The lower the marginal cost in terms of labor units, the higher labor produc-
tivity and the output that firms can obtain (1=c(I)). As for c(I) and φ(I), we
assume, consistently, that c' < 0, c(0) = 1; and lim c = 0; and that φ' > 0
I!1
and ≠(0) = 0. Moreover, in order to have a well behaved maximum problem we
further assume that φ" > 0. For concreteness, we use the following functional
forms that satisfy the above requirements: c(I) = 1=(1 + I), φ(I) = (1=2)I2.
Solving the MNE problem it is derived that I = (1 — βh). Notice that this
solution is suboptimal: the MNE underinvests due to a hold-up problem. If the
parties were able to contract ex-ante, the optimal investment would amount to
I =1; and they could reach a Pareto-superior solution. However, due to con-
tract incompleteness, investment and total rents are suboptimal. Note finally
that MNEs’ profits can differ across countries only because of differences in bar-
gaining power βh. Moreover, this result will hold also allowing for any possible
cross-country asymmetry resulting in productivity differences. The differential
in productivity would be compensated by an equal difference in wages.16
3.3 Modeling MAI
Countries differ in their bargaining power with respect to MNEs. Though, it is
not a-priori obvious which countries may be “high-beta” and which countries
may be “low-beta” countries. Quite often, the actual (economic, fiscal, legal,
administrative) costs of doing business in a country are only known ex-post,
once negotiations with local authorities are dealt or establishments are set.17
Hence, we assume in the following analysis that βh is revealed to MNEs only
after their commitment to invest in country h. MNEs, however, know ex-ante
how β h is distributed across countries. For simplicity, we will assume a uniform
distribution of bargaining power on the range [βι,βu].
The MAI imposes a cost on joining countries. If a country decides to enter
MAI, it has to accept a limitation of its own policy discretion vis-à-vis MNEs.
We model this loss of discretion as a reduction in countries’ bargaining power.
The reduction is assumed to be of size °, so that, if country h enters MAI,
its bargaining power falls to βh — °.18 We assume the reduction of bargaining
power ° to be exogenously given. In the subsequent analysis the term ° will be
referred to as the “strictness” of MAI. The higher is ° , the higher is the foregone
share of FDI rents extracted by countries.
We can now describe the whole game. In the first stage, countries choose
whether or not to participate in MAI and announce it to MNEs. In a second
stage,firms choose in which country to locate without knowledge of βh , but with
16So, as will be clear, the results that follow would hold also allowing for ex-ante cross-
country asymmetries.
1 7 Anecdotal evidence on MNEs that shut down operations in foreign countries or that regret
ab out their FDI decisions is quite abundant.
18The assumption of a fixed loss of bargaining power matters for our results. Of course, in
reality high-b eta countries might be those that have to accept larger concessions. However,
as long as the reduction in bargaining p ower is less than proportional for high-beta countries,
our results hold qualitatively unchanged.
10