FOREIGN DIRECT
INVESTMENT IN
SINGAPORE
An empirical study of the effect of foreign direct
investment on GDP growth in Singapore
Name:
Niall Ruiter
Student number:
10399860
Education:
University of Amsterdam
Study programme: Economics
Supervisor:
Egle Jakucionyte
Abstract
This paper examines the effect of foreign direct investment on GDP growth in Singapore over the period 1970-2014. Using Ordinary Least Squares, a positive and significant impact of FDI on GDP growth is found. Interaction terms relating the level of trade and FDI and human capital and FDI are also positive and significant found which suggests the importance of trade and schooling in relation to foreign investments for Singapore. To control for possible endogeneity, an IV regression is applied. This gives no significant and clear results.
Table of contents
1. Introduction ... 2
2. Literature review ... 3
3. Economic history of Singapore ... 5
4. Empirical analysis ... 7
4.1 Data analysis ... 7
4.2 Model and hypothesis ... 10
4.3 Analysis and results ... 11
4.4 IV regression ... 16
4.5 IV results ... 16
5. Conclusion ... 18
6. References ... 19
1. Introduction
Since independence from Britain in 1959 Singapore has developed from a poor country with a low-skilled workforce to a modern industrial economy with a workforce whose educational level matches the highest in the Organisation for Economic Co-operation and Development (OECD) countries (OECD, 2011). According to a report from the World Trade Organization (WTO), Singapore is one of the most open economies in the world, with a strong export-orientated focus and a market-orientated economy (WTO, 2012).1 On a list ranking the economies in which it is easiest to do business
in, based on regulatory environment, the World Bank has Singapore ranking number one (World Bank, 2015). The International Monetary Fund stated that 70% of Singapore’s industries are export-orientated (IMF, 2015). In addition, Singapore was the fifth biggest receiver of Foreign Direct Investment (FDI) in 2014 and has experienced decades of relatively high GDP growth, even when compared to other fast growing-economies in Asia (World Bank, 2015).
Since Hymer (1960) and Bhagwati (1973) wrote on the subject of FDI, it has become a popular topic of debate among economists and politicians. FDI is believed to stimulate economic growth for a number of reasons. Direct effects of FDI include higher labour productivity, employment creation, higher income and eventually a potential increase in tax revenues (Ramirez, 2006). In addition to these direct effects, Ramirez (2006) argues that FDI causes indirect spill-over effects, such as enhanced competition between firms and transfer of technology and finally managerial expertise and learning-by-doing effects for the domestic firms (see also De Mello (1997)).
According to a study by Balasubramanyam et al. (1996), FDI is comparatively more effective in countries where economic policy allows market forces to allocate resources in the most efficient way.2
Studies by Borensztein et al. (1998) and De Mello (1999) confirm this for countries in South-East Asia. However, a study by Mah (2010) on the relationship between FDI and GDP growth in South Korea does not confirm this view. Yet, South Korea has a less open economy, measured by total trade over GDP, than Singapore and has enjoyed considerable lower FDI inflows over the past decades (World Bank, 2015). This study will therefore examine whether a positive relation does exist between FDI and GDP growth for Singapore, while adding more control variables and interaction terms in order to make an in-depth analysis following Mah (2010).
In this paper, the effect of FDI on real GDP growth in Singapore will be tested using Ordinary Least Squares (OLS). To control for the possible endogeneity of FDI, an Instrumental Variable (IV) regression is applied. To test the hypothesis that FDI influences GDP growth in Singapore, variables that were previously used in similar studies will be included in this study. These include gross domestic investment, openness to trade, government consumption, inflation and human capital.3 In order to
enable a more thorough study, different interaction terms will be added. To tackle the potential simultaneous causality problem, in which GDP growth might influence FDI, an IV regression will be done. The layout of this study will comprise further into four different sections, with the introduction in Section 1. Section 2 presents a short overview of the existing literature on FDI, including the results of these studies. Section 3 outlines a brief history of Singapore since independence in 1959. Section 4 consists of the empirical analysis in which the data is analysed, the regression is done and the results are discussed, with the addition of an IV-regression. The final section consists of a conclusion and a discussion of possible future developments.
1 In this paper, openness of an economy refers to the fact that there are high levels of imports and exports. It
does not imply anything on the political system such as capitalism, communism, and so forth.
2 These countries were mainly in the South-East Asia region including Singapore and various African countries.
These countries were compared to mainly Latin-American countries.
2. Literature review
This section consists of an overview of the existing literature on the relationship between FDI and economic growth.4 The existing literature on FDI consists mainly of empirical cross-country
analyses, with the majority of studies confirming a positive relationship between FDI and GDP growth. A vast amount of literature exists on the relationship between FDI and GDP growth. Most of these studies are empirical cross-country analyses, but there are also country-specific studies. In most of these studies, there exists a positive relationship between FDI and economic growth. In a meta-analysis of different studies on FDI, Ozturk (2007) concludes that the general consensus among these papers is that the effect of FDI on economic growth is strongly related to factors such as human capital, financial market regulation, degree of openness, tax incentives, financial stability and the trade regime. However, factors such as excessive competition by foreign firms, distortion of economic policies and too large reverse flows of profits and dividends can lead to an adverse effect of FDI on economic growth (Mah, 2010).
Country-specific studies using ordinary least squares (OLS) method by Blomström (1986), Kokko (1994) and Ramirez (2006) on the relationship between FDI and economic growth in Mexico all found a positive effect of FDI on economic growth. According to these studies, this positive effect was the result of intense competition by multinational enterprises (MNEs) and technology spill-overs. A similar relationship was found for China using multivariate causality tests (Liu, Burridge, & Sinclair, 2002). A study using OLS and error correction models on Nigerian economic growth gave the result that FDI in manufacturing firms may enhance economic growth (Akinlo, 2004). However, a study by Chakraborty et al. (2002) on the Indian economy, using a vector error correction model, finds that GDP growth causes FDI and not the reverse. Mah (2010), using Granger Causality tests, finds no significant long run effect of FDI on economic growth in South Korea. Furthermore, the effect is also insignificant when an interaction term between human capital and FDI is included, which contrasts with the results found other studies (see for example studies done by Li et al. (2005) and Borensztein et al. (1998)). Cross-country studies also find mixed results. For example, using dynamic panel model, De Mello (1999) finds that developing countries with a relatively strong environment for domestic investment can better host FDI and the effect depends on the characteristics of a country. This effect is especially strong in an economy which is technologically advanced. Other studies give similar results; the effect of FDI on economic growth depends on the level of human capital, for example Borensztein et al. (1998) and Li et al. (2005). Empirical research for a set of different countries finds a positive relation between FDI and economic growth in export promoting countries, see Balasubramanyam et al. (1996) and Basu et al. (2003). Similar findings were also reported for OECD countries (De Mello, 1999). Zhang (2001) and Alfaro et al. (2004), both using error correction models, conclude that FDI stimulates growth when a country has adopted a liberalized growth regime and has well-developed financial markets. Other cross-country studies confirm these results, see for example Nair-Reichart et al. (2001), De Gregorio (1992) and Reisen et al. (2001).5
However, Carkovic et al. (2002) do not find any positive effect of FDI on economic growth, after including lagged effects and country-specific effects, such as exchange rate volatility, changes in terms of trade, and so forth, using OLS and dynamic panel modelling. The inclusion of the degree of openness
4 This paper will not provide an in-depth analysis of the specific types of FDI or reasons for MNE to make this
decision. For more information on FDI, the reader is referred to Krugman et al. (2015).
