Master Thesis. Home-country determinants of outward FDI: Evidence from BRICS economies and five developed countries

Master Thesis Home-country determinants of outward FDI: Evidence from BRICS economies and five developed countries Msc International Financial Management Msc Business and Economics Faculty of Economics and Business University of Groningen Faculty of Social Sciences Uppsala University Student number: S2980991 Student Number: hawa0038 Haiyan Wang Supervisor: Dr. Y.R. Kruse 13 January 2017

Abstract This paper studies the home-country determinants of outward FDI with a focus on nine empirically recognized host-country determinants of inward FDI, namely market size, labor cost, exchange rate, inflation, interest rate, political risks, corruption, openness, and technology. Based on a panel with 183 observations from BRICS and five developed countries (Australia, Germany, Japan, UK, US), evidence is found that market size, inflation, interest rate, political risks, and openness have significant influence on FDI outflows. Moreover, the results of this study show that there are striking differences between developing and developed countries regarding to the drivers for outward FDI. Key words: market size; labor cost; exchange rate; inflation; interest rate; political risks; corruption; openness; technology; outward FDI; BRICS 1

1. Introduction In order to expand business overseas, a firm or individual could conduct Foreign Direct Investment (FDI) to gain effective control of a foreign business. For a long time, developed countries have been the main role of outward FDI, which contributed to the economic growth for both investing and recipient countries. However, in the past decades, rise in investment flows from emerging and transition economies are noticeable. While developed countries conduct large amount of FDI in developing countries, overseas investments from developing countries to developed ones also experience an upward trend. According to the World Investment Report 2016, China is the third largest investor in the world after the United State and Japan with $128 billion FDI outflows. Further, BRICS, a league consists of five major emerging countries: Brazil, Russia, India, China and South Africa, represented approximately a third of FDI flows to developing and transition economies. Moreover, China, Brazil and India are on the top 20 FDI inflows host countries list and Russia and China are on the list of top 20 FDI outflows in 2015. Such phenomena raise a question: what are the drivers for countries to make outward FDI? Is there any differences for developed and developing economies? Previous studies mainly focus on host-country determinants of inward FDI from developed countries (Culem, 1988; Bevan and Estrin, 2004; Buckley et al., 2007). However, the motivations of outward FDI from both developed and developing countries have received few attentions. Therefore, it is open to be questioned that whether economic development is systematically related to outward FDI in emerging countries. This is also supported by institution theory that changes in institutional environment should be considered when doing study in developing economics. Thus, more empirical researches about outward FDI when taking developing economies as home countries are needed. The recognized determinants of inward FDI from prior studies include market size, labor cost, political risk, etc., which are considered significantly influence the location choice of foreign investments (Culem, 1988; Bevan and Estrin, 2004; Buckley et al., 2007). Conversely, do these factors also influence the outward FDI? What are the determinants of outward FDI? Do they vary from country to country? With these research questions, this paper focus on the role of home-country factors on outward FDI from BRICS countries and five developed countries: Australia, Germany, Japan, United Kingdom and the United States. The inclusions of 9 2

determinants (market size, labor costs, exchange rate, inflation, interest rate, political risks, corruption, openness, and technology) are based on empirical work on FDI and other relevant research. This study is undertaken by a panel data model with country fixed effects based on the full sample which contains 183 observations. Also, I run the regressions with two sub samples. One consists of the BRICS economies, the other one contains the five developed countries. The final results suggest that determinants of FDI outflows may base on home country characteristics. With this research, the literature on home-country factors that relate to outward FDI would be enhanced. The remainder of this paper is structured as follows. Section 2 holds an overview of empirical research as well as the determinants have been researched and hypotheses, after which section 3 introduce data collection and the methodology employed. Section 4 presents the regression results and corresponding analysis. Finally, a conclusion with contributions and suggestions for further research is given in section 5. 2. Literature review and Hypotheses In this part, I will first introduce what have been studied on FDI and on the recognized determinants. Besides, the motivations behind the selected determinants would also be discussed using Dunning's eclectic paradigm (1980). Next, hypotheses are built on certain analysis on existing literature. 2.1 Inward and outward FDI FDI is a popular topic on academic literature. It is natural to argue that FDI has positive influences on economic growth in the host countries. For developing countries, FDI could contribute to capital accumulation which is suggested by the Solow growth model and future industrialization. While in developed economies, FDI is a way to absorb new technologies and business practices (Cipollina, et al., 2012). Therefore, policymakers tend to attract foreign investors by providing certain incentives. On the other hand, multinational enterprises engage in outward FDI for many different reasons. According to location advantages proposed in Dunning's eclectic paradigm, three elementary motives are recognized: getting access to foreign market, achieving cost efficiency or lacking supply of natural resources in home country (Dunning, 1979). However, this seems 3

