Is foreign direct investment globalization-induced or a myth? A tale of Africa

Foreign direct investment (FDI) provides many African countries an important source of capital inflow. Despite notable improvements in these capital-scarce countries' economic, political, and social conditions, foreign investors have not considered them viable host locations. Since FDI brings enormous spillovers to its host, some countries have recently institutionalized globalization as the catalyst for reversing the trend. Against this backdrop, we examine the FDI–globalization nexus across 47 African countries for the 1996–2016 period. Using the augmented mean group estimator, the results suggest that FDI in Africa is indeed globalization-induced. Moreover, we find this positive nexus to be driven by the economic dimension of globalization. Overall, we demonstrate the potential of globalization in stimulating an FDI boom in Africa. This paper examines the effect of globalization on FDI inflows in Africa. A worthwhile area for future research is investigating the transmission channels through which globalization influences FDI inflows to the continent.


INTRODUCTION
Globalization entails the process of creating network connections among actors at intra-or multicontinental distances, mediated through a variety of flows, including capital, goods, ideas, and people (Clark, 2000). In defense of globalization, Bhagwati (2004) argues that globalization is instrumental in improving the quality of life. As an indispensable component of globalization, foreign direct investment (FDI) has received considerable attention over the past three decades.
However, Bhagwati (1978) suggests that FDI has mostly benefited highly-globalized countries than their lowly-globalized counterparts.
FDI emerges as an outcome of the resource-, market-, efficiency-and strategic assetseeking investment activities of multinational corporations (MNCs). 1 Narula and Dunning (2010) identify FDI as the most effective way for MNCs to enter developing countries. They argue that MNCs' activities in these countries stimulate domestic capital accumulation (Cipollina et al., 2012;Thangavelu et al., 2009;Gorg and Greenaway, 2004) and narrows the financing gap in investment that impedes economic growth (UNCTAD, 2013). 2 While FDI flows to developing and transitional countries fell in 2018, developing countries still hosted a record 54% of global FDI inflows (UNCTAD, 2019a).
The extant literature holds the view that FDI is primarily driven by globalization. In part, globalization offers MNCs better access to the international factor market (UNCTAD, 1998) and re-shapes the organization of MNCs' offshore activities (Cantwell and Narula, 2001). Empirically, 1 UNCTAD (1998) argues that globalization has changed the way MNCs pursue their investment objectives. 2 Jude (2019) demonstrates that FDI crowds out domestic investment in the transitional countries in the short run, but in the long run, it crowds in domestic investment. Noting the importance of the mode of FDI entry, Chen et al. (2017) find that greenfield FDI crowds out domestic investment in China while cross-border mergers and acquisitions (M&As) crowd it in. Flores and Aguilera (2007) identify globalization to be the motivation behind the top-100 US MNCs' locational choice.
Over the years, many African countries have attempted to attract FDI by institutionalizing globalization as part of the economic reform package during the post-colonial era. However, the amount of FDI in Africa remained abysmal relative to other regions. According to UNCTADStat (2019), Africa's share of the global FDI inflows stood between 0.71% and 4.83% during the 1990-2018 period. This trend was in sharp contrast to an average of 3.92% reported before 1990. These observations suggest that a positive FDI-globalization nexus might be a myth in Africa.
In this paper, we examine the possible nexus between FDI and globalization in Africa. To date, the extant literature (see, for example, Ito 2006, 2008;Quinn et al., 2011) has mostly captured globalization in terms of trade and financial openness or capital mobility restrictions. However, these measures are unidimensional that neglect the social and political aspects of globalization. With this in mind, we capture globalization by the KOF Globalization Index. The index is an ideal choice because it not only encompasses the economic, political, and social dimensions of globalization but also provides the de facto and de jure measures of globalization (Gygli et al., 2019). This multidimensional index was also used by Aluko et al. (2020) who demonstrate asymmetry in the Dumitrescu-Hurlin panel Granger non-causality test between globalization and FDI in Africa. We depart from Aluko et al. (2020) in that we apply the augmented mean group (AMG) estimator to ascertain the direct effect of globalization on FDI. In theory, this estimator is robust to endogeneity and employs an unobserved common factor to control for cross-sectional dependence and time-variant heterogeneity in the relationship.
Moreover, we distinguish between the de jure and the de facto measures of globalization to ascertain whether the de jure measure is only stringent on paper but ineffectual in reality, making de facto globalization more important (Kose et al., 2009).
Using the data from 47 African countries for the 1996-2016 period, we find globalization to be an important FDI driver in Africa. We decompose globalization into different dimensions and show that economic globalization exerts the most significant effect on FDI. We argue that the strength of the FDI-globalization nexus reflects the dominance of a particular globalization dimension in the host country. As such, the policymakers should reorientate their effort to improving economic globalization, which serves as a precondition for an FDI boom on the continent. Before empirically assessing the FDI-globalization nexus, we now present some stylized facts on the context of FDI inflows in Africa.