does not change this conclusion (Carkovic & Levine, 2002), this in contrast to the results found by Balasubramanyam et al. (1996)
Mencinger (2003) studies the relationship between FDI and economic growth in eight relatively small EU candidates and found a negative causal effect of FDI on economic growth, using OLS and Granger Causality tests.6 In addition, a positive relation was found between FDI inflows and
account deficits (Mencinger, 2003). Mencinger (2003) argues that the negative effect of FDI on economic growth was a result of hasty privatization and excessive consumption of imports. Additional reasons for this negative relationship were negative spill-overs, caused by the fact that the MNE imported their materials from abroad rather than buying them in the country itself. Furthermore, the pressure on small firms in these countries did not lead to more innovation; the firms simply could not compete with the MNE’s and this resulted in the development of oligopolies and even monopolies. Finally, the human capital formation caused by FDI did not seem to have any significant positive effect on economic growth in these countries (Mencinger, 2003). De Mello (1999) also found a negative effect for 17 non-OECD countries, whilst finding a positive relationship for OECD countries. This could be explained by the fact that these countries may be less efficient in the use of the new technology brought in by FDI and country-specific factors, such as institutions, trade regimes, political risk and openness to trade (De Mello, 1999).
It can thus be concluded that the effect of FDI on GDP growth is country-specific for most studies and depends strongly on the level of openness of an economy. In most studies, a positive influence of FDI on economic growth in South-East Asia is found, whilst the influence of FDI is not significant in most Latin-American countries. Furthermore, if a strong domestic environment exists, with a certain level of human capital and technological environment, the effect of FDI on economic growth is significantly stronger. Both of these factors are present in Singapore, thus a positive impact of FDI on GDP growth is expected.
3. Economic history of Singapore
Singapore became independent from Britain in 1959, after which Singapore formed a part of Malaysia from which it separated in 1965. As part of the British Empire, Singapore had developed as an important seaport for ships en route between Britain, India and China. It also attracted an enormous number of immigrants, mainly workers from China and India. After independence, Singapore was a poor state without any real economy and had to import most of the basics such as food and energy. The first Prime Minister of Singapore, Lee Kuan Yew, had a two-fold objective: the development of a modern and well-functioning economy and creation of a national identity for the people of Singapore (OECD, 2011). The process of achieving both goals went through approximately three phases, which are discussed below.
The first phase began in 1960 and lasted roughly twenty years. During this period, the main goal of the government was to attract labour-intensive foreign firms, in particular manufacturing firms. The Republic of Singapore therefore concentrated on attracting FDI from the very beginning of its independence. The government decided on attracting manufacturing firms because Singapore had a growing population and high unemployment. By attracting labour-intensive foreign firms, it tackled this unemployment while workers gained knowledge and expertise from these firms. These knowledge spill-overs were a result of the fact that the foreign firms had superior knowledge compared to Singaporean firms. As domestic demand was limited, the government also decided to create an export based economy. However, as a result of regional competition in labour-intensive industry particularly from China and India, Singapore evolved into a more high-skilled economy (OECD, 2011) focussing less on the labour-intensive industry. This initiated the second phase, starting in 1979 until 1996.
In 1979, a different educational system was introduced. This system was built on the idea of creating an entire scale of different types of workers. This way Singapore could deliver workers at different educational levels. Singapore hoped to attract MNEs with a more desirable technological base, for example computer manufacturers. More technical universities were set up to maintain this attractiveness to MNEs (OECD, 2011). During this period, the economy relied on five main drivers of economic growth: high levels of trade by creating a regional trade hub; export-orientated manufacturing; special services for the global market, such as financial services and banking; shipping and the last driver was tourism (Lepoer, 1989). Meanwhile, the country had almost no tariffs and provided a high-quality infrastructure for MNEs. Until 1985, GDP grew by 8.5% every year. However, in 1985 a minor recession occurred. This was as a result of a slowdown in growth in industrial countries, less trade in Singapore because regional countries were trading directly with each other and import tariffs by regional countries. Singapore made a quick recovery after this recession, returning to GDP Growth of almost 11% in 1987 (NLB, 2014). Growth continued at high levels until the East Asian financial crisis struck South East Asia in 1997, leading to a recession in Singapore (Pilbeam, 2013). The main reasons for the recession in Singapore were a loss in competitiveness relative to China, hasty financial deregulation in Singapore and an appreciation of the US dollar. Singapore had a fixed parity with the US dollar, so when the US dollar appreciated, the real exchange rate of the Singaporean dollar also appreciated which lead to speculative attacks and a decrease in competitiveness (Pilbeam, 2013). The Asian crisis however did not cause a significant growth slowdown in Singapore, relative to the other countries in the region. GDP declined by 0.1%, whilst inflation even turned into deflation in 1998. In addition the current account remained with a surplus of 24% in 1998 and the government continued to have a fiscal surplus (Pilbeam, 2013).
However, following the financial crisis in South-East Asia, the government realized that the world economy was becoming a knowledge-based economy, rather than a labour economy. This knowledge-based economy required more a more innovative, creative and research-based
educational system. Government funding succeeded in attracting top scientists and MNEs, which in turn led to higher FDI. This system still continues to function to this day (OECD, 2011). Contrary to the notion of free trade, the government in Singapore has severely regulated the domestic economy while being ruled by the People’s Action Party since 1959. These regulations have had an overall positive impact. In addition to subsidies for education, the government provides funding for the health services, public transportation and housing. A savings fund, the Central Provident Fund, has also been established to provide for pensions (Lepoer, 1989). This intervention has resulted in decennia of political stability and an increase in general welfare, which has in turn led to both higher private investments and FDI over the years.
The growth of FDI can be readily seen in Figure 1. From 1970 until 1985, growth in FDI inflows was low. However, after their first post-independence recession in 1985, inflows of FDI started to accelerate. This acceleration could possibly indicate a break, considering FDI inflows were relatively stable and started to increase year-on-year after 1985. This increase in FDI was the result of overall investment in the region coming from Europe, Japan and the United States, while investors became less interested in investing in Latin-America due to the many recessions and the debt-crisis (Pilbeam, 2013). The decrease in FDI in 2002 was mainly caused by the SARS-epidemic, global growth slowdown due to the war in Iraq, increased competitiveness of China and India and increased terrorism threat. After 2002, FDI increased from approximately 6.4 billion dollars in 2002 to 47.7 billion dollars in 2007 which is an increase of almost 750%. During this period, MNEs especially wanted to reach new markets in countries such as China and India, using the good infrastructure and the market-friendly economy of Singapore (UNCTAD, 2006). This type of FDI were primarily mergers and acquisitions (UNCTAD, 2006). Furthermore, China started to become a major investor in Singapore, besides the growing intraregional FDI (UNCTAD, 2006). As can be seen in the graph, FDI dropped after that. This was a result of the global financial crisis. After the financial crisis, FDI inflows started to increase significantly again. This was again due to a fast recovery of China and India (which are major investors in Singapore) and other regional players. As can be seen, FDI hit 67 billion dollars in 2015, making Singapore one of the biggest recipients of FDI in the world. As regarding future developments, Singapore and the European Union have signed a free trade agreement, which is most likely give a boost to FDI in Singapore (European Commission, 2015).