to be the case for most developed countries, especially for the last two reasons. For instance, Singaporean government encourage offshore investments by reducing tax on foreign income in order to encounter the domestic limitations of human capitals and natural resources (Lee et al., 2016). Nonetheless, developing countries like China, with abundant natural resources and labor force, has also become a major source of outward FDI. A large amount of previous studies focus on the determinants of FDI inflows to host countries. The main determinants of inward FDI from prior studies include market size, economic growth, labor cost, inflation, corruption, etc. (Kobrin, 1976; Bevan and Estrin, 2004; Benáček, et al., 2014). However, studies on the motivations of outward FDI from home countries are incomplete. Therefore, this paper seeks to examine the home-country determinants of FDI outflows and test whether the inward determinants are also applicable for explaining the outflows. 2.2 Market size According to Kobrin (1976), market size and economic growth of host countries are significantly related to inward FDI. Bevan and Estrin (2004) find that the disparity of market size between home and host countries is a powerful explanation for FDI flows from developed western to central and eastern Europe. They argue that market size of home country is a surrogate for product demand and production capacity, therefore market size of the home country is likely to positively alter FDI flows. Moreover, Egger and Pfaffermayr (2004) suggest that market size could serve as an indicator of capital abundance. Countries with abundant capital are expected to engage in more outward FDI activities. This is supported by Kimino et al. (2007) who argue that countries with larger market size have greater capacity to conduct foreign production, with greater amount of capital reserve and intangible assets such as marketing experience and technologies. On the other hand, it is argued that firms from a large home country could raise capital for foreign investment more easily. Because there are larger amount of firms seek to expand into global market on larger scale (Pan, 2003). Besides, some prior studies suggest that home country size is positively related to outward FDI (Tallman, 1988; Grosse and Trevino, 1996). On the contrary, based Dunning's eclectic paradigm, countries with relatively small market size tend to conduct market-seeking FDI. Therefore, the hypotheses with regard to market are formed as: 4

H1a: The market size of home country and FDI outflows are positively related. H1b: The market size of home country and FDI outflows are negative related. 2.3 Labor cost As labor cost accounts for a considerable portion of production costs, it is very important for investment location decision. Most of prior studies conclude that labor cost in host country is negatively related to FDI from developed to developing countries. For instance, with data of FDI flows from EU to CEEC economics, Janicki and Wunnava (2004) find that countries with relatively low labor cost are more attractive for foreign investors as international firms tend to relocate production to where human capitals are available at lower wages level, especially for firms that engage in labor-intensive production. Likewise, Bevan and Estrin (2004) also state that high labor cost of host country has negative impacts on its inward FDI using FDI data from western Europe to central and eastern Europe. On the other hand, based on data from Germany and United Kingdom, Hatzius (2000) find that high unit labor cost of home country drives FDI outflows to grow. Similarly, Dunning s paradigm suggests that firms engage in outward FDI for cost-reduction purpose. Because of the economic and educational development, labor cost in China has experienced a continuous increase. BRICS countries may share the same experience in this case. Moreover, foreign outsourcing is associated with increase in the wages paid to skilled employees in the United States, Japan, Hong Kong (China) and Mexico (Feenstra and Hanson, 2001). Therefore, the changes of labor cost of home country could be driver for outward FDI. Thus, H2: The labor cost of home country and FDI outflows are positively related. 2.4 Exchange rate Exchange rate volatility could generate both challenges and chances for multinational enterprises, forcing them to manage the potential risks. From previous research, results about effect of exchange rate uncertainty on FDI flows are mixed. Cushman (1985) examine the impacts of exchange rate risk on FDI and find that devaluation of home currency encourage inward FDI in order to lower capital cost, but such effect may be offset by varied costs of other inputs. Thus, no significant impacts on FDI as results. On the other hand, appreciation of home currency lead to reduced capital costs which enable multinational enterprises 5

associate with foreign investments more easily (Benassy-Quere et al., 2001). Moreover, There are also some theoretical models declare that exchange rate volatility will inhibit foreign investments. Because erratic variation of exchange rate will cause fluctuations in production costs and revenues of foreign affairs (Russ, 2007). Campa (1993) also argue that volatility of home currency will deter inward FDI. Meanwhile, some argue that there is positive relationship between exchange rate volatility and FDI flows. Itagaki (1981) posits that the profits of a foreign affiliate will increase when home currency depreciates and therefore international firms tend to invest abroad under the exposure of exchange rate risk. Additionally, Sung and Lapan (2000) suggest that multinational firm tend to strengthen its production flexibility by transferring production to the countries where costs of inputs are lowest resulted from reduced value of local currency. Thus, exchange rate volatility stimulate FDI flows. Moreover, short-term volatility will drive risk-averse investors to conduct outward FDI to reduce risks (Goldberg and Kolstad, 1995). Overall, empirical studies suggest that FDI activities response to the exchange rate volatility. As results from previous are mixed, three hypotheses are made: H3a: The exchange rate volatility of home country and FDI outflows are positively related. H3b: The exchange rate volatility of home country and FDI outflows are negatively related. H3c: There is no significant impacts of exchange rate volatility of home country on FDI outflows. 2.5 Inflation rate Uncertainty of the economic environment may discourage FDI flows to developing countries. Empirical literature employ plenty of variables to explain such effect, e.g., volatility of inflation and exchange rate (Jun and Singh,1996; Resmini, 2000). High inflation would lead to price volatility which could reduce expected return on investment, and reflect host government failures to conduct proper expansionary macroeconomic policies (Okafor, 2015). Likewise, Udoh and Egwaikhide (2008) and suggest that inflation and inward FDI are negatively related and put such relationship down to poor-functioned macroeconomic policies in host country, which boost the inflation rate. Hence, conversely, if inflation is high at home, domestic investors are likely to avoid potential loss. In sum, there are limited research on the relationship between inflation and FDI, especially with outward FDI. Thus, the two 6

