Recent evidence of foreign direct investment in Thailand.

Brahmasrene, Tantatape ; Jiranyakul, Komain

ABSTRACT

This study examines another significant form of international business, foreign direct investment (FDI). Unlike the theory of international trade and theory of international portfolio investment, there is no well developed comprehensive theory of foreign direct investment. In particular, this paper employs an econometric model to assess the impact of crucial factors that affected foreign direct investment in Thailand from 1973 to2000. Augmented Dickey-Fuller and Phillips-Perron tests for stationarity followed by cointegration tests are implemented. The dynamic responses of foreign direct investment to changes in real income, foreign exchange rate, labor cost and inflation are investigated. The results suggest that among all of these variables only real income play an important role in determining the level of FDI in Thailand. The industrial policy that stimulates economic growth would be imperative to attract more inflow of FDI.

INTRODUCTION

Thailand is currently undergoing reforms and adjustments in fiscal and monetary policies aimed at bolstering market confidence and achieving economic recovery and stability. Amidst this crisis, foreign direct investment (FDI) remained resilient. FDI inflows to Thailand as a whole have somewhat weathered the financial crisis that hit the nation in 1997-1998. Prior to the economic downturn, the gross domestic product (GDP) grew at an annual rate of 5.4 percent from 1981to1986. The average growth rate rose sharply to 9.5 percent per annum during 1987-1995. Economic growth rate peaked at 11.2 percent in 1990. A sustained growth rate of at least 8 percent continued until 1995 then dropped to 5.9 percent in 1996. Asian Development Bank (1996) and Yam (1997) confirmed that this rate was higher than other countries (Singapore, Malaysia, Indonesia, and the Philippines) that also experienced the Asian Financial crisis of the late 1990s. Rapid economic growth prior to the crisis made Thailand a part of the East Asia miracle (Stiglitz 1996). High growth rate was mainly due to a successful industrialization. After decades of uninterrupted growth, financial difficulties led Thailand into an economic crisis in 1997. According to the Office of the National Economic and Social Development Board, economic growth rate was -1.8 percent in 1997 followed by a sharp contraction to -10.4 percent in 1998. In 1997, the International Monetary Fund implemented a rescue package to restructure the financial sector, to attract capital inflow, and to replenish foreign exchange reserves. Recently, the sustained increase in capacity utilization has led the country into an economic recovery in a rather short period of time. The economic growth rate figures in 1999 and 2000 were 4.2 and 4.4 percent, respectively.

The implementation of an import substitution policy between 1960 and 1971 focused on reducing the country's dependence on imports of foreign goods, especially capital goods, raw materials, and semi-finished products. However, this policy was not quite successful. An export-led growth strategy emerged in early 1970s. This resulted in rapid economic growth in the subsequent period. Export composition of Thailand changed more rapidly than the structure of the manufacturing itself. For example, resources were shifted away from the traditional sector to the manufacturing sector. Eggleston (1997) identified other contributing factors to economic growth such as a transition to productivity-based growth and consumer orientation, an active government role in social and economic infrastructure investment, and a commitment to global trade and investment mobilization. Investment mobilization should be accompanied by an adequate domestic savings through the convenience of depository institutions and the profitability of investment opportunities in money and capital markets. However, the nation has experienced the so-called "investment-saving gap" for a long period of time. Besides borrowing from domestic and foreign sources, direct investment from abroad is also an important source of investment funds.

The flow of foreign direct investments began when the Board of Investment (BOI) was established in 1960 to promote private investment in the country. Since then, Thailand has been more reliant on foreign capital, especially foreign direct investments that resulted when priority was shifted to the private sectors in the late 1950s. The structural change in production composition is partly attributed to the inflow of foreign direct investments. Capital inflow in the form of foreign direct investment increased from 102.9 million baht in 1977 to 64,695 million baht in 1990. The amount of foreign direct investment gradually dropped to 14,695 million baht in 1994. It rose again in subsequent years and reached 136,060 million baht in 1999. In contrast to the withdrawal of bank lending and portfolio investment which triggered a downturn in overall private capital inflows, FDI remained relatively stable and increased its importance in the nation's private capital flows. During 1970-1979, the major contributors to foreign direct investment were Japan and the United States. Hong Kong, Singapore, Taiwan, the United Kingdom and Germany joined in later years. These foreign direct investments are mainly in the form of joint ventures with domestic investors rather than acquisitions of existing firms. Nations often intervene in the flow of foreign direct investments to protect their domestic firms, employment and cultural heritages. Factors examined in this study may influence the governmental position regarding intervention in foreign direct investment.