Figure 1. Singapore: FDI inflows in billions of US dollars, 1970-2015
Source: World Bank 2015 (available at http://databank.worldbank.org/data, retrieved 14 December 2015). 0 10 20 30 40 50 60 70 80 1970 1973 1976 1979 1982 1985 1988 1991 1994 1997 2000 2003 2006 2009 2012 2015
4. Empirical analysis
4.1 Data analysis
As stated in the introduction, the main variable of interest is FDI. Other variables included are real GDP growth as the dependent variable and FDI, domestic investment (GDI), openness to trade (Trade), government consumption (Government), inflation (Inflation) and mean years of schooling (Schooling) as independent variables. FDI, GDI, Government and Trade are measured as a share of GDP, thus are all ratios. This holds for the whole analysis from here on. All data is taken from the World Bank online database. In this subsection, the expected relationships between the variables are discussed, after which the variables are analysed in more detail. The relationships between the variables matter, as a variable that correlates both with FDI and GDP growth has to be added to the regression in order to avoid omitted variable bias. If these variables are not added, the coefficients might be biased and inconsistent (Stock & Watson, 2015). This implies that the OLS-regression gives incorrect results.
The variable FDI is included because it is the variable that is of main interest for this paper. Its effects on long term growth are primarily through externalities and productivity spill-overs (De Mello, 1997). This implies therefore the introduction of technologies, managerial skills and new capital goods (Ozturk, 2007). FDI as a factor of growth is not recognized by the neo-classical growth model of Solow because FDI is not incorporated as a variable (Solow, 1956). In this model, long-run economic growth is only possible through technological progress and population growth, which are both exogenous in the model of Solow (1956). However, in endogenous growth models, FDI can be included as a variable that has an effect on economic growth. Thus the regression in part 4.2 can be seen as derived from the endogenous growth model.
Including domestic investment in the regression is obvious, since GDI and FDI together account for total investments in an economy. This is an important part of aggregate demand, as can been seen by Keynesian-type aggregate demand functions. Total demand influences total output because an increase in demand increases consumption and production, which again increases total output. Whether the correlation between GDI and FDI is positive or negative is less obvious. On the one hand, FDI could complement GDI given that a determinant for foreign investors to invest in a foreign country is influenced by the degree of existing domestic endowments (De Mello, 1997). This would imply a positive relationship. However, FDI could also substitute domestic investments. This implies for example an increase in competitiveness as the foreign investors force the less efficient domestic firms out of the market. The latter is seen in more technological developed countries such as Singapore (De Mello, 1997). Therefore, the expected relationship between FDI and GDI is negative.
Many other papers which study the effect of FDI on economic growth have included some measure of openness using a certain variable and the effect of this measure of openness on economic growth.7 Examples of variables that control for openness are total exports, or total exports plus
imports, and so forth. There are several reasons for including total trade into the regression. First, an open economy, ceteris paribus, is more likely to result in dynamic learning and more technological innovation (Salvatore & Hatcher, 1991). Furthermore, it is argued by Salvatore et al. (1991) that an attractive regulatory environment associated with an economy which is export-orientated, leads to the exploitation of economies of scale and higher productivity. According to the World Bank, Singapore is one of the most attractive economies to do business in (World Bank, 2015). Openness is an economic factor that demonstrates the attractiveness of an economy (De Mello, 1997). The correlation between FDI and Trade is expected to be positive, as foreign firms would be more willing to invest when a country is characterized by an open and liberalized economy (Bhagwati, 1973).
The variable government expenditure has been included in several other studies, for example
De Gregorio (1992) and Ramirez (2006). The theoretical argument for including government expenditure is that expenditures by the government in sectors such as housing and other public services account for a significant part of aggregate expenditure. The relationship between FDI and government expenditure is expected to be negative, as government expenditure could crowd-out investment made by foreign investors (De Gregorio, 1992).
Inflation serves as a measure of macroeconomic stability (De Mello, 1997). Inflation increases the cost of capital and thus creates a negative effect on the profitability of FDI. Thus the expectation is that FDI and inflation have a negative correlation.
The final variable is schooling, defined as mean years of schooling of residents of 25 and over. Its main features can be found in the Appendix. This variable serves as a measure of human capital. The importance of adding a variable that relates to human capital is been stressed in most studies relating FDI to economic growth, for example Balasubramanyam et al. (1996) and De Mello (1997). Borensztein et al. (1998) and Li et al. (2005) argue that the level of human capital determines whether or not a country is successful in adopting technology and the necessary skills demanded by foreign firms. Thus the correlation between FDI and schooling is expected to be positive.
Figure 2 and Table 5 present the variables plotted in a graph and summary statistics.8 Figure 2
shows great variation in variables over time. The rate of GDP growth fluctuates around 7% for Singapore. It shows some relatively low growth percentages and recessions during turbulent times, such as the oil crises, the crisis in South-East Asia and the financial crisis. FDI has an upward trend, though fluctuating considerably, as it can be seen in Figure 2. Until 2003, FDI fluctuated between approximately 4-13% over the years. After 2003, FDI inflows started to increase considerably. The maximum of almost 27% was reached just before the financial crisis in 2007, after which it declined to 6%. Domestic investment has a higher mean value than FDI, as can be seen in Table 5. The reason is that GDI fluctuated around 30-40% during the period 1970-2000, but declines since. The ratios of FDI
8 The table of the summary statistics (Table 5) can be found in the Appendix.
Figure 2. Plot of main variables.
Main variables (in %)
Notes:
1. These variables are included as variable/GDP. For example: GDI¹= GDI/GDP.
2. Other aspects of these variables, including Trade and Schooling, can be found in the Appendix.
0 10 20 30 40 50 1970 1980 1990 2000 2010 Date FDI¹ GDPgrowth GDI¹ Government¹ Inflation
and GDI converge towards each other, with FDI moving in an upward trend and GDI in a downward trend. The variable Trade is at least ten times as large as the variables FDI, GDI and Government, which is expected for Singapore.9 This shows how important exports and imports are for Singapore as a
trading hub in South-East Asia. Trade is also very volatile. This in contrast to government consumption, which does not fluctuate greatly and has the lowest standard deviation (see Table 5). Inflation was very stable, except for two periods. Inflation had a value of 19% in 1973 and 22% in 1974, which is considerably higher than in other years. This increase in inflation was a result of quadrupling oil prices, which led to the first oil crisis. Oil imports made up a large share of Singapore’s imports, thus an increase in oil prices caused higher import costs (MAS, 2014). To battle this surge in inflation, a restrictive monetary policy was implemented by increasing the interest rates. While this also affected economic growth it did not result in a recession (MAS, 2014). The second period inflation peaked, up to 8%, was during the second oil crisis. Apart from these peaks inflation fluctuates in the range of 0-5% up to the present. There have been four periods of deflation since 1970. These deflationary periods coincided with recessions. A possible explanation is that during a recession consumers could choose to delay their consumption while causing firms to invest less as a result of the recession. This decrease in consumption and investment could cause deflation, as the demand for goods will decrease and thus less money will be spent. Finally, the variable Schooling has been steadily increasing over the years, from 4.7 years in 1980 rising to 10.6 years in 2014. As a result of this steady increase, Singapore ranked number eleven on the Human Development Index in 2014 (United Nations, 2015). This demonstrates that the government of Singapore has been quite successful in implementing its policy of improving educational standards for its citizens.
Table 1 exhibits the correlation matrix of the main variables. As expected, the correlation between FDI and GDI is negative. This could be the result of the ‘crowding-out’ effects. This implies that investments by MNEs substitute the investments made by domestic private firms in Singapore. However, this negative correlation could also be the result of FDI flows from Singapore into foreign countries. This latter form of FDI, known as FDI outflows, has increased over the years and Singapore had worldwide the twelfth largest outflow of FDI in 2014 (UNCTAD, 2015). FDI outflows are not a part of GDI, so that an increase in these investments in foreign countries could mean less domestic investments. The correlation between FDI and openness to trade is strongly positive, which is a confirmation of the result of most studies and a confirmation of Bhagwati (1973) that firms are more willing to invest when a country is open.10 The correlation between government expenditure and FDI
is likewise negative. As is argued, this may be because expenditures crowds out investment made by
9 The variable Trade has not been plotted for the reason that it is of considerably higher value than the other
variables. Its main features and that of schooling can be found in the Appendix.
