Abstract
In the era of globalization, flow of capital between nations has become inevitable and it is believed that, FDI will brings with it capital, technology and employment opportunities. There is huge competition between nations in attracting FDI. But Foreign Investors are considering various factors for their investment decisions. This study examined the important determinants of FDI in India using annual time series data. Based on the past literature, certain macro economic variables have been selected as the major determinants of FDI. Interestingly, the result of the study seems to suggest that growth factor has significantly influenced FDI into India.
Key Words: Foreign Direct Investment, GDP, Determinants, Growth Factor.
Summary
In the last two decades of the twentieth century, nations compete among themselves in attracting Foreign Direct Investments. Using annual time series data from 1991-92 to 2009-10, this study examined the major determinants of FDI flows into India. Based on the past literature, certain macro economic variables such as Real GDP, Exchange Rate, Index of Industrial Production, Interest Rate, Trade openness, Corporate Tax rate, Average Real Wages and Infrastructure Development has been selected as the major determinants of FDI. The above mentioned variables were classified as growth and cost factors by applying factor analysis to overcome the multicollinearity problem. In order to analyze the influence of the growth and cost factors on the flow of FDI into India, OLS technique was used. The result of the study identifies that, FDI flows into India has been significantly influenced by the growth factor.
Introduction
Last two decades of the twentieth century has witnessed a marked change in the attitude of most of the economies towards neo-liberal economic policies. In the era of globalization, flow of capital between nations has become inevitable. Generally, it is believed that, the flow of investment in the form of FDI will brings with it huge capital, technology, skills, employment opportunities and also market access. There is very high competition between nations in attracting potential foreign investments to increase industrial development and thereby increase the growth of their economy. But investors are not coming in just because an economy is opened up for foreign investors. There are various factors involved in their investment decisions such as the economic and industrial performance of a country, cost of production, tax rates, exchange rate fluctuations, Infrastructure development, local market and political stability of a nation. In spite of implementing full capital account convertibility, many economies have not been considered as the preferable destination for foreign investors due to various factors. An economy which requires massive FDI flows for development has to identify the major host country factors which attract foreign direct investments in to the economy. Since 90's the major flow of world's investment has been shifted to developing countries, particularly to Asian economies and among them China and India has attracted the major share of FDI (UNCTAD Report, 2009).
During the first three decades of economic planning, India remained a stringently controlled economy, both internally and externally. At the end of 1980s Indian economy faced a serious economic crisis due to a huge balance of payment deficit, which had built up over the previous years and had reached a stage where India would have been forced to default on its external obligations. Such balance of payment crisis is not only because of the OPEC oil price hike and Gulf crisis; it was also because of less conservatism in the attitude of the government towards balancing the books during the second half of 1980s (Haggard and Kaufman, 1992). During 1980's fiscal deficit increased drastically year after year. Whereas the average fiscal deficit of the center was 6.3 percent of GDP in the first half of 1980s and 8.2 percent in the second half and 8.4 percent in 1990-91 against the average of 3.9 percent during 1970s (India Development Report, 1997). Persistent fiscal deficits led to inflation which in turn also affected the balance of payment. To overcome the situation the Government started borrowing rapidly, which in turn worsen the balance of payment situation and the government was forced to make structural changes in its economic policy. During 1991 economic policy reforms was launched, since then Indian market has been made wide open to foreign players. In the last two decades of liberalization, the restriction towards FDI has been made liberal in case of almost all sectors except few areas, such as atomic energy, single brand retailing, chit funds, lottery, gambling and sectors not opened to private players. In spite of allowing 100% FDI in many industries, the flow of FDI in to Indian economy is very less when compared to China. To attract potential foreign direct investments in to India, factors influencing it must be studied. Several studies have been conducted to identify the major factors which attract the flow of FDI in to China whereas in case of India very few studies have been conducted. Therefore, this study made an attempt to identify the important factors which attract the flow of FDI into India and suggest the policy makers to improve those factors in order to attract more amount of productive FDI into the nation.
Review of Literature
A large body of empirical research examines the major determinants of FDI in developed and developing countries. This section reviews the main contributions in this regard. Most of the literature on FDI focuses on the linkage between the macroeconomic indicators and FDI.
