rate fosters its foreign affiliate activity at the expense of the NE activity there.
The impact of withholding tax rates on MNE activity: The effects of
an increase in the withholding tax rate are unambiguous and straight-forward.
An increase in the host country’s withholding tax rate increases the tax burden
of the parent country’s MNEs only, irrespective of the method of double taxation
relief. Consequently, MNE activity there is then reduced.
The impact of depreciation allowances on MNE activity: If the parent
country provides more generous depreciation allowances (reflected in a higher
level of δi), the model predicts an increase of the parent’s horizontal outbound
MNE activity. The reason for this result lies in the assumption that fixed costs
are deductible in the country where they are actually paid (firm-specific and do-
mestic plant-specific fixed costs in the parent country and foreign plant-specific
fixed costs abroad). Hence, domestically headquartered MNEs gain the most,
as they face the highest fixed costs. However, this effect exists only under cross-
hauling, i.e., the coexistence of outbound and inbound horizontal MNE activity
at small (zero) relative factor endowment differences. The effects of depreciation
allowances are reversed if large relative factor endowment differences exist and
vertical MNEs prevail.18 Since vertical MNEs do not operate a production facil-
ity at the headquarters’ location, they only can deduct fixed plant set-up costs
(in contrast to domestic NEs or horizontal MNEs). Accordingly, an increase
in domestic depreciation allowances leads to a distortion in favor of domestic
NEs and vertical outbound MNE activity is crowded out. An increase in the
depreciation allowances in an unskilled labor abundant host country attracts
even more affiliate production there and fosters the parent country’s vertical
outbound MNE activity. In the remaining factor configurations, we obtain the
opposite sign to a change in depreciation allowances in the parent country.
5 Specification of bilateral outbound FDI
Our empirical analysis employs a panel data set of real bilateral outbound FDI
stocks among the OECD economies with annual information between 1991 and
2002. To guard against the bias from omitted time-invariant variables and
time-specific common shocks that affect all country-pairs in the same way, we
include fixed country-pair and time effects. The specification of the empirical
18 The effects of δi and δj are also reversed if depreciation allowances are not only applied
to fixed costs but also to variable costs, as indicated in Footnote 6.
18