This study assesses the impact of crucial factors affecting an inflow of foreign direct investment in Thailand during 1973-2000. Following is a review of related literature. Econometric models are constructed to determine the crucial factors. This analytical framework is described next and followed by the report of the empirical evidence. The last section provides concluding remarks. The expected outcome should suggest recommendations related to industrial policy.

REVIEW OF RELATED LITERATURE

Earlier FDI models such as Hartman's (1984) assumes that firms make their decisions on where to make capital investments on the real after-tax rates of return available on alternative investments. He found that foreign investment in the United States was strongly affected by changes in domestic tax policy of the host country. Pain (1993) constructs a FDI model with variables that can explain the level of inward foreign investment in the United Kingdom. The results show that relative factor prices especially relative labor costs play an important role in determining FDI. Moreover, the findings suggest that there is a positive relationship between FDI and the level of domestic production and a negative relationship with production elsewhere.

It is a common belief that foreign firms will be able to invest more in a host country when its domestic currency is weaker and vice versa. Barrell and Pain (1996) specified several determinants of foreign direct investment undertaken by U.S.-based multinational enterprises during 1970s and 1980s. These determinants of outward FDI are the dollar effective exchange rate, the dummy variable of exchange control, the real level of U.S. corporate profits, the volume of U.S. exports of goods, and the overall level of demand in the home and host countries. They found that the level of GNP and relative factor costs, both labor and capital, are crucial determinants of the outward foreign direct investment made by U.S. multinational firms in seven major OECD countries. Furthermore, the expectation of short-run fluctuations in the dollar is also a significant determinant of outward FDI. Blonigen (1997) reported in his study that in general inward foreign direct investment is positively affected by a depreciation in the real exchange rate. His study shows that there is a linkage between exchange rate movement and inward FDI. However, this FDI is in the form of foreign acquisition involving firm-specific assets which can generate returns in currencies. Additional finding indicates that domestic production does not induce inward FDI flows in the form of acquisition.

DATA AND METHODOLOGY

The annual data from 1973 to 2000 were retrieved from Bank of Thailand and International Monetary Fund (IMF) International Financial Statistic Yearbook. The estimation procedures are similar to those employed in Bajo-Rubio and Sosvilla-Rivero (1994). The models constructed in this study are shown below:

(1) FDI = f(GDP, R , LC, P)

(2) FDI = f(MGDP, R , LC, P) where,

FDI is real foreign direct investment or nominal value of FDI deflated by the wholesale price index.

GDP is gross domestic product adjusted for inflation by the wholesale price index.

MGDP is the real gross domestic product contributed by manufacturing activities.

P is an inflation rate or the percentage change in the consumer price index.

LC is the minimum wage in the Bangkok metropolitan area and its vicinities.

R is the real exchange rate measured in terms of domestic currency (baht) per US dollar multiplied by relative prices (the ratio of US CPI/Thai CPI).

Gross domestic product and real gross domestic product contributed by manufacturing activities are used as a proxy for the market size and for the degree of infrastructure development and production capacity. A depreciation in domestic exchange rate would mean an increase in FDI inflows. This depreciation increases relative wealth and leads to foreign acquisition involving certain assets. The real exchange rates in terms of U.S. dollars are used because the U.S. dollar is a dominant currency for settling international transactions. No matter how multipolar international finance may become and how much the Euromarket expands U.S. dollars remain the premier international currency. For example, a growing proportion of Japan's trade is conducted in yen, but most countries continue to prefer payment in US dollars. Furthermore, the role of yen deposits for settlements was extremely limited (Nakao, 1995). Even among the Asian countries, the percentage of official reserves held in yen was far less than dollar reserves. Labor cost is used to capture the popular belief that at the final stage of international product life cycle, the companies build production facilities in low-cost developing countries to reduce production costs in response to increased competition. Labor cost should be negatively related to the FDI inflows. The firms' unit labor cost is not available. Thus, the unit labor cost is proxied by the real minimum wage which is nominal minimum wage deflated by the consumer price index. Inflation rate is included to reflect the instability and uncertainty associated with the change in real asset value. Case in point, higher inflation rate discourages FDI inflows.