THE CONTEXT
From the outset, Africa is the world's second most populous continent and has an ample reserve of natural resources, both of which make it an attractive host location for the market-and resourceseeking FDI (Adams & Opoku, 2017;Sakyi & Opoku, 2014 What makes this trend even more striking is that by 2012 Egypt had recovered from the fall of $0.5 billion in FDI inflows during the 2011 Arab Spring (OECD, 2020). In fact, relative political stability in the subsequent 2012-2016 period saw Egypt enjoying an annual average of 24% growth rate in FDI inflows (OECD, 2020). In a latest report by RMB (2020), it identifies a large market, a sophisticated business sector, and favorable policies to be the competitive advantages that propel Egypt as the most desirable FDI destination in Africa. In the 2010-2014 period, the average inflows again increased slightly to US$50,618 million or 3.52% of the global inflows. However, the trend was reversed in the 2015-2018 period, with the average inflows of US$47,662 million or 2.83% of the global FDI inflows. In short, the average FDI inflows consistently grew for most of the 1990-2018 period, but its share never exceeded 4% of the global inflows across all periods. In part, this dismal performance reflects the MNC's perception that investing in locations with inherently weak economic institutions and macroeconomic conditions is a risky proposition (Asiedu, 2002;Dupasquier & Osakwe, 2006).
[ Table 1 here] [ Table 2 here] We now describe our research methodology, including the model, data, and preliminary analyses.

The model
In this paper, we examine the relationship between FDI and globalization in Africa through Dunning's (1978Dunning's ( , 1979Dunning's ( , 1998) eclectic paradigm of international production. In essence, the paradigm argues that FDI serves as a vehicle for a firm to combine locational advantages with its ownership and internalization advantages. Since locational advantages include the comparative and competitive advantages that a country possesses over its rivals, they are country-specific and must be controlled for when examining globalization as an FDI attractor. Therefore, we propose the following model: and significant  indicates that globalization induces FDI in Africa. To control for endogeneity and path dependency, we extend Equation (1) by including the lagged FDI as an additional To select the host country-specific factors captured by ' Z , we identify the robust determinants of MNCs' locational choice from the extant literature. For example, Du et al., (2008a,b) and Asiedu (2006) find that foreign investors prefer locations with strong institutional quality that protects private property rights in Africa. Meanwhile, Asiedu (2006) and Agbloyor (2019) show that investment uncertainty under macroeconomic instability deters FDI in Africa.
As expected, Asiedu (2006) and Agbloyor (2019) find that those resource-rich African countries attract more resource-seeking FDI. Meanwhile, Anyanwu (2012) and Asiedu (2006) (2019) show that infrastructural development, such as stable utility supplies and accessible transportation networks, stimulates FDI inflows in Africa. Following this, our control variable set includes institutional quality, macroeconomic instability, natural resources, market size, and infrastructure accessibility.

Data
5 Path dependency in FDI occurs when the level of present FDI inflows increases with the amount of accumulated FDI stock in the host location.
We utilize a panel comprising of annual observations for 47 African countries over the period 1996-2016. We are unable to extend the sample (countries and period) due to unavailability of data. For example, the governance indicator only became available from 1996 and onwards.
Similarly, the most recent publication of the globalization index was in 2016. Table A1 in the Appendix presents a list of countries included in this study.
In line with the extant literature, we measure FDI by the net inflows of FDI as a share of GDP. We obtain the series from the World Development Indicators (WDI). We capture globalization by the KOF Globalization Index constructed by Gygli et al. (2019). Unlike the traditional measures of globalization like trade openness, the KOF Index approaches globalization through a multidimensional lens by including international trade and capital flows, as well as crossborder interactions between citizens and governments. As a result, the index can be further decomposed into economic, social, and political globalization. Specifically, the economic globalization index is constructed based on information from the long-distance flow of goods, capital, services, and information and the perception about market exchanges. Meanwhile, the political globalization index is developed from information that reflects the degree of government policy diffusion. Finally, the social globalization index is built on information that indicates the spread of ideas, information, images, and people. For consistency, each index is scaled between 0-100, with a larger value representing a higher degree of globalization. Another feature of the KOF Index is that it reports the de facto and de jure measures of each globalization dimension.
Whereas, the de facto measure recognizes the realized international flows and activities relating to trade, capital, people, and information and ideas, the de jure measure encapsulates the policies, resources, conditions, and institutions that influence these realized flows and activities (Gygli et al., 2019). For completeness, we separately examine the effect of each globalization dimension as an FDI driver. We obtain the data from the KOF database administered by the Swiss Economic Institute (http://www.kof.ethz.ch/globalisation/). 6 In line with Kaufmann et al. (2011), we measure institutional quality by averaging the six governance dimensions from the World Governance Indicators (WGI). 7 The estimates of the governance indicators occupy a range between -2.5 and 2.5. Since many studies have gauged macroeconomic instability by the volatility of the inflation rate, we select the annual growth rate of the implicit GDP deflator (Feeny et al., 2014;Gui-Diby, 2014;Wisniewski and Pathan, 2014;Habyarimana and Opoku, 2018). We measure the availability of natural resources as the ratio of natural resources rent to GDP (Asiedu, 2013;Agbloyor, 2019). We capture market size by the share of the urban population, on the basis that a large urban population provides a stable labor force supply and represents a huge virgin market for MNCs (Poelhekke and van der Ploeg, 2009).
We represent financial development by the Svirydzenka (2016)   [ Table 3 here]

