metrically at ti = tj = 0.3, which roughly resembles the average corporate tax
rate in the OECD countries in 2004. We account for the fact that bilateral tax
treaties prevail among the countries of interest and set the withholding tax rate
at a low level, tiw = tjw = 0.05. We assume that about 20 percent of fixed costs
are tax deductible so that δ = 0.2.9 In the case where tax revenues are used
to finance public infrastructure, the scaling parameter determining the relative
importance of public infrastructure is set at β = 0.1.
4 Simulation results and hypotheses
Without corporate taxation (i.e., setting all parameters of taxation to zero) the
chosen calibration of the model leads to a pattern of affiliate production which
is virtually identical to the one in Carr, Markusen, and Maskus (2001) and in
Markusen and Maskus (2002). Below, we measure foreign affiliate activity as
the share of outbound affiliate production of country i (defined as (hi +vi)Xijj +
viXiji ) in the world production of X . Horizontal multinational firms prevail,
if country size and relative factor endowments are not too different. Vertical
multinationals come into existence only, if relative factor endowment differences
(i.e., production cost differences) are large enough. Higher trade costs (foreign
plant set-up costs) discourage NEs (MNEs).
Figure 1 displays foreign affiliate production in an Edgeworth box with fac-
tor endowments on the axes for the benchmark case without corporate taxation.
First, consider factor endowments along the main diagonal indicating countries
with identical relative factor endowments. Here, NEs and horizontal MNEs
prevail, and foreign affiliate production is higher the more similar the countries
are in size. Second, moving from the center towards the North-Western corner
implies larger differences in relative factor endowments, inducing an increase in
foreign affiliate production by vertical MNEs, all else equal. In the subsequent
analysis we introduce corporate taxation and primarily focus on the case where
tax revenues are used to finance public infrastructure. However, in qualitative
terms, the effects are similar for lump-sum transfers. The reason is that cor-
porate taxation affects the equilibrium plant configuration even if tax revenues
are redistributed in a lump-sum fashion.10
9 In our empirical analysis below, we measure tax base reducing allowances by depreciation
rates including first-year extra depreciations (see Devereux and Griffith, 1999, 2003). The net
present value of depreciation allowances for tax purposes is about 30 percent higher than the
assumed depreciation rate in the model simulation. The means of the periodical depreciation
rates in the sample are about 22 percent (machinery) and 5 percent (buildings), respectively.
10 Figures and other results for the lump-sum transfer case are available from the authors
12