where i represents the country’s index; t denotes the time-period. α0 to α12 are parameters to be estimated. μi are countries’ fixed effects; ωit is a well-behaving error term. The analysis uses an unbalanced panel dataset comprising 123 countries over the period 2002–2017. The choice of this set of countries and the period is dictated by data availability. In order to mitigate short term fluctuations and capture medium term effects of variables under analysis (in particular the variable of interest), the analysis uses non-overlapping sub-periods of 3-year average, which include 2002–2004; 2005–2007; 2008–2010; 2011–2013; and 2014–2017.
The dependent variable “FDI” is the measure of the inward foreign direct investment in stock. It can be the real values of inward FDI (constant 2010 US$ prices), computed by multiplying the inward Foreign Direct Investment stock (% of GDP) by the real GDP (constant 2010 US$) (e.g., Nagel et al., 2015 and Herzer, 2011). This variable is denoted “FDICST”. For robustness check, the variable “FDI” is also measured by “FDIGDP”, which is the inward FDI stock, expressed as a percentage of the recipient-country’s GDP.
“AfT” is the real gross AfT disbursement that accrues to a country. It can be either the total real gross AfT disbursements - denoted “AFTTOT” - or its components. The latter include AfT for economic infrastructure, denoted “AfTINFRA”, AfT for building productive capacity, denoted “AfTPROD”, and AfT allocated for trade policies and regulations, denoted “AfTPOL”. All AfT variables have been expressed in constant prices 2016, US Dollar.
“EXCONC” is the index of export product concentration. The latter is primarily measured by the Herfindahl-Hirschmann Index of export product concentration computed by the United Nations Conference on Trade and Development (UNCTAD). Its values range between 0 and 1, with higher values reflecting a higher export concentration on a few products. In contrast, values of this index closer to 0 reflect a homogenous distribution of exports among a series of products. For robustness check analysis, we use the index of export product diversification computed as the absolute deviation of a country’s export structure from world’s export structure, based on the modified Finger-Kreinin (1979) measure of similarity in trade. This indicator, denoted “FKIEDI”, takes values between 0 and 1, with values closer to 1 reflecting a greater divergence of a country’s export structure from the world’s export structure. Values of the indicator closer to 0 indicate a greater convergence of a country’s export structure towards the world’s exports structure.
All variables contained in model (1) have been standardized so as to avoid measurement problems of variables, and make comparable the related estimates (arising from the estimations). The standardized procedure involves, for each variable, the computation of the ratio of the difference between the variable and its mean (average) to the standard deviation of this variable. This procedure leads to the removal of time dummies from model (1) as their standardized values amount to zero.
Control variables are drawn from the literature of the macroeconomic determinants of inward FDI and consist mainly of those determinants that are likely to influence the impact of AfT on inward FDI. These variables include: the host-countries’ market size proxied by the host-countries’ real their capita income (which simultaneously acts as a proxy for the country’s development level, denoted “GDPC”), the host-country’s economic growth rate (denoted “GDPGR”) and the host-country’s size of their population (denoted “POP”). Other control variables include the level of human capital accumulated, proxied by the education level, i.e., the secondary school enrolment rate; trade policy (denoted “TP”), and the institutional and governance quality (denoted “INST”). Indeed, many studies (e.g., Chakrabarti, 2001; Asiedu, 2002; Busse and Hefeker, 2007; Vo and Daly, 2007 and Boateng et al. 2015) have underlined that domestic market size of a host-country influences inward FDI to this country. Likewise, many studies (recent ones include for example, Asiedu, 2006, Trevino et al. 2008; and Okafor et al. (2015) have shown that human capital accumulation influences positively Inward FDI. The impact of trade policy liberalization/or trade openness (which is an outcome of several policies, including trade policy liberalization) on inward FDI has been largely investigated in the literature. This impact depends on the types of inward FDI that enter a country. For example, according to the tariff-jumping hypothesis, higher tariffs (or higher trade restrictive measures) would attract horizontal type Inward FDI, which aim to be protected from import competition in the host-country (e.g., Markusen, 1984; Markusen and Venables, 1995). At the same time, vertical Inward FDI are likely driven by trade policy liberalization (e.g., Helpman and Krugman, 1985). Similarly, export platform FDI and complex-vertical Inward FDI increase in the context of trade policy liberalization (e.g., Fugazza and Trentini, 2014). Many empirical analyses (e.g., Mina, 2007; Trevino et al. 2008, Asiedu and Lien, 2011; Boateng et al. 2015; Gnangnon, 2017; Gnangnon and Iyer, 2017) have uncovered a positive impact of trade liberalization (or trade openness) on Inward FDI. Overall, while a positive impact of trade policy liberalization on inward FDI could be expected in the current analysis, it is worth emphasizing that a negative or a statistical nil impact could also be obtained, because the observed impact might reflect the impact of trade policy on the different types of Inward FDI.
