3
domestic or foreign markets more cheaply than those from countries with weaker currencies.
Therefore, FDI is expected to flow from strong-currency countries to weak-currency countries.
In this study we consider the prominent FDI determinants, the market factor, the costs of
production and the real exchange rate, based on the results found in the previous empirical studies.
DATA, MODEL AND METHODOLOGY
Data
This study works with Thailand’s annual time series data from 1970 to 1996. The
employed data are defined as follows.
LUSFDI
LJFDI
,
and
LEUFDI
stand for the net
inflows of foreign direct investments from Japan, the United States and the European Union,
respectively.
LGDP
is real GDP, which reflects the size of the Thai market.
LTWR
and
LTUC
are the monthly earning of Thai employees in the manufacturing sector and the average
long-term lending interest rates in Thailand, respectively. These two variables represent the
production costs in Thailand. The real exchange rate (
LRER
) is an average value of the FDI home
countries’ currency against the Thai baht. We use the value of the
Pound Sterling
against the baht as
the representative currency for the
EU
because compared to other
EU
members, the United Kingdom
holds the largest share in
EUFDI
. Thus, the variable
LRER
is defined as the prices of yen (
LJPY
), the
US dollar (
LUSD
) and the Pound Sterling (
LUKP
) in terms of baht. All series are measured in real
terms of 1988 prices and in logarithm. The raw data are obtained from various sources as shown in
endnote.
Model
The estimating model used in this paper is constructed based on the assumptions that FDI
inflows to Thailand in the period of study are determined by the economic factors in Thailand,
namely the market size (
LGDP
), the real wage rates (
LTWR
), the unit user cost of capital (
LTUC
),
and the real exchange rate (
LRER
). The mathematical representation of our estimating model is
expressed as:
t
t
t
t
t
t
LRER
LTUC
LTRW
LGDP
LFDI
ε
β
β
β
β
α
+
+
+
+
+=
4
3
2
1
(1.1)
The expected signs of the coefficients are:
0 ,0 ,0 ,0
4
3
2
1
f p p f
β
β
β
β
.
Medthodology
This study employs cointegration test and error correction model (ECM
).
The
cointegration requires that relevant variables be integrated of the same order. Therefore, the first step
for the cointegration test is to pretest each variable to determine its order of integration. This can be
done by unit root tests. After determining the order of integration of each variable and finding that
all variables are
I(1)
, the next step is to check the presence of a long-run relationship between
variables via OLS method. However, application of cointegration analysis is only to test for the