10 Balasubramanyam et al. (1996); Borensztein et al. (1998) and Carkovic et al. (2002).
Table 1. Correlation matrix.
Correlation matrix
GDPgrowth FDI GDI Trade Gov. Infl. Sch.
GDPgrowth 1.0000 FDI 0.1960 1.000 GDI 0.1579 -0.5668 1.000 Trade 0.1211 0.4688 -0.5684 1.000 Government -0.4201 -0.1474 -0.0401 -0.0106 1.000 Inflation 0.2796 -0.0053 0.2246 0.3912 -0.4013 1.000 Schooling -0.2253 0.6419 -0.7769 0.5016 -0.1301 -0.0069 1.000
foreign firms. For example, if the government, as mentioned above, invests heavily in public goods such as housing, foreign firms do not have investment opportunities in that sector. The expectation of a negative relationship between FDI and inflation is also confirmed. During both oil crises, inflation increased while FDI decreased, often with a lag of one year. Thus investors would indeed be unwilling to invest in Singapore when inflation is high. The correlation between FDI and schooling is relatively high. This could be a sign that a more highly educated workforce attracts higher investments by MNEs. Finally, the correlation between FDI and Schooling is strongly positive. This possibly implies that FDI and Schooling are becoming more and more important for Singapore, as both are in an upward trend.
4.2 Model and hypothesis
In this subsection, the regression model is set up, the hypothesis is formulated and the expected signs of the coefficients are discussed. The regression model below is the main model that will be analysed, not including any lagged terms nor any interaction terms. Thus, this study will examine the effect of FDI on GDP growth by estimating the following model:
𝑌𝑡 = 𝛽0+ 𝛽1 𝐹𝐷𝐼𝑡+ 𝛽2 𝐺𝐷𝐼𝑡+ 𝛽3 𝑇𝑟𝑎𝑑𝑒𝑡+ 𝛽4 𝐺𝑜𝑣𝑡+ 𝛽5 𝐼𝑛𝑓𝑙𝑎𝑡𝑖𝑜𝑛𝑡+ 𝛽6 𝑆𝑐ℎ𝑡+ 𝜀𝑡 In this equation, 𝑌𝑡 stands for the annual real GDP growth of Singapore in period 𝑡, 𝐹𝐷𝐼𝑡 stands for FDI as a share of GDP in period 𝑡, 𝐺𝐷𝐼𝑡 stands for total domestic investments as a share of GDP in period t, 𝑇𝑟𝑎𝑑𝑒𝑡 stands for total exports plus imports as a share of GDP in period 𝑡, 𝐺𝑜𝑣𝑡 stands for total government consumption as a share of GDP in period 𝑡, 𝐼𝑛𝑓𝑙𝑎𝑡𝑖𝑜𝑛𝑡 stands for inflation in period 𝑡,
𝑆𝑐ℎ𝑡 stands for mean years of schooling in period 𝑡 and finally, 𝜀𝑡 (the error term) takes on all the other
factors that are not in the equation that affect economic growth in period 𝑡.
My hypothesis is the following: the estimated coefficient in front of FDI is expected to be positive and statistically significant, thus implying that it is expected that FDI has a positive impact on GDP growth in Singapore. This expectation is based on arguments discussed above, with the main argument that the introduction of technologies, managerial skills and new capital goods, which are a common feature of FDI, generate positive spill-overs which lead to long term GDP growth. Thus the externalities of FDI are considered to be the most growth-enhancing, as they create the spill-overs which the domestic investments might lack (De Mello, 1997). A theoretical argument made in some papers is that FDI could also have a negative impact on GDP growth, in which reverse flows such as dividends or profits outweigh the inflows.11 However, FDI is defined as net inflows such that the total
inflows of FDI is corrected for outflows such as dividends or repatriation of profits. Another way in which FDI could have a negative impact on GDP growth is through too severe competition, distortions of economic policies and possible social and cultural norms which do not coincide with the norms in the recipient country (Zhang, 2001). However, those effects should not be of any significance for Singapore as its economic policies are generally designed to attract foreign investors and Singapore heavily relies on these foreign investors for economic growth. Furthermore, the model does not focus on these possible distortions as a good approximation of these distortions is hard to find.
The coefficient of the GDI effect is expected to be positive, as investments determine a large share of GDP growth (De Long & Summers, 1990). In conformity with Bhagwati (1973) and many other studies which test the effect of openness on economic growth, the coefficient in front of Trade is expected to be positive.12 An interaction term with FDI and Trade is also expected to be positive,
confirming the study of Balasubramanyam et al. (1996). Government expenditures also affect GDP growth. Aschauer (1990) and Barro (1988) argue that government spending can affect economic growth positively, if it is spent in a productive manner. Government spending necessarily must be
11 Gregorio (1992), Ramirez (2006) and Mah (2010) use these arguments in their papers.
financed with taxes or via lending, which distort choices and thus effect growth negatively.13 However,
if the expenditures are productive, for example the expenditures are used to invest in education, infrastructure or complementing private investment, these government expenditures stimulate economic growth (Aschauer, 1990). Therefore, the coefficient in front of Government is expected to be positive. There are many papers on the effect of inflation on GDP growth. Some papers conclude that inflation never influences GDP growth positively, for example Andrés et al. (1999) and Barro (1995), whilst other papers find a positive effect of inflation on GDP growth only when inflation is low, see Fischer (1993). Furthermore, most central banks target a small positive inflation rate. The European Central Bank and the Federal Reserve are examples that target a moderate annual inflation rate (Mishkin, F., S., Matthews, K., & Giuliodori, M., 2016). Thus the expected sign of the coefficient in front of inflation is indeterminate. The coefficient in front of schooling is expected to be positive. As discussed, the reason for this expectation is that human capital is a significant factor influencing GDP growth. An interaction term between FDI and schooling is also expected to be positive, as human capital and FDI complement each other in influencing GDP growth. Thus by including the discussed variables, which correlate both with FDI and GDP growth, the possibility of omitted variable bias is begin reduced.
4.3 Analysis and results
In this subsection, the model will be estimated. First the variables will be checked for stationarity, after which the regression will be adjusted if the variables are non-stationary. Then various models will be estimated and checked for heteroskedasticity and autocorrelation. All the tests and estimates are made using Stata 13.
Before estimating the model that is outlined in the previous section, the variables must be checked for stationarity. A variable is stationary if the probability distribution of the variable does not change over time, the mean and variance are thus constant over time (Stock & Watson, 2015). Often time series on macroeconomic data tend to exhibit a stochastic time trend, which implies that the variables are nonstationary. OLS-based tests, for example a hypothesis test, can be misleading if applied for nonstationary variables (Stock & Watson, 2015). If the variable has a stochastic trend, which implies having a unit root, then the first difference of this variable might not have a stochastic trend. In other words, by differencing a variable that is nonstationary, the variable could change from nonstationary to stationary (Stock & Watson, 2015).14 After this correction, the regression can be
modified, after which the standard OLS-based tests can be applied. First, the number of lags will be determined. The lags are past values of the variable, for example the first lag of 𝐹𝐷𝐼𝑡 is 𝐹𝐷𝐼𝑡−1 (Stock & Watson, 2015). The most commonly used are the Akaike information criterion (AIC) and the Bayes information criterion (BIC or SBIC). AIC often overestimates the number of lags, whilst BIC might give a model with too few lags (Stock & Watson, 2015). These likelihood statistics determine the optimal number of lags. These two information criteria determine how large the increase in the R² must be to justify using an additional lag (Stock & Watson, 2015). The model which shows the lowest value of the AIC or BIC is then chosen. In other words, the optimal lag length is given by the value that has the lowest value of the AIC or BIC. The result of the tests can be seen in Table 2.