hypotheses are proposed as followed: H4a: The inflation rate of home country and FDI outflows are positively related. H4b: The inflation rate of home country and FDI outflows are negatively related. 2.6 Interest rate If rise in exchange rate or interest rate volatility derive from home country, such changes will attract inward FDI. Vice versa, if changes of interest rate originate in the host country, foreign investment would be discouraged (Russ, 2007). Empirical research suggest that business cycle has significant effect on FDI outflows. Cavallari and d'addona (2013) view interest rate as one of the drivers of outward FDI based on two mechanisms: imperfect capital market and sunk cost of entry. As for the former one, there is an positive income effect during cyclical expansion period. Increasing prices raise the net value of the company and therefore promote the financing for new investments. Sunk cost of entry, on the other hand, influence the entry decision at the first place. Both models advocate that interest rate do matter for the FDI decisions. As results, Cavallari and d'addona (2013) find that interest rate mainly affects the amount of FDI. Moreover, Yeyati et al. (2007) also suggest that interest rate cycle of home country is a crucial determinant of FDI flows. To be specific, interest rate and outward FDI are negatively related in US and Europe. Cuts in interest rate of home country reduce financial costs when financing mainly at home, which also contribute to FDI outflows. This is also supported by Kyrkilis and Pantelidis (2003) that domestic investors can gather capital for overseas investments more easily with decreased opportunity cost of capital. In a similar vein, Cushman (1985) also find that international firms from countries have cost advantage over local competitors with competitively low interest rates. Thus, the hypotheses are developed as : H5a: The interest rate of home country and FDI outflows are negatively related. H5b: The interest rate of home country and FDI outflows are positively related. 2.7 Political risk Political risk is a popular topic in international business literature. It could be viewed as the probability of occurrence of political events that will influence expected profitability of investment activities or governmental interference with business operations (Kobrin, 1979). In other words, political risk could alter investors investment decisions in order to reduce risks. 7

According to Hatzius (2000), the average level of FDI barriers, which relate to the costs of producing overseas, is determined by policy. By studying FDI flows to central and eastern Europe, Benacek et al. (2000) state that political risks may have power on the distribution of investment across countries and on the investment location decisions. FDI is affected by political risk along with volatility of future cash flow. For host countries, previous studies generally suggest that a less risky investment environment could contribute to the ability to attract inward FDI and political instability reduce inward FDI significantly (Schneider and Frey, 1985; Janicki and Wunnava, 2004). However, Pan (2003) find the reversed result in the case of China. The explanation is that foreign investors have to acquire more equity in order to take control of the venture during risky periods. Interestingly, Méon and Sekkat (2012) find that inward FDI are negatively related to political risks, but become less sensitive to political instability when the global amount of FDI flows are larger. Therefore, the effects of political risks on FDI could be neglected or even come along with greater volume of inward FDI. For home country, the extent to which political risks influence FDI flows rests primarily on the attitude towards risk and the political characteristics of home country (Kimino et al., 2007). Further, Tallman (1988) states that a poor investment climate with higher political risks of the home country would encourage outward FDI to the relatively stable countries. As there is no mainstream suggestions, two hypotheses are formed: H6a: The political risk of home country and FDI outflows are positively related. H6b: The political risk of home country and FDI outflows are negatively related. 2.8 Corruption A corrupt government does not provide impartial market competition opportunities to all the players. Payers engage in bribery for purposes like reducing transaction costs or getting priorities and advantages for competition. And political interference could change competition on illiberal market with power of local bureaucracy (Benáček, et al., 2014). Nonetheless, according to Bardhan (1997) and Aidt (2003), corruption could serve as either a grabbing hand or a helping hand for inward FDI. On one hand, bribery is costly and therefore firms are less willing to conduct business in countries with high level of corruption. Egger and Pfaffermayr (2004) find a negative 8

relationship between corruption of host countries and inward FDI. Further, they argue that corruption significantly affects OECD FDI but not non-oecd FDI. For extra-oecd countries, the growth of FDI is mainly induced by economic growth rather than corruption. On the other hand, corruption can offset negative effects of government failures under certain circumstances. For instance, Lui (1985) argues that firms placing high value on time are more likely to pay bribes to accelerate business process. In this way, corruption could further allocation efficiency. Moreover, Beck and Maker (1986) find that the most productive companies are the ones offer most brides in bidding competition. As it is well-known, corruption is a serious issue for BRICS countries. According to Transparency International s Corruption Perception Index 2016, Brazil and India rank 76th, Russia ranks 119th, China ranks 83th, south Africa ranks 61th, among 168 countries. Corruption would make room for flexibility and improve competitiveness of the market and limit opportunities for productivity (Ngunjiri, 2010). Therefore, corruption problems of home country may force domestic investors choose overseas investment. However, as mentioned above, corruption could also serve as a helping hand. Therefore, H7a: Corruption in home country and FDI outflows are positively related. H7b: Corruption in home country and FDI outflows are negatively related. 2.9 Openness In general, empirical studies show a positive correlation between the level of openness and FDI inflows. International orientation as a measure of openness is an crucial determinant to FDI and could contribute to national competitiveness (Habib and Zurawicki, 2002). The more open a country is to international trade, the more appealing it is for inward FDI (Chakrabarti, 2001). Rodriguez and Pallas (2008) also argue that liberal trade regime of host countries can encourage inward FDI with given internalization advantages for investing firms. In a similar vein, the ease of entry plays an important role and trade barriers would inhibit FDI (Culem, 1988). Higher level of openness could help domestic firms conduct foreign investments more easily. As openness regarding to both import and export, it is expected to be positive related to FDI outflows as inflows. Thus, a hypothesis is formed as: H8: Trade openness of home country and FDI outflows are positively related. 9