Preliminary analyses
Although conventional estimators typically assume stationarity during the analysis, they can generate spurious results if the variables are nonstationary (Greene, 2003). Since it is not uncommon for macroeconomic time series to exhibit nonstationarity (Hamilton, 1989), we employ the Pesaran (2007) cross-sectionally augmented Im-Pesaran-Shin (CIPS) panel unit root test.
Essentially, the CIPS test is a cross-sectional augmented Dickey-Fuller (CADF) regression model with the cross-sectional means of the lagged levels and first differences of the variable. According to Im et al. (2003), the CADF model is given by: T is the CADF for the ith cross-section unit given by the t-ratio of the coefficient of 1 it y  in Equation (4).
Apart from testing for the panel unit root, we also test for cross-sectional dependence (CSD) in the error terms, which can arise from common shocks to, and unobserved idiosyncrasies of, the cross-sectional units (De Hoyos & Sarafidis, 2006). If CSD is present in the model, standard panel data estimation could produce inconsistent estimates (Kapetanios et al., 2011). To address this concern, we select the Pesaran (2004) CSD test, which is based on the average of the pairwise correlation coefficients of the ordinary least squares (OLS) residuals from the cross-sectional unit regressions. Specifically, the CD test statistic is given by: where T is the time interval, N is the number of cross-sectional units, and ˆi j  is the pairwise correlation coefficient between cross-sectional units. [ Table 4 here]
Specifically, AMG controls for non-stationarity and accounts for the slope heterogeneity common in cross-country panels. It also controls for CSD by introducing the cross-sectional means of the unobservable factors over time into the regression.
The AMG estimator is best described by a two-stage process: where X is a vector of the independent variables consisting of lagged FDI, globalization, and the control variables, i  is the constant term, it  is the error term, ˆt  is the common dynamic process, and ˆA MG  is the AMG estimate. To carry out the AMG estimator, we begin by estimating the firstdifference ordinary least squares (OLS) model and obtain the coefficients on the year dummies, ˆt  . Next, we add ˆt  to the individual-country OLS regressions and compute the AMG estimates by averaging the coefficients on all regressions.

The benchmark results
In this section, we discuss the significant results of the FDI-globalization nexus in Africa. 9 Specifically, we only focus on the models that passed the Wald test, yielded zero-order integration in residuals, and exhibited weak CDS. 10 In Table 5, each column represents a model, and a bold number denotes a model with significant results. In general, we find that the coefficient on globalization is positive and significant in columns (1) and (2), suggesting that globalization induces FDI inflows in Africa. Our finding challenges Bitzenis (2003) who argues that MNCs' locational choice is only influenced by firm-specific motives and downplays the effect of the 9 We do not report and discuss the results of the control variables but are accessible from the authors upon request. 10 Pesaran (2015) notes that, in panel model estimation, residuals that are weakly cross-sectionally dependent do not pose serious limitation to estimation and statistical inference. However, statistical inference may be inaccurate when strong cross-sectional dependence in the residuals exists. Pesaran (2015) further notes that assuming cross-sectional independence in residuals, in Lagrange Multiplier (LM)-based CD tests, may be quite limited for large N panels and it would be more appropriate to hypothesize weak cross-sectional dependence in the residuals. Thus, due to the application of large N panels, we test for weak cross-sectional dependence in the residuals. globalization agenda on FDI. Our finding does not utterly debunk these arguments; however, we provide evidence to suggest that globalization encourages the internationalization of MNCs' activities. We argue that MNCs are likely to invest more in countries with higher flows of people, information and ideas, capital, and goods.
[ Table 5 here] Next, we examine how globalization influences FDI inflows in two subsamples; sub-Saharan Africa (SSA) sample and non-highly-globalized African countries sample. 11 We are interested in SSA because it hosted an average of 71% of the FDI inflows over the 1970-2018period (UNCTADStat, 2019. In defining the highly-globalized countries, we refer to countries with a mean globalization index exceeding the world globalization mean value of 57.62. Columns (3)-(6) in Table 5 show that overall globalization exerts a positive and significant effect on FDI in both subsamples. This finding indicates that globalization promotes FDI inflows and echoes our earlier assertion for the full-sample case. Overall, our results suggest that the positive FDIglobalization nexus in Africa is not sensitive to sample selection bias.

