Page 126 - 34-1

In-House Provision of Corporate Services: The Case of Property-Casualty Insurers and In-House Actuarial
Loss Reserve Certification

Table 4 Weak and Healthy Subsample Regression Results
Variable Full sample PSM sample
(1) Full (2) Weak (3) Healthy (4) PSM (5) Weak (6) Healthy
Independent Variables Sample Sample Sample Sample Sample Sample
Group Affiliation Indicator (=1 if member of -0.001 0.019*** -0.001* 0.008*** 0.026*** 0.002***
group)
(0.001) (0.001) (0.001) (0.001) (0.004) (0.001)
Publicly-traded Stock Indicator (=1 Publicly- -0.008*** -0.033*** -0.011*** -0.022*** -0.040*** -0.019***
traded)
(0.001) (0.001) (0.001) (0.001) (0.004) (0.001)
Privately-held Stock Indicator (=1 Privately- -0.001 -0.043*** 0.002*** -0.014*** -0.065*** -0.003***
held)
(0.001) (0.001) (0.001) (0.001) (0.005) (0.001)
Wald chi-squared 38469.34 71006.30 347036.83 203614.07 35358.95 427781.86
Number of Observations 7455 1399 5872 2822 455 2199
Note: This table reports results on the relationship between an In-House Actuary Indicator, organizational
form variables, and SOX variables with loss reserve error for the sample period 1999-2010. Columns
1 to 3 report results using the full sample, while columns 4 to 6 report results using the Propensity
Score Matching (PSM) sample. Feasible generalized least squares (FGLS) with a panel specific AR (1)
autocorrelation structure is used in all models. The dependent variable is the value of the reserve error
scaled by total admitted assets. All other variables are defined in Table 3. Year dummies are added in all
models. Robust standard errors are reported in parentheses below each coefficient. *, **, and *** indicate
statistical significance at the 10%, 5%, and 1% level, respectively.

the dependent variable. Recall that over-reserving is associated with a positive coefficient,
while under-reserving is associated with a negative coefficient. Column 1 contains results
using the full sample of (weak and healthy) insurers, while Columns 2 and 3 contain full
sample results for weak versus healthy insurers separately. Column 4 contains results for

the full (weak plus healthy) PSM sample, while columns 5 and 6 contain the results for the
separate PSM samples of weak and healthy insurers.
The coefficients for the In-House Actuary Indicator variable are negative and
significant in Table 4 for weak insurers in the full and PSM samples in columns 2 and

5 respectively. These results support Hypothesis 1a; weak nonpublic insurers using in-
house actuaries are more under-reserved (less over-reserved) than nonpublic insurers using
external actuaries. Note that the coefficient of intercept captures the effect of nonpublic
insurers using external actuaries. The coefficients for the in-house actuary variable are

also negative and significant for healthy nonpublic insurers in columns 3 and 6 of Table
4. These results do not support Hypothesis 1b, which predicts that the coefficient for the

118