Limitation

The change in tax policy variable as mentioned by Hartman's paper (1984) is omitted due to the unavailability of the data. Money market rate can be used as a proxy for the cost of capital. However, the money market rate for Thailand has just been available beginning in 1990. Prior to 1990, Bank of Thailand (the central bank) published discount rate, deposit rates and government bond rates. Moreover, the Thai government issues a small amount of bonds that cannot well represent the cost of capital. In addition, the higher interest cost in Thailand discouraged the multinational corporation to borrow from the Thai local financial institutions. For these reasons, the cost of capital was excluded.

Estimation Procedures

Many researchers use unit root test to investigate the dynamic nature of economic times series data. The unit root test of stationarity and the cointegration test are two procedures employed to test the properties of time series data used in the model:

First, two standard unit root tests of stationarity is performed--Augmented Dickey-Fuller (ADF) test (Dickey & Fuller 1979 & 1981) and Phillips-Perron (PP) test (Phillips & Perron 1988). Both examine the null hypothesis that a unit root at level of a variable exists. A time-series that has a unit root is a non-stationary time-series.

(1) Augmented Dickey-Fuller (ADF) test:

[DELTA][X.sub.t] = [b.sub.0] + [b.sub.1]T + z[X.sub.t-1] + [SIGMA][a.sub.i][DELTA][X.sub.t-i] + [e.sub.t]

X is a variable being tested for stationarity.

T is the time or trend variable.

e is the error term.

i is the number of lagged differences of X needed to make the error term serially uncorrelated.

If z = 0, X has a unit root.

(2) Phillips-Perron (PP) Test:

[DELTA][X.sub.t] = [b.sub.0] + [b.sub.1]T + z[X.sub.t-1] + [e.sub.t]

The null hypothesis is that b1 = z =0 or a unit root exists in X. Second, despite the fact that each of them are nonstationary but are integrated in the same order, the cointegration test is performed to find out that the linear combination of these variables might be stationary. The theory of cointegration discussed by Engle and Granger (1987) states that if X and Y are both integrated of order one, I(1), but their linear combination [Z.sub.t] = [Y.sub.t]-A[X.sub.t] is stationary, i.e., order zero, I(0), then X and Y are said to be cointegrated or have a long-run relationship. In order to accomplish this task, the log-linear of the FDI equation is run by the Ordinary Least Square (OLS) method. Then the residual obtained from the estimated equation is tested using the Dicky-Fuller residual based tests for a unit root shown below:

[DELTA][e.sub.t] = (p-1)[e.sub.t-1] + [v.sub.t]

where e is the residuals obtained from the OLS regression and v is the error term. In short, this procedure tests the null hypothesis of p=1. Cointegration exists if the null hypothesis is rejected. In other words, there is a stable linear steady-state relationship between the aggregate foreign direct investment and its explanatory variables. The t-ratio can be compared with the McKinnon critical values as found in Davidson and MacKinnon (1993).

Finally, without the presence of cointegrating relation, the first difference of log-linear FDI equation will be estimated by the OLS regression. However, if the test result indicates cointegrating relation, the dynamic equations suggested by Stock and Watson (1993) will be applied in both FDI equations. Banerjee et. al. (1986) stated that the estimation of dynamic equations should be employed because this method is more efficient with relatively small sample size than the Granger-Engle two-stage procedure.

EMPIRICAL RESULTS

Using the Augmented Dicky-Fuller test accompanied by Phillips-Perron unit root test, the results are shown in Table 1. The estimated test statistics are compared with McKinnon 5 percent level of critical values to reject a unit root hypothesis. Table 1 presents the ADF and the PP tests for the null hypothesis that each series contains a unit root against the alternative hypothesis that it does not. Test with and without trends are performed to ensure accuracy since the series may or may not exhibit deterministic trends. The ADF and PP tests with and without a linear trend show that the null hypothesis of a unit root is accepted for almost all series. These two tests (ADF and PP) only give contradictory results on real labor cost. For this particular variable under no trend, ADF test indicates the presence of unit root but PP test rejects the null hypothesis of a unit root at 5 percent level of confidence. Overall, the results suggest that there is little evidence that each series will contain more than one unit root. Therefore, each series is nonstationary.