2.10 Technology While FDI inflows can bring benefits like advanced technology and innovative management to the host country (Kimino et al., 2007), international firms may also undertake outward FDI for improving their competitiveness in technology and innovation, especially for firms in developing countries. For example, there is evidence that Chinese multinational firms internationalized through M&As in order to acquire foreign technology (Buckley, et al., 2007). Because Chinese firms that following an mimic strategy and producing low-end products are stuck in a standstill period. And such situation at home will drive firms to conduct FDI activities abroad (Deng, 2009). Furthermore, as Deng (2004) argued, Chinese firms prefer to acquire an existing foreign company than start a new business abroad. In this way, they may transfer superior technologies back to home country and enhance their competitiveness in a relatively short time. Therefore, it is reasonable to predict that countries with relatively low level of technological development are more motivated to undertake outward FDI. Hence, H9: There is a negative relationship between technology development in home country and FDI outflows. 3. Data and Methodology 3.1 Dependent and independent variables FDI outflows is measured by capital received, either directly or indirectly, by a foreign direct investor from a FDI enterprise. The interpretation of the results is then discussed in relation to the independent variables as followed. With respect to market size, GDP per capita is used to capture its effects. To determined labor costs, a Relative Unit Labor Costs Index is used, which takes the year of 2008 as standard. Unit labor costs, calculated as the ratio of total labour costs to real output, measure the average cost of labour per unit of output. Annual interest rate of government securities is indicator of interest rate in this paper. As for political risk, a indicator for government effectiveness which comprises the quality of public and civil service, and the degree of its independence from political stress, the quality of policy formation and application, and the integrity of the government's commitment to such policies. The higher score means lower risks. The perceived levels of public sector corruption is ruled by Corruption Perceptions Index (CPI) based on expert opinions. In particular, CPI, published by Transparency 10

International annually, currently ranks 168 countries on a scale from 100 (very clean) to 0 (highly corrupt). Regarding to openness, it is measured by the ratio of merchandise trade (imports and exports) of GDP. The sum of patent applications from both nonresidents and residents is employed as indicator for technology development. 3.2 Data collection The data comprise a panel of 10 investing countries with 183 observations for the period 1996 to 2014. The sample home countries and time period are mainly decided by the extent to which collected information is capable to develop persistent measures of the selected variables over time. For the research, 10 countries are selected, including 5 developing countries (Brazil, Russia, India, China and South Africa) and 5 developed countries (Australia, Germany, Japan, United Kingdom and United States). All these sample countries play an important role for global FDI flows. As mentioned above, most empirical studies focus on host-country determinants of FDI from developed to developing countries. Therefore, analysis based on these sample countries may provide a preliminary understanding of the relationship between home-country determinants and outward FDI. Further, it is worth to study whether the drivers of FDI outflows are the same between developed and emerging economies. DataStream is used to collect data. Details and sources of data are described in table 1. 11

Table 1. List of variables and sources of data Variables Proxy for Source OFDI Outward foreign direct investment, flows United Nation Conference on Trade and Development (UNCTAD) MS GDP per capita (US dollar) World Bank Development Indicator LABOR Relative unit labor cost Oxford Economics ER Exchange rate (home currency per US dollar) WM/Reuters INR Interest rate Oxford Economics INF Inflation rate International Monetary Fund (IMF) POR Political risk index Oxford Economics CORR Corruption Perceptions Index Transparency International OPEN Openness World Trade Organization TE Technology World Intellectual Property Organization (WIPO) 12

3.3 Panel data regression model with country fixed effect As Kimino et al. (2007) argued, the application of panel estimation with fixed effect could control the heterogeneity of individual country and therefore reduce the probability of misspecification and yield more precise delivery of end results. For double confirm, a Redundant Fixed Effects Tests is undertaken. The results showed in Appendix A suggest that for this model, cross-section fixed effects are statistically significant and needed to be controlled, while period fixed effects are insignificant. Moreover, a Hausman Test is employed to determine whether a random effects model is more appropriate. As shown in Appendix B, the p-value of the test is 0.051 implying that the fixed effects model is to be preferred. Hence, based on Kimino et al. (2007) and Buckley, et al. (2007), a model is created for the research: OFDIit = α + β1msit + β2laborit + β3erit + β4inrit + β5infit + β6porit + β7corrit + β8openit + β9teit + uit where i = subscript for each individual country, t = subscript for time, α is a constant, β1,2,,10 are the coefficients for the independent variables. The error term uit accounts for any unobserved individual home-country effect that is implicitly included in the regression. The regressions analyses are conducted with the use of EVIEWS 9.5. In many cases, time series appear to be non-stationary, that means, the values of time series do not fluctuate with a constant variance or with a constant mean. Such data may lead to spurious regressions, which could be identified by high R 2 and high residual autocorrelation. In order to avoid spurious regressions and results, I transform data for OFDI, MS, LABOR, TE into growth rates. In this way, the data are smoothed and scaled down, thereby, accountability of the regression results would be improved. 4. Results and analysis This part consist of four sections. First, summary statistics would be introduced, followed by the correlations test results as the second section. In the third section, regressions results of a panel data model with fixed effect would be presented. Except for regression based on the 13

