident managers.

KEY WORDS:

• corporate investment
• investment–cash flow sensitivity (ICS)
• overconfidence
• self-attribution bias

Notes

^1. Psychology research has suggested that the overconfidence syndrome overlaps considerably with narcissism, as a combination of both attitudes and behaviors (see Petit and Bollaert 2012).

^2. Hirshleifer (2001) describes the association between overconfidence and self-attribution bias as follows: “overconfidence and self-attribution bias are static and dynamic counterparts: self-attribution causes individuals to learn to more overconfident” (Billet and Qian 2008, 1037).

^3. Besides the proxy of Lin, Hu, and Chen (2005) using management’s forecasting error, Malmendier and Tate (2005) measure managerial overconfidence as a reflection of CEOs’ option-exercising behavior, while Ben-David, Graham, and Harvey (2008) and Santner and Weber (2009) both use survey results on the distribution of CEO forecasts.

^4. We use an adjusted version of the sticky model of Anderson, Banker, and Janakiraman (2003) and Chen, Gores, and Nasev (2013) from the cost accounting literature. After Anderson, Banker, and Janakiraman (2003) empirically suggested the sticky cost phenomenon, subsequent studies demonstrated that cost stickiness pervades under different settings by incorporating a decreased performance dummy into their models (Via and Perego 2014; Yang 2015).

^5. Following Lin, Hu, and Chen (2005), Huang et al. (2011) eliminate companies that disclose overestimated earnings to obtain the approval of external financing such as bank loans rather than because of managerial overconfidence.