13 Lending must be repaid in future periods, thus this also distorts choices.
14 Differencing a nonstationary variable does not necessarily make it stationary, although this is the case for this
For every variable, the optimal lag length is given by both the AIC and BIC. For example, for GDI the AIC and BIC are calculated for one, two, three or four lags. This gives four different values of the AIC and four different values for the BIC. Then the minimum value of AIC is chosen, which is 5.37564 and this correspondents to one lag. This is also done for BIC, which also gives one lag. This procedure is then done for each variable, giving the optimal lag length for each variable. This lag length is the same for all, except for FDI. As can be seen, the AIC gives as optimal lag length three lags, whilst the BIC gives one lag length as optimal. After this procedure, the unit root test can be done for each variable on the level form and the first difference, given the lag lengths determined above.
As stated previously, to be able to do OLS-based tests, the variables in the regression must be tested for stationarity. Testing for stationarity will be done with the Augmented Dickey-Fuller (ADF) test for a unit autoregressive root. Multiple studies have shown that the ADF test works better with too many
Table 2. Lag length criteria
Optimal lag length
Variable AIC AIC lag length BIC BIC lag length
GDPgrowth 5.5591 0 5.6009 0 FDI1 6.10178 3 6.24391 1 GDI 5.37564 1 5.45923 1 Trade 9.27614 1 9.35973 1 Government 2.3146 1 2.39818 1 Inflation 3.92077 6 4.23184 6 Schooling -1.38258 2 -1.23985 2 Notes:
1. For this variable the AIC and BIC give a different optimal lag length.
Table 3. Augmented Dickey-Fuller test results for stationarity1.
Unit root tests
Variable Level form First difference
GDPgrowth -5.143*** -8.675*** FDI -1.106 -5.223*** GDI -1.206 -4.487*** Trade -2.440 -4.642*** Government -3.118 -4.475*** Inflation2 -1.786 -2.818* Schooling -0.386 -0.004*** Notes:
1. Under the null hypothesis, the variable is non-stationary. Under the alternative the variable is stationary.
2. Inflation is significant in first differenced-form at the 10% level. However, it is statistically significant at the 1% level in second-differenced-form. Its test-statistic is given by -4.762, with a p-value corresponding to 0.0001 in second-differenced form.
*. Statistically significant at 10% level. **. Statistically significant at 5% level. ***. Statistically significant at 1% level.
lags than too few.15 For this reason, the lag length for each variable is chosen by the AIC information
criteria. Under the null hypothesis of the ADF test, the variable has a stochastic trend, and is thus non-stationary. Under the alternative hypothesis the variable is stationary and has no stochastic trend. This test is done for the level form of the variable and the first difference. The reason is that if the level form is nonstationary, then the first difference might be stationary, after which further testing can be done. The outcomes of the ADF test are presented in Table 3.
The results in Table 3 show that the hypothesis that each variable is non-stationary in level form cannot be rejected except for GDP growth. However, for the first differences the null hypothesis of non-stationarity is rejected for all variables. This suggests that the all but GDP growth follow a stochastic trend in level form which implies that GDP growth is integrated of order zero, whilst all the other variables are integrated of order one. This in turn implies that the main regression has to be modified to estimate with OLS. The dependent variable, GDP growth, remains unchanged and is thus still in a level form. The other variables, however, are first-differenced in order to correct for the stochastic trend. The new regression is as follows:
𝑌𝑡 = 𝛽0+ 𝛽1 ∆𝐹𝐷𝐼𝑡+ 𝛽2 ∆𝐺𝐷𝐼𝑡+ 𝛽3 ∆𝑇𝑟𝑎𝑑𝑒𝑡+ 𝛽4 ∆𝐺𝑜𝑣𝑡+ 𝛽5 ∆𝐼𝑛𝑓𝑙𝑎𝑡𝑖𝑜𝑛𝑡+ 𝛽6 ∆𝑆𝑐ℎ𝑡+ 𝜀𝑡 All the variables are defined in the same way as in the original regression. The ∆ stands for difference operator. The first difference of, for example 𝐹𝐷𝐼𝑡, is thus ∆𝐹𝐷𝐼𝑡 = 𝐹𝐷𝐼𝑡− 𝐹𝐷𝐼𝑡−1. This is thus the difference between the value of FDI in period 𝑡 and FDI in period 𝑡 − 1. This equation, including interaction terms, is tested in Table 4.
Furthermore, several tests on heteroskedasticity and serial correlation have been computed. The results can be seen in the tables in the Appendix.16 To test for heteroskedasticity, three tests have
been computed for each of the seven regressions, the results can be seen in Table 6. These tests are a test for autoregressive conditional heteroskedasticity (ARCH), the Breusch-Pegan Test for Heteroskedasticity, and finally White’s General Test for Heteroskedasticity. White’s test is a special case of the Breusch-Pegan test, with relaxation of the assumption of normally distributed errors (White, 1980). As can be seen, there is not enough evidence to conclude that the error terms are heteroskedastic, i.e. it can be assumed that the error terms are homoskedastic and thus have a constant variance. All three tests confirm this result. To tests for autocorrelation, two different tests have been applied and the results can be seen in Table 7. The first test for autocorrelation is Durbin’s alternative test for autocorrelation (King, 1981). This test is more general than the conventional Durbin-Watson test and also has a small-sample correction, which is useful as all regressions have relatively few observations. Furthermore, the Breusch-Godfrey test is also computed. This test is statistically more powerful than the Durbin-Watson test (Baum, 2006). These two tests are carried out for each regression under the null hypothesis that the error terms are not serial correlated. This hypothesis cannot be rejected for any equation, thus there is not enough evidence to conclude that the error terms are serial correlated. Summarizing, there is not enough evidence that the models exhibit any serial correlation (sometimes referred to as autocorrelation) nor is there any statistical evidence that the error terms are heteroskedastic. This implies that the Gauss-Markov theorem can be applied and the OLS-estimators are thus BLUE (Best Linear conditionally Unbiased Estimators) (Stock & Watson, 2015). In short, the Gauss-Markov theorem is conditional on the Gauss-Markov conditions. These conditions state that the OLS-estimators are BLUE if the following conditions hold: the error term has a mean of zero, given the independent variables; the variance of the error term is constant, i.e. the error term is homoskedastic and finally, the error terms are uncorrelated, i.e. the
15 See the studies done by Stock (1994) and Haldrup et al. (2006).
16 See Table 6 and Table 7 in the Appendix for results of the tests for heteroskedasticity and serial correlation,
error terms exhibit no serial correlation. These Gauss-Markov conditions are based on the least square assumptions plus homoscedasticity (Stock & Watson, 2015). As a result of meeting these conditions, it can be stated that the OLS-estimators that are estimated in the regressions above are unbiased, consistent and efficient.