Let us turn to the expected effect of NonAfT flows on inward FDI. As noted above, Selaya and Sunesen (2012) have found that development aid allocated for the enhancement of human capital is associated with higher inward FDI. As this type of aid is included in the category on NonAfT flows, one can expect a rise in NonAfT flows to be associated with a rise in inward FDI. As NonAfT flows are primarily invested in the non-tradable sector, they might affect inward FDI through their effect on the real exchange rate. On the one hand, development aid flows (including flows invested in the non-tradable sector) can be associated with the appreciation of the real exchange rate (e.g., Addison and Baliamoune-Lutz, 2017; Adu and Denkyirah, 2018; Elbadawi, 1999; Ouattara and Strobl, 2008). As an appreciation of the real exchange rate can lead to lower FDI inflows (e.g., Caves, 1989; Blonigen, 1997; Froot and Stein, 1991; and Vijayakumar et al. 2010), one can expect NonAfT flows to be associated with lower FDI inflows.
With respect to the inward FDI effect of institutional and governance quality, there is a large consensus in the literature that good institutional and governance quality promotes inward FDI (e.g., Busse and Hefeker, 2007; Bevan et al. 2004; Ali et al., 2010; Buchanan et al., 2012).
In all regressions performed with the two-step system GMM estimator, variables measuring “AfT” and the variables “TP”, “INST” are considered as endogenous, whereas the variables “EDU” and GDPGR” have been considered as predetermined. The other variables have been considered as exogenous.
We provide an insight into the relationship between “AfTTOT” and “FDICST” variables (unstandardized variables), by using non-overlapping sub-periods of 3-year average data for the period 2002–2017 over the full sample. Thus, we present in Figure 1 the developments of “AfTTOT” and “FDICST”. Figure 2 shows the correlation pattern (in the form of cross-plot) between total AfT flows and the real inward FDI stock. It can be noted from Figure 1 that total AfT flows and real inward FDI have moved in the same direction. Total AfT flows increased, on average over the full sample, from US$ 93.8 million in 2002–2004 to US$ 272 million in 2014–2017. At the same time, real FDI moved from US$ 2060 billion in 2002–2004 to US$ 4910 billion in 2014–2017. Both the left-hand side graph (based on unstandardized variables) and the right-hand side graph (based on standardized variables) of Figure 2 show a positive correlation between total AfT flows and real inward FDI stock. However, the outliers present on the left-hand side graph do not appear on the right-hand side graph.
Overall, the empirical analysis proceeds as follows. First, we estimate variants of model (1) without the variable “ECI” and its interaction with the variable “AfT”. The variants of model (1), therefore, include total AfT flows as well as its components. These variants of model (1) merely aim to examine the effect of AfT flows variables on real inward FDI stock. The results of these estimations are presented in Table 1. Second, we estimate model (1) as it stands, including with both total AfT as well as its components (this leads to the estimation of several specifications of model (1)). The results of these different model specifications with the variable “ECI” (as measure of “EXCONC”) are reported in Table 2, whereas results of these model specifications with “FKEDI” as measure of “EXCONC” are provided in Table 3. Finally, we use “FDIGDP” as the measure of “FDI” and estimate different other specifications of model (1), including with total AfT flows and the components of the latter. The outcomes of these different specifications of model (1) are displayed in Table 4.
Appendix 1 provides the description and source of all variables used in the analysis, while Appendix 2 presents the list of the 123 AfT recipient-countries. Appendix 3 reports descriptive statistics on unstandardized variables.