In general trade flows between countries is primarily a function of the GDP of the host country. Shapiro (1998) has discussed that market size (GDP) of the host country directly affects the expected revenue of investments. Raymond MacDermott (2007) reveals that FDI will rise with host country GDP. Many researchers have used GDP as a proxy for market size. The general assumption is that larger market size of a region attracts more FDI in to an economy. Kravis and Lipesey (1982), Blomstrom and Lipsey (1991) proved this hypothesis through their research, that market size has positively influenced the flow of FDI. Lv Na and W.S. Lightfoot (2006) examined the determinants of FDI by region in China. GDP has been used as a proxy for market size and potential. The result of the study reveals that market size and potential attracts FDI to a larger extent and the study has also identified that quality of labour and degree of openness are also the important determinants of the distribution of FDI and high labour cost deters the inflow of FDI. In line with the above mentioned study, Makki et.al (2004) analyzed the determinants of foreign direct investments by US food processing industry in developed and developing countries. They found that market size, per-capita income and openness significantly affect US food processing firms decisions to invest abroad, but their influence differs between developed and developing countries. Using annual time series data James B. Ang (2007) analyzed that real GDP is found to have a significant positive impact on FDI inflows in Malaysia. The study also suggested that increase in the level of financial development, infrastructure development, and trade openness promoted FDI and higher statutory corporate tax rate and appreciation of the real exchange rate seem to discourage FDI inflows. Interestingly, the results also appear to suggest that higher macroeconomic uncertainty induces more FDI inflows. The literature has derived important and interesting results of how infrastructure development in host country can affect FDI flows. Asiedu (2002) reveals that better infrastructure in the host economy have a positive impact on non-sub Saharan African economies and trade openness seem to have a positive impact on Sub Saharan African countries. Cheng and Kwan (2000) also confirm that good infrastructure brings in more FDI in China. They also found out large regional market had a positive effect and wage cost had a negative effect on FDI in china. Markusen (1984) and Helpman (1984) discussed that access to markets (horizontal FDI) and low wages for part of production process (vertical FDI) are the important motivations for FDI. Luger and Shetty (1985) has analyzed the determinants of foreign direct investments into North America and identified that state government spending policies has significantly influenced foreign investors and higher wage rates discouraged FDI in North America. Coughlin et al. (1991) has also identified that higher wages deterred foreign direct investment but higher unemployment rates attracted foreign direct investment in US and also factors such as higher per capita income, higher densities of manufacturing activity, extensive transportation infrastructures and larger promotional expenditures attracted foreign direct investment. He also found out that higher taxes deterred foreign direct investment in US. Hines (1996) confirms the importance of high state tax rates in deterring FDI in America significantly. Hines (1999) also investigated the distribution of FDI in US and examined the tax sensitivity of FDI into a state of "non-credit-system" foreign investors relative to that of "credit-system" foreign investors. He found that higher tax rates are associated with larger FDI decrease by the non-credit-system investors relative to the credit-system investors. Hartman (1984) examined the behavior of foreign affiliates in the United States by using the data on the host country (US) tax rates and returns. He regressed retained earnings FDI and new transfer FDI on the host country (US) tax rate separately and found out that retained earnings FDI responds significantly to the host country tax rate but new transfer FDI, however, does not respond significantly to host country tax rates. An obvious hypothesis is that appreciation of currency discourages FDI. Froot and Stein (1991) analyzed that there is increase in inward FDI with currency depreciation in US. Klein and Rosengren (1994) provided empirical evidence for exchange rate depreciation increases US FDI using various samples of US FDI. Blonigen (1997) confirms that, there is increase in inward US acquisition FDI by Japanese firms in response to real dollar depreciations relative to the yen by using industry-level data on Japanese mergers and acquisition FDI into US. Various studies discussed about the complicated pattern of FDI such as export platform FDI. Hanson, Mataloni, and Slaughter (2001) discussed the importance of export platform FDI by using data on the foreign operations of U.S. multinationals. They found out that export platform FDI is encouraged by low host-country trade barriers and deterred by large host-country markets. Later Ekholm et al.(2003) and Bergstrand et al.(2004) estimated that US outbound FDI into certain host countries serves as a production platform for exports to a group of (neighboring) host countries. Baltagi et al. (2004) analyzed about function of affiliates of Multi National Enterprises which is located in various host countries. The study reveals these affiliates are shipping intermediate goods for further processing and shipping back the finished products to the parent country. The recent literature has also emphasized the importance of neighboring country effect. Coughlin and Segev (2000) analyses that FDI into neighboring provinces increases FDI into a Chinese province. Baltagi et al. (2004) extended this approach by building up a model of MNE activity in a multi-country world, which predicts major neighboring country characteristics such as GDP, trade costs, distance, labor skills, investment risk, etc. should affect FDI into a focus country using data on US outbound FDI into OECD countries.