Equation (1) presents the main regression. Noticeable is that FDI has a positive effect on GDP growth and this effect is highly significant, i.e. significant at the 1% level. Based on this regression, it can be stated that an increase of FDI over GDP of 1 percentage point increases GDP growth by approximately 0.278percentage point, holding other factors constant.17 Thus, the hypothesis that FDI
has a positive and significant effect on GDP growth, is confirmed by equation (1). In other words, the null hypothesis that FDI has no significant impact on GDP growth can be rejected at the 1% level.18 The
coefficient in front of Trade is small and negative, which is in contrary to the expectation, although no conclusion can be made as it is insignificant. The coefficients for government expenditure and domestic investment are likewise insignificant. Inflation and schooling, however, are significant. Both have a positive impact on GDP growth. For inflation, this means that an increase of inflation with 1 percentage
17 Holding all other factors constant, i.e. the ceteris paribus condition, implies that the other variables in the
regression being analysed do not change when looking at, for example, the impact of FDI on GDP growth. This holds for the rest of the analysis of the coefficients.
18 This result is also found in similar studies. Ramirez (2006) found an effect of 0.67 for Mexico and Gregorio
(1992) found an effect of FDI on GDP growth of 0.55 using panel data. Table 4. OLS regression. Dependent variable is GDPgrowtht , 1970-2014.
OLS Regression
Variables (1) (2) (3) (4) (5) (6) ΔFDIt 0.278 (3.25)*** 0.170 (1.86)* -0.026 (-0.16) 0.024 (0.14) 0.309 (3.41)*** 0.186 (1.96)* ΔGDIt 0.207 (1.26) 0.316 (1.99)* 0.315 (1.92)* 0.348 (2.15)** 0.237 (1.42) 0.437 (2.87)*** ΔTradet -0.020 (-0.69) -0.035 (-1.28) -0.002 (-0.06) -0.020 (-0.65) -0.013 (-0.44) -0.005 (-0.23) ΔGovt. -1.060 (-1.64) -0.928 (-1.56) -1.268 (-2.06)* -1.085 (-1.76)* -1.061 (-1.65) -1.416 (-2.11)** ΔInflation 1.093 (2.87)*** 1.084 (3.09)*** 0.866 (2.31)** 0.955 (2.57)** 0.949 (2.34)** 0.157 (1.30) ΔScht. 9.482 (2.24)** 9.750 (2.49)** 6.223 (1.45) 7.794 (1.79)* 9.212 (2.17)** - ΔFDIt x ΔTradet - 0.007 (2.30)** - 0.005 (1.42) - - ΔFDIt x ΔScht. - - 1.348 (2.04)* 0.7801 (1.03) - - GDPgrowtht-1 - - - - 0.138 (1.03) - Constant 4.886 (5.65)*** 5.059 (6.32)*** 5.737 (6.27)*** 5.504 (6.05)*** 4.018 (3.32)*** 7.180 (14.77)*** R² 0.68 0.74 0.73 0.75 0.69 0.43 Adj. R² 0.60 0.66 0.65 0.66 0.60 0.35 F-statistic 8.53*** 6.65*** 8.86*** 8.35*** 7.48*** 5.70*** Notes:T-ratios are in parenthesis
*. Statistically significant at 10% level. **. Statistically significant at 5% level. ***. Statistically significant at 1% level.
point leads to an increase of GDP growth with approximately 1.1 percentage point. According to this result, high inflation stimulates GDP growth for Singapore, which is contrary to what would be expected. Schooling, as expected, also has a positive and significant impact on GDP growth. Increasing the mean years of schooling by one year increases GDP growth by almost 9.5 percentage point. The R², which measures the fraction of the variance that is explained by the model, gives 0.68. in other words, a large fraction of the variance in GDP growth is explained by the explanatory variables.
In equations (2)-(4), different interaction terms have been added to the regression. In equation (2) an interaction term between FDI and Trade has been added, which is positive and significant at the 5% level. This is a conclusion found in most studies which look at the interaction effect of FDI and a measure of openness. Looking at the marginal effect of FDI on growth, which is the first derivative of the equation with respect to FDI, gives: 𝜕(∆𝐹𝐷𝐼𝜕𝑌𝑡
𝑡)= 0.17 + 0.007 × (∆𝑇𝑟𝑎𝑑𝑒𝑡), and thus ∆𝑌𝑡 ≈
(0.17 + 0.007 × ∆𝑇𝑟𝑎𝑑𝑒𝑡) × ∆(∆𝐹𝐷𝐼𝑡). Thus the effect of FDI on economic growth increases as the level of openness also increases. The R² also increases from 0.68 to 0.74, meaning that this new equation fits the data set better. However, the R² is not a perfect measure because it almost always increases (never decreases) as a new variable is added. The adjusted R² corrects for this effect and it increases from 0.60 in (1) to 0.66 in (2), thus indicating a better fit. The coefficient in front of FDI is significant at the 10% in (2), as can be seen by the smaller t-ratio. This could be a result of multicollinearity between the variable FDI and the interaction term between FDI and Trade, which increases the standard error on the coefficient of FDI. In equation (3), an interaction term between FDI and schooling is added. The coefficient in front of this interaction term is significant at the 10% level, while the coefficient in front of FDI is negative and no longer significant. Looking at the marginal effect of FDI in (3), it can be stated that FDI increases as schooling increases. The R² and adjusted R² are slightly lower than in (2). In equation (4) both interaction terms have been added. However, neither of them is significant, nor is the coefficient in front of FDI significant. Furthermore, R² increases marginally compared to (2), while the adjusted R² does not increase. This implies that (4) adds nothing new to regressions (1)-(3). Furthermore, equation (5) is equivalent to (1) including a lagged term of GDP growth. However, this coefficient is not significant at any level below 10% and the adjusted R² is the same as in (1). Finally, in equation (6) the variable schooling is dropped in order to check if this affects the result, as the variable Schooling is only a proxy for human capital. The relatively large decrease in the R² and adjusted R² seems to be a sign that the variable Schooling is important in this regression. This can also be seen from the fact that it is significant at the 5% level in (1) and (2). From all the regressions, only two interaction terms helped to explain more of the effect of FDI on GDP growth. Thus the level of openness and schooling are an important determination in helping explain why FDI has a positive impact on GDP growth.
Comparing equation (2) and (3), it is noticed that (2) has a higher R² and adjusted R², but this difference is only marginal. In regression (2) the coefficients in front of Trade and Government are not significant at any level, while GDI is only significant at the 10% level. For Trade, this could explain the fact that a certain level of FDI is needed in order for the level of openness to be effective. This can be seen by looking at the marginal effect of Trade on GDP growth: ∆𝑌𝑡 ≈ (−0.035 + 0.007 × ∆𝐹𝐷𝐼𝑡) × ∆(∆𝑇𝑟𝑎𝑑𝑒𝑡). This implies that when FDI over GDP is higher than 5%, the effect of Trade on GDP growth is positive.
However, the model could be subject to endogeneity, meaning that GDP growth might influence FDI. This reverse causality could lead to a bias in the coefficients and is thus called reverse causality bias. In order to tackle this problem, an instrumental variables (IV) regression could be applied, as it is discussed in the final subsection.
4.4 IV regression
To apply IV, viable instruments are needed. As Stock et al. (2015) point out, an instrument must be exogenous and relevant. Exogenous instrument means that the instrument must not correlate with the error term of the original regression. A relevant instrument is an instrument that correlates with the endogenous variable, in this case FDI. Papers on FDI recognize the difficulty finding viable instruments, see for example Borensztein et al. (1998), Li et al. (2005) and Carkovic et al. (2002). In these papers, an IV regression is applied using, among other instruments, the lagged terms of the endogenous variable as instruments. However, for lagged values to be a viable instrument, the assumption must be made that lagged values of the endogenous variable do not influence the current value of the dependent variable, in this case GDP growth (Angrist & Krueger, 2001). In a paper by Vahter (2010), analysing FDI in Estonia, FDI inflows of similar and nearby countries are used as an instrument. It is argued that these FDI inflows predict inflows to Estonia, but are not directly correlated with factors influencing economic growth in Estonia. The problem often is, however, that these instruments are weak. He et al. (2009) use weather as an instrument. It is argued that this does not influence economic growth and does have a significant impact on FDI. However, it is also acknowledged that weather is a weak instrument and only influences economic growth in countries that rely heavily on the agricultural sector. This problem is also recognized by Nair-Reichert et al. (2001), with the argument that a good instrument for FDI and economic growth is very hard to find. To test if the instruments proposed by Borensztein et al. (1998) and Vahter (2010) are indeed not viable, an IV regression is computed using their choice of instruments.
