that the Alpha variable remains insignificant, but more importantly, we see that the Alpha2
variable is also not significant. This latter finding is consistent with Mikkelson et al. when
they considered a nonlinear relationship.
In model 3, however, we see that the estimated coefficients on all of the ownership
variables are statistically significant, and of the expected signs. That is, the coefficients
for Alpha and Alpha3 are positive, while the coefficient for Alpha2 is negative. These
results suggest that firms with ‘low’ and ‘high’ levels of managerial ownership experience
positive relationships between managerial ownership and changes in firm performance
(alignment-of-interest hypothesis), while firms with ‘intermediate’ levels of
managerial ownership exhibit a negative relationship between managerial ownership and
changes in firm performance (entrenchment hypothesis). These findings are actually
consistent with the earlier general findings on the relationship between firm performance
levels and managerial ownership (e.g., Morck et al., 1988; Short and Keasey, 1999), but
this cubic form has never been considered in the IPO literature when examining the
change in performance from before to after the IPO. Thus, not differentiating for three
levels of managerial ownership may explain why Mikkelson et al. cannot empirically
support the theoretical relationship between ownership and performance. Entrenchment
can occur at high levels of ownership, but at very high ownership levels, management is
essentially the owners. Here, there is perfect or near-perfect alignment between the
manager and owner.