Results of the unit root tests on first differences are shown in Table 2. Table 2 presents the ADF and the PP tests for the null hypothesis that the first difference of each series contains a unit root against the alternative hypothesis that it does not. The ADF and PP tests without a linear trend show that the null hypothesis of a unit root can be rejected for all series even though the two tests give some contradictory results for real exchange rate and real manufacturing GDP. For real exchange rate with trend, PP test indicates the presence of unit root but ADF test rejects the null hypothesis of a unit root. With respect to real manufacturing GDP, ADF test indicates the presence of unit root but PP test rejects the null hypothesis of a unit root at 5 percent level of confidence. Nonetheless, each series is integrated in the same order, i.e. I (1). When they are integrated at the same order, they could be cointegrated (Gujarati, 1995).

Furthermore, ADF procedures for cointegration test are employed. This is a residual-based test to test for a unit root of the residuals of the OLS regressions of equations (1) and (2). The results are reported in Table 3. The residuals of both equations do not contain a unit root because the null hypothesis of a unit root could be rejected at 5 percent level. Note that the McKinnon critical value for rejecting a unit root hypothesis of at the 5 percent level is -1.955. Therefore, the time series variables in the estimated equations are cointegrated.

In view of the fact that the series are nonstationary and cointegrated, Stock and Watson's dynamic ordinary least squares method are employed. This method includes lead and lags operators to correct for serial correlation and simultaneity bias in small samples. A precondition to apply the dynamic equations is that all the time series variables are nonstationary. The results are reported in Tables 4 and 5. The results suggest that among all of these variables only real income (real GDP or real manufacturing GDP) play an important role in determining the level of FDI in Thailand. Both variables are significant at less than 5 percent level of confidence. All variables except the real exchange rate variable in model 2 exhibit expected sign. Real exchange rate variable was also replaced by nominal exchange rate. However, all tests indicate that real or nominal exchange rate, real labor cost and inflation rate are insignificant in determining FDI in Thailand.

CONCLUSIONS

Contradictory to the traditional view that cheap labor costs induce FDI in Thailand, this study finds that only domestic income is the crucial determinants of FDI in Thailand. The pinnacle of the economic crisis led the nation into the shortage of capital for financing production and trade. Amid this problem, it is imperative for the government to maintain a stable and growing economy. Moreover, it should recognize the FDI role in restoring economic growth and development. The intensified efforts to attract FDI may include the opening of certain industries, such as service sector; relaxing the rules regarding financing, mode of entry and ownership, and granting special incentives and privileges. Further investigation of the driving forces of FDI may include factors such as the availability of cheap assets and the long term prospects.

REFERENCES

Asian Development Bank (1996) Asian Development Outlook 1996. Manila.

Bajo-Rubio, Oscar & Sosvillar-Rivero, Simon (1994) An Econometric Analysis of Foreign Direct Investment in Spain. Southern Economic Journal, 61(1), 104-120.

Banerjee, Anindya, Juan J. Dolado, David F. Henry & G. W. Smith (1986) Exploring Equilibrium Relationships in Econometrics through Static Models: Some Monte Carlo Evidence. Oxford Bulletin of Economics and Statistics, 46, 253-277.

Barrel, Ray, and Pain, Nigel (1996) An Econometric Analysis of US Foreign Direct Investment. Review of Economics and Statistics, 78(2), 200-207.

Blonigen, Bruce A. (1997) Firm-Specific Assets and the Link between Exchange Rates and Foreign Direct Investment. American Economic Review, 87(3), 447-465.

Davidson, Russell, and MacKinnon, James G. (1993) Estimation and Inference in Econometrics, Oxford: Oxford University Press.

Dicky, David & Fuller,Wayne (1979) Distribution of the Estimators for Autoregressive Time Series with a Unit Root. Journal of the American Statistical Association, 74(336),427-431.

Dicky, David & Fuller,Wayne (1981) Likelihood Ratio Statistics for Autoregressive Time Series with a Unit Root. Econometrica, 49(4), 1057-1072.

Eggleston, Karen (1997) The Sustainability of East Asian Growth. ASEAN Economic Bulletin, 14(1), 14-31.

Engle, Robert F. & Granger, Clive W. J. (1987) Cointegration and Error Correction: Representation, Estimation, and Testing. Econometrica, 55(2), 251-76.

Gujarati, Damodar N. (1995). Basic Econometrics, New York: McGraw-Hill, Inc., 709-730.

Hartman, D. G. (1984) Tax Policy and Foreign Direct Investment in the U. S. National Tax Journal, 37(4), 475-487.