4.5 IV results
In Table 8 in the Appendix, three IV regression have been computed. In equation (I), lagged values of FDI have been used as instruments.19 To test the exogeneity of the instruments, the Hansen
test has been applied with the corresponding J-statistic.20 This statistic is approximately 𝜒
𝑚−𝑘2 distributed, where m is the number of instruments and k is the endogenous variable. As can been seen, the null hypothesis of exogeneity cannot be rejected in regression (I). To test if the instruments are weak, the F-test is calculated in the first stage of TSLS. If this F-statistic is greater than 10, then the instruments are relevant.21 The value of the F-statistic is less than 10 in (I), thus the lagged terms of
FDI which were used as instruments, are weak. This implies that the instruments do not correlate sufficiently with FDI and that the coefficients in the corresponding regression yield no or inconsistent results (Stock & Watson, 2015). This can also be seen from the signs of the coefficients and their significance in (I): FDI is positive and significant at the 10% level, while Schooling is negative and significant 5% level. This implies that higher mean years of schooling of the workforce would reduce GDP growth. All the other coefficients are insignificant in regression (I).
In regression (II), the FDI flows of countries close to Singapore have been used as an instrument, as proposed by Vahter (2010). These countries are Thailand, Vietnam, Korea and the Philippines. This regression shows similarities to (I), as the instruments are exogenous but weak in (II). The coefficients on all variables are insignificant, except for government consumption and schooling. Government consumption is positive and significant at the 10% level, while schooling is negative at the 5% level, which is again in contrast to the results obtained using OLS. In the final IV regression, regression (III), both the instruments of (I) and (II) are combined as a last resort. Although FDI is significant at the 5% level, the regression exhibits the same complications as (I) and (II): weakness of
19 The instruments used were the three first lagged values of FDI. Using a different set of lagged values did not
change the outcome significantly.
20 Sometimes referred to as the Sargan-test or the J-test (Stock & Watson, 2015). The value of the J-statistic is
calculated based on the corresponding F-Statistic: 𝐽 = 𝑚𝐹~𝜒𝑚−𝑘2 .
the instruments. This can be seen by the value of the F-statistic, which should be larger than 10 for a strong instrument, but has a value of 0.47 in (III). This weakness implies that the coefficients are inconsistent and possible biased (Stock & Watson, 2015). What can be concluded as a result of this attempted IV regression is that a good and useable instrument is difficult to find, as has been previously acknowledged. Although the null hypothesis of exogeneity for the used instruments cannot be rejected, the instruments cannot be said to be relevant, i.e. the instruments are weak and thus do not correlate sufficiently with FDI. This weakness of the instruments applies for all three regressions in Table 8. The signs and the significance of the coefficients is also in contrast to the results obtained by the OLS-regressions, implying a possible bias as a result of the weakness of instruments.
5. Conclusion
There are several reasons why FDI could have a positive effect on GDP growth. This is mainly through more advanced technology and increased management expertise (Borensztein, De Gregorio, & Lee, 1998). These types of spill-overs have led many economists to believe that FDI can influence on economic growth. This relationship has therefore been extensively researched and is the subject of many papers and books. Human capital and an open economy have been recognized as being important determinants of the willingness of foreign investors to invest and this has also been proved empirically in many papers. With an ever increasing focus on the importance of schooling and continuing concentration on remaining an important regional trading hub, using its location as a strength, Singapore has put a strong focus on those elements that are recognised as being important in attracting FDI.
In this paper it has been attempted to analyse the effect of foreign direct investment on GDP growth in Singapore. This empirical analysis has been performed by running an OLS-regression. Several tests have also been applied to check the robustness of these OLS-based inferences. Based on the outcome of the regressions, the hypothesis that FDI has a significant and positive impact on GDP growth cannot be rejected. Furthermore, this study has shown that the effect of FDI depends on the level of openness and that a certain level of human capital is needed for FDI to have a positive effect on GDP growth. There is a significant positive interaction between FDI and the level of openness. Thus FDI does not only influence GDP growth on itself, it also does so indirectly via interaction terms. These interaction terms confirm the spill overs generated by FDI.
This analysis has to be interpreted with caution, however. For example, the data set is quite small, as schooling is only measured from 1983 on. This could give a higher R² than justified (Stock & Watson, 2015). Furthermore, the variable Schooling is not a perfect measure for measuring human capital, as human capital is so much more than just mean years of schooling. For example, the mean years of schooling does not take into account the level of schooling, which is divided into three levels in Singapore (OECD, 2011). There is also a possibility of omitted variable bias, in which a variable that correlates with FDI and GDP growth is not taken into the regression. This could imply that the coefficients in front of the variables do not fully represent their true effects (Stock & Watson, 2015). For example, a variable controlling for infrastructure could be added, although this effect of infrastructure could already be taken into account by government expenditure and domestic investments. Finally the data for FDI is taken from the balance of payments. However, foreign firms also finance their investments via equity and debt on the domestic market. Thus FDI underestimate investment done by foreign firms.
Although the regression performed is not perfect, it can be shown to establish several basic facts on the relationship between GDP growth and FDI, schooling and the level of openness of Singapore. Any policy implications of this study are that Singapore, and any economy like Singapore, are likely to benefit from extensively investing in human capital. Continually focussing on making the economy attractive for foreign investors and thus FDI inflows, the GDP of Singapore is likely to grow. Further research, including a more appropriate measure of human capital, would improve the analysis of FDI on GDP growth. Furthermore, it would also be interesting to see if the results obtained above are relevant for countries that are less open than Singapore and rely more heavily on domestic investments and government expenditures. A final point for further research would be to investigate which sector benefits most from foreign investment. Focussing on increasing foreign investment in these sectors would result in an even greater impulse to GDP growth.
6. References
Akinlo, A. E. (2004). Foreign direct investment and growth in Nigeria: An empirical investigation.
Journal of Policy Modeling, 26(5), 627-639.
Alfaro, L., Chanda, A., Kalemli-Ozcan, S., & Sayek, S. (2004). FDI and economic growth: the role of local financial markets. Journal of International Economics, 64(1), 89-112.
Andrés, J., & Hernando, I. (1999). Does inflation harm economic growth? Evidence from the OECD. In The costs and benefits of price stability (pp. 315-348). University of Chicago Press. Angrist, J., & Krueger, A. B. (2001). Instrumental variables and the search for identification: From
supply and demand to natural experiments (No. w8456). National Bureau of Economic
Research.
Aschauer, D. A. (1990). Is government spending stimulative? Contemporary Policy Issues, 8(4), 30-46. Balassa, B. (1985). Exports, policy choices, and economic growth in developing countries after the
1973 oil shock. Journal of Development Economics, 18(1), 23-35.
Balasubramanyam, V. N., Salisu, M., & Sapsford, D. (1996). Foreign Direct Investment and Growth in EP and IS Countries. The Economic Journal, 106(434), 92-105.