whole data set, for comparison, results of two different groups: BRICS and developed countries would also be presented. Fourth, a number of robustness tests would be conducted by testing the model one by one country. Discussions are along with the corresponding results. 4.1 Descriptive statistics Table 2 presents the summary statistics based on 10 countries within the period 1996-2014. Regarding to the dependent variable OFDI, the mean of 0.747 is comparatively close to the median -2.145, which means the data is resilient against extreme values and the distribution is normal. With respect to the independent variables, it is reasonable that difference between the mean and median of ER is large as exchange rate are based on local currency per US dollar. Besides, except for CORR, all factors has similar mean and median respectively. To be more specific, descriptive statistics of dependent variable OFDI are also showed on Table 3. Generally, all countries have experienced considerable fluctuations of FDI outflows. Among all countries, Brazil has largest volatility of FDI outflows with a minimum of - 190.674 and a maximum of 189.487, while China, India and Russia have positive noticeably positive mean and median. Similar to Brazil, outward FDI of South Africa has fluctuated a lot with negative mean (-5.867) and median (-31.093). Japan is the only country with both positive mean and median among the five developed country. Meanwhile, Australia have fluctuated from -155.910 to 155.510. UK also has negative mean and median growth rate as Australia, but at a smaller scale. Both Germany and the US have positive mean but negative median. Taking the mean of every country as reference, FDI outflows from Russia increased at most while the ones from Brazil decreased at best. Furthermore, the countries listed on top 20 FDI outflows in 2015, namely the US, Japan, China, Germany, Russia, all have positive mean growth. 4.2 Correlations According to Brooks (2014), high correlations between independent variable could lead to multicollinearity issue. Therefore, the correlation matrix showed on table 3 is used to present a rudimentary check for multicollinearity. All correlations of 0.5 or higher would be considered whether they should be dropped out of the model. There are four correlations are higher than 0.5, namely between inflation rate and interest rate (0.576), between inflation and political risks (-0.613), between inflation and corruption (-0.618), and between political risks 14

and corruption (0.679). Determination on keeping which variables would be based on results of a Variance Inflation Factor (VIF) test. As could be seen in the Appendix C, the VIF of all variables are far below a threshold of 10 with a highest VIF score of 1.630 for political risks. Therefore, no independent variables would be excluded in the regression. Table 2. Descriptive statistics of the full sample for the period 1996 to 2014 Mean Minimum Median Maximum Std. Dev. Observations OFDI 0.747-190.674-2.145 189.487 71.958 183 MS 2.759-7.849 2.081 13.600 3.265 183 LABOR 0.576-38.624 0.766 28.133 8.405 183 ER 19.952 0.503 2.422 133.775 33.344 183 INR 7.031 0.550 5.370 36.400 5.924 183 INF 3.863-1.408 2.826 15.757 3.462 183 POR 5.032 3.449 5.288 6.447 1.006 183 CORR 58.032 21.000 65.700 88.600 22.611 183 OPEN 36.061 12.293 35.953 72.625 14.656 183 TE 4.998-85.590 3.462 70.063 14.966 183 Table 3. Descriptive statistics of OFDI of individual country for the period 1996 to 2014 Country Mean Minimum Median Maximum Std. Dev. Observations Brazil -41.396-149.295-74.338 189.487 90.581 19 China 18.274-63.424 14.165 123.012 51.043 19 India 15.446-80.217 15.971 171.639 65.361 19 Russia 40.884-41.559 25.979 174.126 66.044 12 South Africa -5.867-136.744-31.093 139.429 93.358 19 Australia -25.344-155.910-46.018 155.510 88.066 19 Germany 6.778-70.602-17.728 163.233 64.974 19 Japan 13.370-41.651 9.796 91.242 33.723 19 UK -7.841-190.674-8.424 99.421 77.440 19 US 7.958-94.788-3.516 127.986 46.641 19 Note: Data of OFDI are transformed into annual growth rate. 15

Table 4. Correlations among variables OFDI 1.000 OFDI MS LABOR ER INR INF POR CORR OPEN TE MS 0.209*** 1 LABOR 0.125* 0.330*** 1 ER 0.108-0.057-0.144* 1 INR -0.082-0.098-0.039-0.254*** 1 INF -0.069 0.139* 0.019-0.124* 0.576*** 1 POR -0.035-0.442*** -0.193*** -0.094-0.483*** -0.613*** 1 CORR -0.081-0.461*** -0.195*** -0.096-0.452*** -0.618*** 0.679 1 OPEN 0.106 0.205*** 0.081-0.299*** -0.292*** 0.023 0.049 0.038 1 TE 0.095 0.350*** 0.098-0.079 0.061-0.011-0.239*** -0.251*** -0.013 1 *** Indicates significance at the 1% level; **Indicates significance at the 5% level; * Indicates significance at the 10% level 16

