Page 127 - 34-1

NTU Management Review Vol. 34 No. 1 Apr. 2024

in-house actuary variable would be insignificant for healthy nonpublic insurers. Thus,
there is very strong evidence (after controlling for firm characteristics and other incentive
conflicts) that both weak and healthy nonpublic insurers that use in-house actuaries to
certify loss reserves are under-reserve more (or over-reserve less) than nonpublic insurers

39
using external actuaries. However, the degree of under-reserving is economically very
small for healthy nonpublic insurers.
According to Hypothesis 2, weak publicly-traded insurers using in-house actuaries

should be less under-reserved (or more over-reserved) than weak nonpublicly-traded
insurers using in-house actuaries pre-SOX. The coefficient for In-House Actuary
Indicator×Publicly-traded Stock Indicator is positive for weak insurers in Table 4
column 5. The coefficient is 0.035, and it is significant at the 1 percent level. This result
is consistent with Hypothesis 2: weak publicly-traded insurers using an in-house actuary

are less under-reserved than weak nonpublic insurers using an in-house actuary pre-
SOX. Interestingly, weak publicly-traded insurers using an in-house actuary overall tend
to over-reserve than their peers using an external actuary (0.035-0.030 = 0.005, although

not statistically significant at the conventional level). This suggests that monitoring
from capital market seems to be able to constrain the incentives of under-reserving for
financially weak insurers with an in-house actuary to some extent.
The coefficient for the External Actuary×Post SOX Indicator×Publicly-traded
Stock Indicator is positive and significant in Table 4 column 5 (i.e., 0.013), signifying that

weak publicly-traded P-C insurers using external actuaries became less under-reserved
40
after SOX than weak publicly-traded insurers pre-SOX. These results for weak insurers

39 Further, Wald tests indicate that the coefficients for weak insurers in the full and PSM samples are
significantly larger (in absolute value) than for healthy insurers. For example, in columns 5 and 6, the
coefficient for the In-House Actuary Indicator is larger in absolute value for weak (-0.030) compared
to healthy insurers (-0.003). The chi-squared value for the test of equality of coefficients for the In-
House Actuary Indicator variable between the weak sample and the healthy sample is 2.76 for the full

sample, and this result is significant at the 10% level. The chi-squared value for the test of equality
of coefficients for the In-House Actuary Indicator variable between the weak sample and the healthy
sample is 4.29, and it is significant at the 5% level for the PSM sample.
40 We also estimate regressions in which the post-SOX indicator is interacted with a nonpublicly-traded
indicator variable, where the nonpublicly-traded indicator is equal to one if the insurer is a mutual or
privately-held, and it is set to zero otherwise. Interestingly, the results in the PSM samples indicate
that the coefficient of the nonpublicly-traded indicator and the post-SOX indicator is negative and

119