Methodology
This study attempt to evaluate the specific determinants of foreign direct investment inflows in to India for a period of 19 years from 1991-92 to 2009-10. By analyzing data on a macroeconomic level this study provides the vision about the macro economic variables that favour the flow of FDI in to India. Based on the earlier work certain macro economic variables have been selected for the study to analyze the important determinants of FDI flows in India. The performance of an economy plays a vital role in attracting foreign investors. One of the most important performance variables is the Real GDP of an economy and considered as the important determinant of FDI (Raymond MacDermott (2007), Lipesey (1982), Blomstrom and Lipsey (1991), Lv Na and W.S. Lightfoot (2006), James B. Ang (2007), Makki et.al (2004)). Infrastructure development has considered as another important determinant of FDI because it could increase the capital productivity and expand the availability of resources in the host economy and government spending on infrastructure is used as a proxy for Infrastructure development (James B. Ang (2007), Asiedu, E. (2002), Cheng and Kwan (2000), Luger and Shetty (1985), Coughlin et al. (1991)). Generally investors prefer economies which has greater trade openness and the volume of imports and exports together over a period of time has been considered to measure trade openness (Asiedu, E. (2002), Fedderke, J. W., & Romm, A. T. (2006)). Real effective exchange rate has been considered to examine the assumption that, diminished currency value of the host country increases the flow of FDI (Froot and Stein (1991), Klein and Rosengren (1994), Blonigen (1997)). Reducing tax rate is an effective measure to boost FDI and corporate tax rate for foreign companies is used as one of the determinants of FDI in this study (Hines (1999), Coughlin et al. (1991), Hartman (1984)). By using Average real wage Index an attempt has been made to measure the effect of labour cost on the flow of FDI (Coughlin et al. (1991), Luger and Shetty (1985), Markusen (1984) and Helpman (1984), Leonard K Cheng and Yum K Kwan (2000). In this study we have also considered interest rate and Index of Industrial production as the determinants of FDI in India. The lag of all the exogenous variables has been considered to identify their influence on the flow of FDI into India. Generally in economics the dependence of endogenous variables on exogenous variables is rarely instantaneous, they mostly respond with a lapse of time. Countries with a positive track record of previous GDP, economic growth rate, trade openness and government expenditure are expected to attract foreign investors in the forthcoming period (Biglaiserand and Derouen ( 2006)). It is also argued that, there exists a time lag relationship between the flow of FDI and its determinants such as interest rate and real effective exchange rate (Xing (2006)). Hence, we have substituted all the exogenous variables with one period lag and specified an empirical model for the study.
FDIt = β0 + β1 RGDPt-1 + β2REERt-1 + β3IIPt-1 + β4 IRt-1 + β5TOt-1 + β6TR t-1 + β7RWt-1 +
β8INF t-1 + et
(where, FDI - Foreign Direct Investment inflows, RGDP - Real Gross Domestic Product, REER - Real Effective Exchange Rate, IIP - Index of Industrial Production, IR - Interest Rate, TO - trade openness, TR - Corporate Tax rate, RW - Average Real Wages, INF - Infrastructure Development, t - time, t-1 - first lag, e - error term.)
In order to analyze the influence of the above mentioned determinants on the flow of FDI in to India, Ordinary Least Squares technique (OLS) has been applied. But for applying an OLS technique, the model should be free from the problems of Multicollinearity, Heteroscedasticity, Auto-correlation and Endogeneity (Jose Miguel Giner and Graciela Giner, 2004). The residuals should be normally distributed and the time series should also be Stationary. In this paper, the problem of Multicollinearity has been tested by comparing the tolerance level and variance inflation factor with the R² of the auxiliary regression which was found out by regressing each of the regressors with that of the remaining regressors. Tolerence value of less than (1-R square) and VIF (Variance Inflation factor) of greater than 1/ (1-R square) will indicate that the existence of multicollinearity among the regressors. Another important assumption is that, there should not be any heteroscedasticity and it is tested using the White's General Test for hetroscedasticity. If the critical value of chi-square (n*R²) which has been obtained from an auxiliary regression (u² = α1+ α2x1+ α3x2+ α4 x1²+ α5 x2²+ α6 x1x2 +ν)
is less than the table value, there is no heteroscedasticity. The normality of residuals has been tested using the histogram and Shapiro Wilk test. Then the presence of auto-correlation has been checked by applying Durbin-Watson d test. A d value between du and 4 - du indicates that there is no auto-correlation and stationarity has been tested using the unit root test (∆Yt = δYt- 1+ut). If the coefficient (δ) is not equal to zero, then the time series is said to be stationary. Endogenity problem was also checked by looking at the correlation value of the regresssors and residuals. (Damodar N.Gujarati and Sangeetha, 2007).
