Luxembourg from the data set as their FDI data are not trustable. As a result, we obtained
our balanced panel data set sample with 690 observations.
3.2. Simultaneous Equation System
A simultaneous equation system consists of a number of structural equations involving
several endogenous variables whose values are determined by exogenous variables and
lagged values of variables, known as predetermined variables. After each of the endogenous
variables is solved in terms of the exogenous and predetermined variables, we obtain a system
of reduced form equations.
Although the implications of simultaneity for econometric estimation were recognized
long time ago, e.g., Working (1926), the first major contribution to the area of estimating
simultaneous equation system has been made by Trygve Haavelmo (1943). According to
Haavelmo (1943), if one assumes that the economic variables considered satisfy,
simultaneously, several stochastic relations, it is usually not a satisfactory method to try to
determine each of the equations separately from the data, without considering the restrictions
which the other equations might impose upon the same variables. That this is so is almost
self-evident, for in order to prescribe a meaningful method of fitting an equation to the data, it
is necessary to define the stochastic properties of all the variables involved. Otherwise, we
shall not know the meaning of the statistical results obtained. Furthermore, the stochastic
properties ascribed to the variables in one of the equations should, naturally, not contradict
those that are implied by other equations.
If the simultaneity is ignored and ordinary least squares applied, the estimates will be
biased and inconsistent. Consequently, forecasts will be biased and inconsistent. In addition,
tests of hypotheses will no longer be valid (Ramanathan, 1998).
Our illustrative framework suggests that FDI contributes positively to the growth rate of
FDI receiving economy, and that positive growth rate stimulates FDI inflows positively. That
means, on theoretical ground, there is a bi-directional relationship between variables. Hence,
we need to consider the determination of FDI growth and growth rate together as it would not
be correct to use unidirectional relationship between these variables.
4. Econometric Analysis
In this part of the paper, we present our results out of simultaneous equation system analysis.
Our simultaneous equation system is composed of two equations:
gFDI,it =β0+β1gY,it+β2gX,it+β3hcit(-5)+β4gFDI,it(-1)+uit (10a)
gY,it =α0 +α1gFDI,it +α2gX,it +α3hcit (-5) +α4gY,it (-1) +vit (10b)
In (10a), gFDI,it is the growth rate of foreign direct investment of the ith country at time t, gY,it
is the growth rate of GDP, gX,it is the growth rate of exports, hc(-5) is five year lagged value
of human capital and gFDI,it(-1) is one year lagged value of FDI growth rate. In (10b),
gY,it (-1) is one year lagged value of GDP growth rate. Growth rate of exports is the annual
percentage change of goods and services exports. GDP growth rate is defined as annual
percentage change in GDP. Lastly, FDI growth rate is the growth rate of foreign direct
investment inflows to countries. Finally, human capital variable is the five-year lagged values
8