4.3 Regression results and analysis Table 4 shows that there are four significant variables for FDI outflows, namely market size, interest rate, inflation rate and trade openness based on regressions on the full sample. The same estimations are also undertaken for two country group: BRICS and 5 developed countries. Corresponding results are included in table 4. The R 2 of these three regressions are 0.218, 0.331, and 0.172, respectively. The two country groups show striking different results. 4.3.1 Market size Hypothesis 1a predicts a positive relationship between market size and outward FDI. It is proved by the results that the variable market size is statistically significant at the 10% level with a positive coefficient of 4.638. This indicates that home countries with larger market size are more likely to conduct outward FDI. Further, it also implies that economic growth could be a driver force for increasing outward FDI. The results are corresponding with the findings of previous studies (Bevan and Estrin, 2004). The underlying reason could be that countries with larger market size have greater capital abundance to engage in overseas investment activities (Egger and Pfaffermayr, 2004). As the market size is measured by GDP per capita, which is also an indicator of income level. It is common that developed countries have relatively mature domestic market with less opportunities for business expansion. Hence, domestic investors in these countries may have incentives to expand business in foreign market by FDI. As aforementioned, BRICS countries have become major investors among the world in recent with relatively low GDP per capita level. However, the overall results derived from 10 sample countries do not provide explanations for this situation. Surprisingly, results from two country groups show that market size is significant neither for BRICS and the five developed countries. Therefore, the overall results may be driven by several individual sample countries. Additionally, according to Dunning's eclectic paradigm (1980), countries with relatively small market size have more intentions to engage in outward FDI for market-seeking purpose, which is contrary to results. Thus, the insignificant results of two country groups may due to offset effects of combined reasons. 4.3.2 Interest rate The results with respect to hypothesis 5b confirm that interest rate serve as a driver of 17

outward FDI, with a coefficient of 5.735 significant at 1% level. Interestingly, interest rate significantly influence BRICS countries significant at 1% level with a beta of 5.710, but is insignificant for the five developed countries. This is could be explained by the findings of Cavallari and d'addona (2013) that competitively high interest rate has positive income effect on the net value of the firm and therefore financing for new investments becomes more easier. However, such arguments are based on the cyclical expansion period. In the past decade, BRICS countries have experienced great economic progress, while developed countries have relatively lower growth rate as shown. Thus, interest rate has more explanatory power for the changes of FDI outflows from BRICS. Another reason could be volatility of interest rate rather than interest rate itself affects the decisions of engaging in overseas investment activities (Sung and Lapan, 2000). As shown on Appendix D, exchange rate volatilities of BRICS are larger than the ones of the developed countries. Besides, the absolute interest rates of BRCIS are also higher. An imperfect capital market could create uncertainty which leads to interest rate volatility. When comparing to BRICS, the five developed countries have relatively complete and integrated capital market. Therefore, it is expected that there are higher uncertainty of interest rate changes in BRICS markets. 4.3.3 Inflation rate The results regarding to hypothesis 4b suggest a negative relationship between inflation rate of home country and FDI outflows. Many previous studies argue that high inflation rate of host country would discourage inward FDI as price uncertainty may reduce expected return on investments (Udoh and Egwaikhide, 2008). For home country, high inflation rate implies failure of macroeconomic policies and would reduce the rate of investment as the market becomes unstable. However, results from two different country groups are similar to interest rate that inflation significantly affects BRICS, but not the developed countries. As shown on Appendix E, BRICS have experienced more fluctuations of inflation rate compared to the five developed countries. In a similar vein, volatility of inflation rate derived from an imperfect capital market would deter domestic investors. Furthermore, the value of home currency may depreciate due to high inflation, which reduces capital abundance and makes financing for investments more costly. 4.3.4 Political risks As higher rating means lower political risks in this paper, the results of the full sample 18

confirm that lower political risks relate to more FDI outflows as hypothesis 6b proposed. This finding add new evidence to current studies as most of them state that political risks of host countries would either attract or deter foreign investors (Janicki and Wunnava, 2004; Pan, 2003). However, such negative relation between political risks and outward FDI is contrary to the argument of Tallman (1988) that high political risks would push domestic investors to invest overseas. Therefore, the underlying reasons may lie with other aspects related to political risks. One possible explanation as Kimino et al. ( 2007) argued, the extent to which political risks affect FDI flows primarily determined by the investors attitude towards risk and the political characteristics of home country. On the other hand, political risks are not significant for the two sub-sample groups. 4.3.5 Openness Hypothesis 8 suggests a positive relationship between openness of home country and outward FDI. With a coefficient of 1.878 at 5% level, the independent variable openness significantly influence FDI outflows. Basically, this is matched with the previous literature findings (Rodriguez and Pallas, 2008). As openness is measured by the ratio of merchandise trade (imports and exports) of GDP, higher level of openness provides greater opportunities for both inward and outward FDI. Besides, opening of an economy could contribute to economic growth, which is positively related to FDI outflows as mentioned above. Likewise, trade openness could narrow the income differences between poor and rich countries by diffusion of knowledge and technology, as results, contribute to greater national productivity and economy (Apergis and Cooray, 2015). Hence, high level of openness also has indirect influence on the dependent variable. Apergis and Cooray (2015) also find that countries at similar development stage tend to have same patterns for both inward and outward FDI changes. Nonetheless, regression results from two country groups show distinct outcomes again. While openness is one of the significant drivers for outward FDI from BRICS (β = 3.191, p-value = 0.018), FDI outflows from developed countries do not seem to be affected by openness. 4.3.5 Further discussions From the full sample results, except for the significant results discussed above, the other variables are not statistically significant for FDI outflows. However, it is interesting to notice that for BRICS countries, exchange rate and corruption significantly influence outward FDI, 19