Unbundling globalization: de facto versus de jure measures
Next, we decompose globalization by the de facto and de jure measures of the KOF overall globalization index. Table 6 shows that the coefficients on de facto globalization are positive across all the samples, indicating a positive FDI-globalization nexus. However, these coefficients are only statistically significant at the 5-10% level when the trend component, which captures 11 We create the SSA sample by dropping Algeria, Egypt, Morocco, and Tunisia from the full sample. In terms of the non-highly-globalized countries sample, we exclude Egypt, Mauritius, Morocco, South Africa, and Tunisia. time-variant unobservables (Eberhardt, 2012;Eberhardt et al., 2013) is excluded from the estimations.
[ Table 6 here] In a similar vein, Table 7 shows that the coefficients on de jure globalization remain positive and statistically significant across all estimations and subsamples. de jure globalization attracts more FDI than its de facto counterpart. For example, in the overall sample, the coefficient on de jure globalization is 0.208 compared to 0.124 for de facto globalization. We also observe a similar pattern for the SSA countries (0.225 for de jure globalization compared to 0.186 for de facto globalization) and for the non-highly-globalized countries (0.209 for de jure globalization show that the effect of the de jure measure of globalization is more pronounced than the de facto measure. In short, the positive effect of globalization on FDI inflows is not sensitive to the de facto or de jure measure.

Unbundling globalization: Economic, political, and social dimensions
Since globalization is a multidimensional concept, we separately examine the effect of economic, political, and social globalization in this section. Table 8 shows that economic globalization exerts a positive and significant impact on FDI inflows. This finding is consistent with Majocchi and Strange (2007), who identify trade openness and capital mobility as the key determinants of FDI in a country. We also find that the coefficient on political globalization is positive and statistically significant, indicating stronger political globalization exerts positive influence on FDI inflows.
This catalytic role of political globalization supports Büthe and Milner (2008), who show that developing countries with international political engagements like international trade agreements record more FDI inflows than those which do not. Unlike the previous two dimensions, we find a negative but statistically insignificant coefficient on social globalization. The statistically insignificant impact of social-globalization that we find may follow from the fact that African countries are not well socially integrated into the world. Their cultures are not diffused into the world relative to some western countries and Asian countries such as China.
In comparing the results of the various dimensions of globalization, the effect of economic globalization is found to be more pronounced. For example, Table 8 shows that the coefficient on economic globalization (0.158) is more than 1.5 times larger than the coefficient on political globalization (0.093). The coefficient of social globalization is however statistically insignificant.
The results therefore imply that higher economic globalization is more important in attracting FDI relative to political and social globalization. Thus, the policymakers should embrace economic globalization by improving trade and financial openness, market exchanges, market competition and production, all which will not only benefit the domestic economy but also generate spillovers in attracting FDI inflows.

CONCLUSION
The advocates of globalization believe that increasing globalization is a necessary condition for attracting FDI inflows. This belief has led to many African countries embracing globalization since the 1990s. However, the evidence suggests that FDI inflows to the continent remained relatively low during this period. An attempt to understand this mismatch between belief and reality is what motivated our paper.
Using the data from 47 African countries from 1996 to 2016 and the AMG estimator, we find a significant effect of globalization on FDI. This finding remains robust to either the SSA countries or the non-highly-globalized countries sample. Undoubtedly, these findings suggest that FDI is globalization-induced in Africa. Against this backdrop, we argue that policymakers must prioritize globalization on their agenda. Moreover, our findings on de jure globalization suggest that policymakers must create a favorable perception that they are designing policies and regulations to narrow the gap between the domestic market and the global economy.
In terms of globalization dimensions, we show that economic globalization is, by far, the largest force shaping FDI inflows in Africa. This finding suggests that African countries should continue to open their borders for investment and trade. Although social globalization wields no effect on FDI inflows, it could be the case that its level is too low to jump start the FDIglobalization nexus. In other words, it would be a missed opportunity for African countries not to pursue social globalization, and in the process, improve overall globalization. We also find that political globalization fosters FDI inflows, suggesting that broadening the international engagement profile could be a useful strategy.
This paper examines the effect of globalization on FDI inflows in Africa. A worthwhile area for future research is investigating the transmission channels through which globalization influences FDI inflows to the continent.