In assessing the empirical result summarized in Table(7-1), the following methodological constraints must be borne in mind.
Firstly, in this empirical study, the production technologies of home firms and host firms are assumed to be Cobb-Douglas. This means that the elasticity of substitution between labor input and capital input is assumed to be constant, which might be the strong assumption. The reason why the Cobb-Douglas production functions are used in this empirical study is very practical, that is, it is easy to be handled empirically. The empirical study that uses other-type production function- the trans-log production function, for example-may be needed.
Secondary, in this empirical study, the technology, which is represented by the technological level(A), the capital share rate(¦Á) and the labor share rate(¦Â), is assumed to be the same between domestic production(home firms) and foreign production(host firms). Is this assumption acceptable? This question is hard to answer. As described briefly in Section 3, one must examine what ``technology'' means. Since this empirical study is characteristically macroeconomic, ``techonology'' is assumed to be represented by the structures of production functions. The FDI model that incorporates the technology factors explicitly needs to be built in the further research.
Thirdly, in estimating the foreign capital stock(K*), the lifetime
of capital(
) is assumed to be 20 years.
The author must admit that this
assumption is too arbitrary and too strong to be accepted.
More reasonable and acceptable estimation of the lifetime of capital
will be needed in the further research.
Fourthly, in estimating the foreign labor input(h*L*) by the equation(5-4-5), the competitiveness of the product market and labor factor market is assumed, that is, the product price(P) and the labor price(w) is assumed to be exogenous. Is it acceptable? it is thought to be acceptable if one looks into consideration the difficulty of empirical work on the incompetitive market.
Fifthly, in estimating the host firm's labor input(h*L*) by the
equation (5-4-5), the wage rate of the United States(
) are
used in this empirical study as a 
value in
the equation(5-4-5). Ideally, what can be called `` the
wage rate of North America'' should
have been used. But in this empirical study this method
was not used due to the statistical problem.
Because of this, both comparing MPK with MPK* and
regressing FDI flow on the difference between MPK and MPK* are
impossible in this paper.
But it is possible to discuss the
time-series movement of MPK* itself.
Sixthly, the technology level(A)
and labor share rates(¦Â) in Cobb-Douglas production functions are
assumed to remain unchanged over time(1984-93). This methodological
assumption can be considered to be acceptable because up-and-downs of
and 
from 1984 to 1993 is very small.