with a coefficient of -5.129 and -5.097 significant at a 10% level respectively. Consistent with Itagaki (1981) who posits that devaluation of home currency would be beneficial for foreign investment returns. In addition, exchange rate volatility of home countries would drive domestic investor to conduct foreign investments in order to hedge exchange rate risks. On the other hand, developed countries have experienced less exchange rate volatility in the past decades as Appendix F shows. Therefore, exchange rate has limited explanatory power for FDI outflows from developed economies. Furthermore, Cushman (1985) suggests that appreciation of home currency may lead to lower capital cost of foreign investment, but increase costs of other input. As results, exchange rate has no significant impacts on outward FDI. Such argument may help to explain the overall results with regard to the hypothesis 3c. As for corruption, higher CPI score indicates less corruption in this study. Bardhan (1997) and Aidt (2003) suggest that corruption could serve as a grabbing hand or a helping hand for inward FDI. In the case of BRISC, the result indicate that higher corruption would drive outward FDI increase. Hypothesis 7a is therefore confirmed with respect to the findings of Ngunjiri (2010). As it is well known, BRICS have serious problems with regard to corruption issue. Although their government made great efforts on improving this situation. Corruption level of these countries is still rather high compared to developed countries, which could be seen on Appendix G. The overall results show that labor costs and technology are insignificant for the selected sample countries. Buckley et al. (2007) argue that bilateral FDI are associated with the cultural and geographic proximity, which are not taken under consideration in this paper. Therefore, the regression results could be biased using data of the 10 selected countries, which have large cultural or geographical differences with each other. For example, most current literature about labor cost based on studies within European countries (Hatzius, 2000; Janicki and Wunnava, 2004; Bevan and Estrin, 2004). Moreover, determinants of inward FDI are not necessary the ones for outward FDI. These could explain why the results show unexpected outcomes. 20

Table 5. Regression results Note: The full sample consists of 183 observations from ten selected countries for the period 1996-2014. Sub-sample 1 contains data 88 observations from Brazil, China, India, Russia, and South Africa. Sub-sample 1 contains 95 observations from Australia, Germany, Japan, United Kingdom, and United States. OFDI Full sample Sub-sample 1 Sub-sample 2 Constant -262.030 53.230-416.086 (172.005) (242.117) (265.957) MS 4.638* 2.619 4.868 (2.494) (3.790) (4.303) LABOR 0.972 0.879 1.406 (0.678) (1.028) (1.056) ER -0.230-5.129* 0.437 (0.978) (2.676) (1.090) INR 5.735*** 5.710*** 8.253 (1.842) (2.073) (7.642) INF -7.835*** -9.789*** 2.754 (2.763) (3.166) (8.124) POR 34.787* -22.520 53.646 (32.399) (56.944) (42.759) CORR -2.727-5.097* 1.437 (1.710) (2.632) (2.546) OPEN 1.878** 3.191*** 1.256 (0.873) (1.316) (1.614) TE 0.236 0.119-0.288 (0.374) (0.410) (1.522) R-Squared 0.218 0.331 0.172 Adjusted R-Squared 0.132 0.213 0.051 Observations 183 88 95 Country fixed effects YES YES YES Time fixed effects NO NO NO *** Indicates significance at the 1% level 21

** Indicates significance at the 5% level * Indicates significance at the 10% level 4.4 Robustness test In order to study whether there is a common pattern for both developing and developed countries to conduct outward FDI, this paper examines the selected variables based on data from 10 countries, which are the main actors of global FDI activities. For further analysis, I separate the countries into two groups. And the results of developed country group are strikingly different from the primary results of all sample countries. Therefore, in this part, I would do robustness check by using the same estimation for every sample countries individually. By this way, I try to find out whether the overall results stay robust for individual country. Table 5 shows the combined regression results for the 10 sample countries. With respect to Brazil, inflation rate is negatively related to outward FDI with a beta of -14.291 significant at a 10% level. Besides, the variable openness has a coefficient of 23.360 significant at a 10% level. Both results are consistent with the results of the whole data set and the BRICS country group. Additionally, exchange rate have significantly negative impacts on FDI outflows from Brazil with a beta of -137.518 significant 10% level. When it comes to India, labor cost has significant influence with a coefficient of -6.020 significant at 10% level. Besides, inflation rate has a negative relation with outward FDI with a beta of -13.903 significant at 5% level. Similar to India, labor cost are the only significant variable with a coefficient of -6.601 significant at 10% level in China. Conversely, labor cost is positively related to FDI outflows from UK, consistent with the empirical findings of Hatzius (2000). Inflation and political risks are also positively related to FDI flows from UK. As for South Africa, only corruption has significant influence on FDI. While interest rate is the only significant determinant for Australia, market size is the only one for Germany. Regarding to Japan, there are three significant variables, namely inflation rate (β = -21.152, p-value = 0.022), openness (β = 10.698, p-value = 0.001), and technology (β = 8.789, p-value = 0.004). Lastly, political risks, corruption and technology have significant impacts on outward FDI from the US. In sum, every sample country have different drivers for outward FDI. This provides evidences that home-country determinants of outward FDI may based on region or country characteristics. Table 6. Robustness check for individual sample country. 22