Nakao, Shigeo (1995) The Political Economy of Japan Money (Chapters 4 and 9), Tokyo: University of Tokyo Press.

Pain, Nigel (1993) An Econometric Analysis of Foreign Direct Investment in the United Kingdom. Scottish Journal of Political Economy, 40(1), 1-23.

Phillips, Peter C. B., and Pierre Perron (1988) Testing for a Unit Root in Time Series Regression. Biometrika, 75(2), 335-346.

Stiglitz, Joseph E. (1996) Some Lessons from East Asian Miracle. World Bank Research Observer, 11(2), 151-177.

Stock, James H., and Watson, Mark W. (1993) A Simple Estimator of Cointegrating Vectors in Higher Order Integrated Systems. Econometrica, 61, 783-820.

Yam, Tan Kong (1997) ASEAN in a New Asia: Challenges and Opportunites in Yue, Chaia Siow and Pacini, Marcello, ed. ASEAN in the New Asia: Issues and Trends. Singapore: Institute of Southeast Asian Studies, 1-33.

Tantatape Brahmasrene, Purdue University North Central

Komain Jiranyakul, National Institute of Development Administration

Table 1 Test of the Unit Root Hypothesis

Variables ADF Statistic PP Statistic

 No Trend Trend No Trend Trend

Log of Real FDI -0.606 -3.456 -0.997 -2.926
Log of Real -1.070 -3.176 -0.434 -2.736
Exchange Rate
Log of Real GDP -0.916 -1.762 -0.849 -2.820
Log of Real -2.771 -3.267 -3.687 -3.481
Labor Cost
Log of Inflation -0.755 -3.435 -0.467 -1.740
Rate
Log of Real MGDP -0.422 -1.896 -0144 -1.909
5% Critical Value -2.985 -3.603 -2.980 -3.594

Note: lag of ADF test =1 and lag of PP test=2

Table 2 Unit Root Test of First Differences

Variables ADF Statistic PP Statistic

 No Trend Trend No Trend Trend

Log of Real FDI -4.099 -3.997 -5.714 -5.730
Log of Real -4.426 -4.313 -3.348 -3.292
Exchange Rate
Log of Real GDP -4.020 -4.209 -7.157 -7.143
Log of Real -5.142 -5.164 -3.686 -3.953
Labor Cost
Log of Inflation -3.929 -3.980 -3.277 -3.610
Rate
Log of Real MGDP -2.757 -2.698 -3.938 -3.849
5 Critical Value -2.991 -3.612 -2.985 -3.594

Note: lag of ADF test=1 and lag of PP test=2

Table 3 Cointegration Tests

Estimated Equation ADF Statistic

Equation (1) -3.711
Equation (2) -3.874
Critical value 5 percent level -1.955

Table 4 FDI Model 1

Variable Coefficient Std. Error

Constant -4.245792 8.188820
Log of Real GDP 2.608145 1.038793
Log of Real Exchange Rate 1.192959 2.226055
Log of Real Labor Cost -3.890238 2.204727
Log of Inflation Rate -0.338639 0.423948
R-squared 0.921252 F-statistic
Durbin-Watson Statistic 1.957730 Prob (F-statistic)

Variable t-Statistic Prob.

Constant -0.518486 0.6154
Log of Real GDP 2.799542 0.0188
Log of Real Exchange Rate 0.535907 0.6037
Log of Real Labor Cost -1.764499 0.1081
Log of Inflation Rate -0.798903 0.4429
R-squared 9.748935
Durbin-Watson Statistic 0.000544

Table 5 FDI Model 2

Variable Coefficient Std. Error

Constant 10.11060 14.94726
Log of Real MGDP 2.136392 0.923711
Log of Real Exchange Rate -0.408484 2.576425
Log of Real Labor Cost -3.885865 2.710192
Log of Inflation Rate -0.285409 0.588858
R-squared 0.921252 F-statistic
Durbin-Watson Statistic 0.921252 Prob (F-statistic)

Variable t-Statistic Prob.

Constant 0.676622 0.5140
Log of Real MGDP 2.312837 0.0433
Log of Real Exchange Rate -0.158547 0.8772
Log of Real Labor Cost -1.433979 0.1821
Log of Inflation Rate -0.484682 0.6383
R-squared 9.291406
Durbin-Watson Statistic 0.000668