Results and Implications
The results in Annexure I explains that in case of the above mentioned model except multi-collinearity problem other conditions were met. The tolerance value in this model is less than (1-R square) in case of GDP, Trade Volume, corporate taxes and equal to (1-R square) in case of the remaining independent variables, but variance inflation factor is greater than 1/ (1-R square) in case of GDP, Trade volume, Corporate taxes, Average real wages and slightly less than 1/ (1-R square) in case of the remaining regressors. Therefore, the result indicates that the independent variables specified in the model are significantly related among themselves. So the above mentioned model cannot be used as such, certain variables can be removed from the model to avoid the problem of multicollinearity but then it will end up in specification bias. The data can be transformed to apply OLS but we used factor analysis to reduce the variables in to factors which would eliminate the problem of multicollinearity because correlations between factors are zero.
Annexure II shows the result of KMO and Bartlett's Test which explains the sampling adequacy for applying factor analysis and tests whether the correlation matrix is an identity matrix respectively. The Kaiser-Meyer-Olkin Measure of Sampling Adequacy is 0.680 which is above 0.5 indicates a satisfactory factor analysis to proceed further. The probability value of Bartlett's test of sphericity is 0.000 which is significant and also indicates that the correlation matrix is not an identity matrix. Varimax orthogonal rotation has been used to clarify the factor pattern in order to better understand the nature of the factors. After applying factor analysis two factors has been extracted from eight variables. The first and second factors have an Eigen value of 4.638 and 2.376 respectively. The rotated component matrix shows the loadings of eight variables on the two factors. The greater the absolute value of the loading, the higher is the variable's contribution to the factor. Variables such as Real GDP, IIP, Trade openness and Infrastructure Development loaded higher on factor 1 and lower on factor 2. The remaining variables, Real Effective Exchange Rate, Interest Rate, Corporate Tax rate and Average Real wage Index and Infrastructure Development loaded higher on factor 2 and lower on factor 1. The growth of an economy can be measured using macro-economic factors like GDP, IIP, Trade Volume and Infrastructure Development which are generally considered by foreign investors who would like to know about the trend of these factors to assess the strength and stability of the host economy. Therefore, factor 1 has been named as Growth factor. Another important assumption is that investors generally look for low cost destinations. There are various cost related factors involved in an investment decision and when we look at macro level cost factors, attributes such as labour cost, tax rate, interest rate and exchange rate plays an important role. Hence factor 2 has been named as Cost factor.
After solving the problem of multicollinearity in the original model by reducing the eight regressors in to two factors i.e, growth and cost factors, we have used the factors as regessors in the new model, as follows:
FDIt = β0 + β1GFt-1 + β2CFt-1+ et
The above mentioned conditions for applying an OLS technique have also been checked for this new model and it is also free from the problem of multicollinearity. In order to check the presence of heteroscedasticity in the model, White's General Test for Heteroscedasticity has been applied. The value of chi-square
is 6.81 and its probability value is 0.24, so there is no heteroscedasticity present in this model. The residuals in this transformed model are normally distributed and this has been proved using the histogram of residuals which delivers a bell shaped curve and Shapiro Wilk test. The calculated value is 0.927 and the probability value is 0.175 which explains that the residuals are normally distributed. The model is also free from the problem of auto-correlation, which has been confirmed by the Durbin-Watson d statistic 1.592 which lies between du and 4 - du (1.535 and 4 - 1.535). The time series model is also stationary in nature and this has been established by the unit root test and, the coefficients for all the variables were found to be significant at 10 percent. The residuals are not correlated with the exogenous variables so there is no endogenity problem. Hence the above mentioned model has satisfied all the major conditions for applying the OLS technique (Table: 1). Therefore, this model has been considered as the optimum model to identify the major determinants of FDI.