OFDI Brazil China India Russia South Africa Constant -1392.472-207.903-1051.006-734.616-1684.797 (832.829) (456.013) (930.750) (1328.642) (1893.230) MS -16.891 3.502 4.188 25.379-15.265 (14.560) (9.984) (7.208) (19.864) (20.490) LABOR 3.005-6.601* -6.020* 0.510 2.645 (2.081) (4.293) (3.112) (18.187) (2.997) ER -137.518* -0.463 1.838 48.722-15.599 (64.651) (19.287) (3.558) (23.244) (15.725) INR 5.073 14.494 12.237-108.028 11.187 (3.446) (15.003) (8.462) (95.095) (21.211) INF -14.291* -3.100-13.903** 56.645-29.398 (7.174) (6.526) (5.292) (29.494) (17.194) POR 309.522 7.915 255.575-276.047 234.420 (232.723) (93.119) (207.163) (457.161) (369.701) CORR -1.021 7.499-6.870 24.779-21.558* (10.237) (5.850) (8.376) (9.534) (11.596) OPEN 23.360* 3.319-1.854-40.028 15.984 (12.529) (2.019) (1.996) (36.472) (9.215) TE -0.177 0.376 0.545-1.823-0.155 (1.445) (0.709) (0.554) (4.437) (1.065) R-Squared 0.555 0.699 0.668 0.698 0.371 Adjusted R- Squared 0.110 0.459 0.402 0.106 0.132 Observations 19 19 19 12 19 23

Table 6 (continued) Note: As exchange rate is measured by home currency per US dollar and data were collected from the same database, data for US are not available. OFDI Australia Germany Japan UK US Constant 730.545-668.431 141.609-2854.215** -898.3109 (1357.439) (764.870) (288.874) (1247.026) (1033.377) MS -11.603 14.566* -6.879-6.242-8.337 (24.392) (8.024) (3.798) (12.522) (62.474) LABOR 4.117 1.937 0.632 5.162* -0.184 (3.984) (4.946) (0.696) (2.419) (21.153) ER -38.109-259.561 0.372 133.378 NA (86.059) (167.137) (0.448) (309.197) NA INR 47.281* -7.921 17.702-25.429 160.585 (25.318) (24.809) (12.434) (17.011) (155.492) INF 36.006-51.186-21.152** 102.507*** 35.625 (27.767) (36.171) (7.827) (26.941) (159.239) POR -74.838 167.105-77.152 450.819* 448.492*** (204.456) (108.771) (51.792) (196.184) (117.930) CORR 6.271-1.737 0.026 8.525-21.641** (14.847) (13.212) (1.815) (6.458) (8.884) OPEN -17.085-0.267 10.698*** -14.658 10.093 (15.889) (4.280) (2.231) (8.502) (71.118) TE -5.541-3.098 8.789*** 3.364-4.462* (3.650) (6.202) (2.366) (3.621) (2.051) R-Squared 0.507 0.651 0.768 0.765 0.717 Adjusted R- Squared 0.113 0.371 0.583 0.577 0.491 Observations 19 19 19 19 19 *** Indicates significance at the 1% level ** Indicates significance at the 5% level * Indicates significance at the 10% level 24

5. Conclusion Due to globalization, the world economy have become more integrated. Developing countries gain more and more attentions as the main actors. Whether the rules are applicable all the players is needed to be studied. Although a lot of research have been done on the FDI topic, especially on inward FDI, it is still unclear what are motivations for conducting outward FDI. The goal of this paper is therefore provide insights into the home-country determinants of FDI outflows with a focus on comparing emerging and developed countries. An empirical analysis of outward FDI from 10 countries for the period 1996 to 2014 is undertaken, where five are developing countries and the other five are developed for comparison. Besides, a panel data model with country fixed effects is applied. When the entire sample is concerned, the market size of home country have significant and positive impacts on FDI outflows, thereby confirming prior studies on the subject. However, when the same estimators were used for the two country groups. Market size is insignificant for neither of them. With respect to interest rate, it is significantly and positively related to outward FDI. Particularly, it is persuasive that both absolute level and the volatility of interest rate affect the outward FDI, as different regression results were gained from the two country group. Regarding to inflation rate, it follows similar pattern as interest rate but with significantly negative effects on FDI outflows. Both interest rate and inflation rate are related to cost-reduction purpose for engaging in FDI with regard to Dunning's eclectic paradigm. Unexpectedly, lower political risks relate to more FDI outflows. Finally, trade openness of home country is beneficial to both inward and outward FDI combining the findings of previous research. Moreover, exchange rate and corruption are only significant for BRICS, but not for the full sample. Unexpectedly, labor cost and technology do not have significant influences on the dependent variable. As most of existing literature focus on single country or at least countries within the same region, how relevant the findings of this paper to other countries is needed to be further examined. Last but not least, robustness checks are conducted for every sample country individually and the results intimate that there is no common home-country determinants for all countries within the scope of selected independent variables and limited data. Though for policy makers, determinants of inward FDI are more noteworthy. Studies on outward FDI provide guidance that is useful for domestic investors to make foreign 25