Barbara Leitch Lepoer. (1989). Singapore: A Country Study. Washington: GPO for the Library of Congress. Retrieved 13-12-2015 from: http://countrystudies.us/singapore/
Barro, R. J. (1988). Government spending in a simple model of endogenous growth (No. w2588). National Bureau of Economic Research.
Barro, R. J. (1995). Inflation and economic growth (No. w5326). National bureau of economic research.
Basu, P., Chakraborty, C., & Reagle, D. (2003). Liberalization, FDI, and Growth in Developing Countries: A Panel Cointegration Approach. Economic Inquiry, 41(3), 510-516.
Baum, C. F. (2006). An introduction to modern econometrics using Stata. Stata press.
Bhagwati, J. N. (1973). The theory if immiserizing growth: further applications. International Trade
and Money. Toronto: University of Toronto Press.
Blomström, M. (1986). Foreign Investment and Productive Efficiency: The Case of Mexico. The Journal
of Industrial Economics, 35(1), 97-110.
Borensztein, E., De Gregorio, J., & Lee, J.-W. (1998). How does foreign direct investment affect economic growth? Journal of International Economics, 45(1), 115-135.
Carkovic, M., & Levine, R. (2002, June). Does Foreign Direct Investment Accelerate Economic Growth?
University of Minnesota Department of Finance Working Paper, p. 23.
Chakraborty, C., & Basu, P. (2002). Foreign direct investment and growth in India: a cointegration approach. Applied Economics, 34(9), 1061-1073.
De Long, J. B., & Summers, L. H. (1990). Equipment investment and economic growth (No. w3515). National Bureau of Economic Research.
De Mello Jr., L. R. (1997). Foreign direct investment in developing countries and growth: A selective survey. The Journal of Development Studies, 34(1), 1-34.
De Mello Jr., L. R. (1999). Foreign direct investment-led growth: evidence from time series and panel data. Oxford Economic Papers, 51(1), 133-151.
European Commission (2015). Trade: Singapore. Retrieved 15 December 2015 from: http://ec.europa.eu/trade/policy/countries-and-regions/countries/singapore/
Feder, G. (1983). On exports and economic growth. Journal of Development Economics, 12(1), 59-73. Greenaway, D., & Sapsford, D. (1994). What does liberalisation do for exports and growth?
Weltwirtschaftliches Archiv, 130(1), 152-174.
Gregorio, J. D. (1992). Economic growth in Latin America. Journal of Development Economics, 39(1), 59-84.
Haldrup, N., & Jansson, M. (2005). Improving size and power in the unit root testing. Aarhus University
Economics Paper, (2005-02).
He, Q., & Sun, M. (2009). FDI and Economic Growth: the Role of Gradual Financial Deregulation. Hymer, S. H. (1976). The international operations of national firms: A study of direct foreign
investment (Vol. 14, pp. 139-155). Cambridge, MA: MIT press.
International Monetary Fund. (2015). IMF Country Report: Singapore. Retrieved 13 December 2015 from: http://www.imf.org/external/pubs/ft/scr/2015/cr15200.pdf
Jordaan, J. A. (2004). Estimating FDI-induced externalities when FDI is endogenous: A comparison between OLS and IV estimates of FDI-induced externalities in Mexico. London School of
Economics and Political Science, Research Paper in Environment and Spatial Analysis, 92.
King, M. L. (1981). The alternative Durbin-Watson test: An assessment of Durbin and Watson's choice of test statistic. Journal of Econometrics, 17(1), 51-66.
Kokko, A. (1994). Technology, market characteristics, and spillovers. Journal of Development
Economics, 43(2), 279-293.
Krugman, P. R., Obstfeld, M., Melitz, M. J. (2015). International Economics: Theory and Policy (10th
edition). Harlow: Pearson Education Limited.
Li, X., & Liu, X. (2005). Foreign Direct Investment and Economic Growth: An Increasingly Endogenous Relationship. World Development, 33(3), 393-407.
Liu, X., Burridge, P., & Sinclair, P. O. (2002). Relationships between economic growth, foreign direct investment and trade: evidence from China. Applied Economics, 34(11), 1433-1440.
Mah, J. S. (2010). Foreign Direct Investment Inflows and Economic Growth: The Case of Korea. Review
of Development Economics, 14(4), 726-735.
Monetary Authority of Singapore (2014). Economics Explorer Series No.3 – Inflation.
Retrieved 29-12-2015 from: http://www.mas.gov.sg/~/media/manual%20migration/ Economics%20Explorer%20Series/explorer3.pdf
Mencinger, J. (2003). Does Foreign Direct Investment Always Enhance Economic Growth? Kyklos,
Mishkin, F., S., Matthews, K., & Giuliodori, M. (2013). The Economics of Money, Banking & Financial
Markets. (1st edition). Harlow: Pearson Education Limited.
Nair-Reichert, U., & Weinhold, D. (2001). Causality Tests for Cross-Country Panels: a New Look at FDI and Economic Growth in Developing Countries. 63(2), 153-171.
National Library Board (2014). Singapore experiences its first post-independence recession. Retrieved 13 December 2015 from: http://eresources.nlb.gov.sg/history/events/9f9489cf-5432-4797-bf66-fd1b3bab7a2b
OECD (2011), “Singapore: Rapid Improvement Followed by Strong Performance”, in Lessons from
PISA for the United States, OECD Publishing. Retrieved 06-12-2015 from:
http://dx.doi.org/10.1787/9789264096660-en
Ozturk, I. (2007). Foreign Direct Investment - Growth Nexus: a Review of the Recent Literature.
International Journal of Applied Econometrics and Quantitative Studies, 4(2), 79-98.
Pilbeam, Keith. (2013). International Finance. (4th edition). Basingstoke: Macmillan Publishers
Limited.
Ramirez, M. D. (2006). Is foreign direct investment beneficial for Mexico? An empirical analysis, 1960– 2001. World Development, 34(5), 802-817.
Reisen, H., & Soto, M. (2001). Which Types of Capital Inflows Foster Developing-Country Growth?
International Finance, 4(1), 1-14.
Sarel, M. (1995). Nonlinear Effects of Inflation on Economic Growth. IMF Working Paper. Retrieved 28 December 2015 from: http://ssrn.com/abstract=883204
Salvatore, D., & Hatcher, T. (1991). Inward oriented and outward oriented trade strategies. Journal of
Development studies, 27(3), 7-25.
Solow, R. M. (1956). A Contribution to the Theory of Economic Growth. Quarterly Journal of
Economics, 70(1), 65-94.
Stock, J., H., & Watson, M., W. (2015). Introduction to Econometrics: updated third edition. (3rd edition).
Harlow: Pearson Education Limited.
Stock, J. H. (1994). Unit roots, structural breaks and trends. Handbook of econometrics, 4, 2739-2841. UNCTAD (2015). World Investment Report 2015.
United Nations (2015). Human Development Report 2015.
Vahter, P. (2010). Does FDI spur innovation, productivity and knowledge sourcing by incumbent firms? Evidence from manufacturing industry in Estonia.
White, H. (1980). A heteroskedasticity-consistent covariance matrix estimator and a direct test for heteroskedasticity. Econometrica: Journal of the Econometric Society, 817-838.
World Bank (2015). World Bank Indicators. Retrieved 29 November 2015 from: http://data.worldbank.org/indicator?display=default
WTO (2012). Trade Policy Review: Singapore. Retrieved 29 November 2015 from: https://www.wto.org/english/tratop_e/tpr_e/tp367_e.htm
Zhang, K. H. (2001). Does foreign direct investment promote economic growth? Evidence from East Asia and Latin America. Contemporary Economic Policy (19), 175-185.