Table: 1 - Determinants of FDI
Coefficient
t-statistic
Constant
38094
11.706
***
Growth Factor
46391.3
11.855
***
Cost Factor
1364.19
0.416
R-squared
0.904
Adjusted R-squared
0.891
S.E. of regression
13618.81
Sum of squared residuals
2.78e+09
F test
70.719
***
Durbin-Watson for Autocorrelation
1.592
White's test for Heteroskedasticity
6.81
Shapiro Wilk for Normality of residuals
0.927
Unit Root Test
δ
p-value
FDI
0.179
0.025
**
Growth Factor
0.141
0.084
*
Cost factor
-0.534
0.003
***
***, **, * Level of significance 1%, 5%, 10% respectively.
Apart from satisfying all the conditions for applying OLS technique, the overall explanatory power of the model is high. From F- statistic (70.719) it can be understood that all the Betas are not equal to zero, meaning at least one of the variables in the model is of use. The adjusted R-squared is 0.891 implying 89.1% of variation in FDI is explained by the exogenous variables in the model. The OLS result shows that Constant and Growth Factor has a significant influence on FDI. Even if Growth and Cost Factor does not change FDI will change by Rs 38,094 crs. If the Growth factor changes by 1 unit FDI will change by Rs 46,391.3 crs. Cost factor cannot be interpreted this way because the variable is not significant (t stat). There is evidence for the influence of Growth factor on FDI but the same cannot be said for the Cost factor with this model. Hence it can be inferred that foreign direct investment inflows in to India has been influenced more by the growth factor than by the cost factor. (Table: 1)
In this study the growth factor consists of Real GDP, Index of Industrial Production, Trade volume and infrastructure development which has been compressed and considered as a single factor. All the variables come under growth factor has equal importance and it has been proved from the factor loadings in the rotated component matrix (Annexure II). There is very little variation in the values of factor loadings which is less than 0.022. Hence it is inferred that all the variables considered as growth factor has an equal influence on inflow of foreign direct investments in India. Thus the analysis highlights that the growth factor is the important determinant which attracts FDI flows into India rather than the cost factors. Obviously every investment decision will be made based on the growth trajectory of the host economy and no investor would prefer an investment destination where there is no or very less growth, even though the cost of production and other investment related expenses are very low. Therefore it is evident that most of the global investments are going to developing economies where there are high growth rate and infrastructure development rather than to low cost under-developed economies.
Conclusion
Foreign Direct Investment is an inevitable phenomenon in the recent wave of globalization and every nation is competing with each other in attracting foreign direct investments to enhance development. This study has empirically tested the major determinants of FDI in India and identified growth factor is the important determinant of foreign direct investment in to our economy. The study suggests that economic performance attracts foreign investors to invest in India. A boost in the growth factor might bring in more Greenfield investments and also increase our bargaining power among other developing economies.
Annexure 1:
OLS results of original model
Coefficient
t-ratio
Constant
115212
1.5537
RGDPt-1
0.128
2.974
**
REER t-1
-1382.31
-3.514
***
IIP t-1
-1304.9
-2.167
*
IR e t-1
4922.3
3.058
**
TOt-1
0.011022
0.271
TR t-1
633.986
1.108
RWt-1
-126.634
-1.091
INF t-1
-0.556
-5.609
***
R-squared
0.991392
Adjusted R-squared
0.984
S.E. of regression
5267.807
Sum squared residuals
2.50e+08
F statistics
129.57
***
Durbin-Watson
2.047
White's test for Heteroskedasticity
17.907
Normality of residuals
2.526
***, **, * Level of significance 1%, 5%, 10% respectively.
Collinearity Statistics of original model
Exogenous variables
Auxiliary Regression R²
Tolerance
VIF
(1- R²)
1/ (1- R²)
GDP RS. Crs
0.999
.000
2097.605
0.001
1000
REER
0.540
.460
2.173
0.460
2.174
IIP
0.999
.001
740.459
0.001
1000
Interest Rate
0.906
.094
10.608
0.094
10.638
Trade Volume
0.997
.002
365.829
0.003
333.333
Corporate Taxes
0.939
.059
16.441
0.061
16.393
Average Real wage Index
0.610
.390
2.567
0.390
2.564
Infrastructure
0.986
.014
69.421
0.014
71.429
Annexure 2:
Results of Factor Analysis
KMO and Bartlett's Test
Kaiser-Meyer-Olkin Measure of Sampling Adequacy.
.680
Bartlett's Test of Sphericity
Approx.Chi-Square
275.291
Sig.
.000
Total Variance Explained
Component
Rotation Sums of Squared Loadings
Eigen value
% of Variance
Cumulative %
1
4.638
57.970
57.970
2
2.376
29.698
87.668
3
.606
4
.248
5
.106
6
.021
7
.006
8
.000
Rotated Component Matrix (Varimax)
Component
1
2
RGDPt-1
.980
-.189
REER t-1
-.222
.697
IIP t-1
.958
-.262
IR e t-1
-.489
.789
TOt-1
.986
-.087
TR t-1
-.569
.748
RWt-1
.078
.900
INF t-1
.965
-.142
