271 /302
271
臺大管理論叢
第
28
卷第
2
期
Table 6 Panel A reports the results. We find that the coefficient on ∆
ln(Sales)
it
*FCF
it
*Dec
it
is significantly negative for
SG&A, COGS
, and
OC
at the 1 percent level. Consistent with
Chen et al. (2012), we find that the cost asymmetry behaviors can increase with agency
costs.
While our results confirm prior studies that the agency problems provide an
important explanation for cost asymmetry, we also find that agency problems do not
subsume the role of customer and supplier relationship in cost asymmetry behaviors. The
costs are less sticky for firms which have higher concentrated customer bases.
Specifically, after controlling for the impact of agency costs on cost asymmetry behaviors,
the coefficient on our main variable of interest ∆
ln(Sales)
it
*CC
it
*Dec
it
remains
significantly positive across the three columns. The coefficient is 1.3632 (
t
-statistic =
7.71), 1.0671 (
t
-statistic = 8.33), and 0.8912 (
t
-statistic = 8.15) for
SG&A, COGS
, and
operating costs (
OC
) respectively.
In addition, we test whether the stronger customer-supplier relationship can reduce
the agency problems in cost asymmetry behaviors. We use the median value of customer
concentration to split the sample, and partition our sample into high customer
concentration and low customer concentration. We compare the coefficients on the
interaction term ∆
ln(Sales)
it
*FCF
it
*Dec
it
between the high customer concentration and low
customer concentration. Table 6 Panel B reports the results. Interestingly, we find that the
coefficient for the high customer concentration sample is insignificant, but the coefficient
for the low customer concentration sample remains significantly negative. The results
suggest that that high customer concentration can mitigate the agency problem manifested
in managers’ empire building or unwillingness to downsize.