Data and sample
The empirical analysis relies exclusively on secondary data from the World Development Indicators (WDI) database published by the World Bank. The use of WDI data is well established in the international business and development literature, particularly in studies examining the macroeconomic determinants and effects of foreign direct investment in emerging and developing economies [2, 3, 18].
The sample consists of eight major African economies—Morocco, Egypt, Tunisia, Nigeria, South Africa, Kenya, Ethiopia, and Ghana—over the period 2000–2024, subject to data availability. The panel is unbalanced, as some variables are not consistently available for all countries across the entire period. In particular, selected structural indicators, such as gross capital formation as a percentage of GDP, are missing for certain countries in some years, which results in variation in the estimation sample across model specifications. All regression tables report the exact number of observations and countries included.
These countries were selected based on three criteria: (i) economic size and regional representativeness, (ii) relevance in terms of trade and investment flows, and (iii) availability of comparable macroeconomic data over a sufficiently long time horizon. Together, they account for a substantial share of Africa’s GDP, inward FDI, and external trade, and are frequently analyzed in the literature as regional anchors or gateway economies (Table 1).
All variables are annual. Percentage and growth indicators are used to enhance cross-country comparability, while monetary values are transformed logarithmically where appropriate. The panel structure allows the analysis to exploit both within-country and overtime variation while controlling for unobserved country-specific characteristics through fixed effects. Differences in data availability across variables imply that some specifications rely on slightly reduced samples, as indicated in the regression tables.
Variables and measurement
The primary dependent variable in the baseline analysis is FDI inflows as a percentage of GDP, a standard measure in the international business and development literature for assessing the relative importance of foreign investment in host economies [5, 7]. To examine developmental outcomes, a second dependent variable—growth in GDP per capita (annual percentage change)—is employed in the extended analysis.
The key explanatory variable is trade integration. Due to incomplete availability of trade-to-GDP indicators for the full sample, trade integration is measured as the logarithm of the sum of export and import volume indices (2015 = 100). This proxy captures the real volume of trade flows and reflects countries’ participation in international markets without being mechanically influenced by GDP fluctuations. In the international business literature, trade integration is often interpreted as an indicator of market access, production scale, and embeddedness in global value chains [4, 26].
Domestic investment capacity is captured through gross capital formation as a percentage of GDP, reflecting the extent to which host economies provide complementary capital and absorptive capacity for foreign investment. Market size is proxied by the logarithm of GDP per capita (current US dollars), consistent with market-seeking motives emphasized in the international business literature [11].
To capture potential differences associated with the recent global environment, the analysis introduces a post-2021 dummy variable, which takes the value of one for years 2021 onward and zero otherwise. This indicator serves as a proxy for a global context characterized by supply chain reconfiguration, heightened geopolitical competition, and the operationalization of AfCFTA. The variable does not identify the causal effect of any single policy or event; rather, it allows the analysis to assess whether observable macroeconomic relationships differ across periods marked by distinct global conditions.
Table 2 reports the descriptive statistics for the main variables used in the empirical analysis. The figures reveal substantial heterogeneity across countries and over time, underscoring the relevance of a panel-data approach with country fixed effects. FDI inflows display considerable volatility, including episodes of net disinvestment, reflecting both global shocks and domestic dynamics. The trade integration proxy exhibits marked dispersion, indicating meaningful differences in the scale of participation in international markets. Gross capital formation and GDP per capita measures also display significant cross-country variation, consistent with differences in domestic absorptive capacity and development levels. Differences in the number of observations across variables reflect the unbalanced nature of the panel and data availability constraints.
Trade integration, measured as the logarithm of the combined export and import volume indices, exhibits substantial cross-country dispersion. The variation in trade volumes over time and across economies reflects meaningful differences in the scale of participation in international markets and global production networks. Such heterogeneity is particularly relevant for the empirical analysis, as higher trade integration may be associated with greater market access and production connectivity, factors frequently emphasized in international business theory.
Gross capital formation, used as a proxy for domestic investment capacity, averages slightly above 20 percent of GDP but also displays considerable variability across countries and years. This dispersion suggests significant differences in domestic absorptive capacity and the ability of host economies to complement foreign investment with local capital formation, consistent with the conditional perspective on FDI emphasized in the literature.
GDP per capita levels and growth rates also exhibit marked heterogeneity, with income levels spanning low-income to upper-middle-income categories and growth rates ranging from contraction episodes to periods of rapid expansion. This diversity supports the inclusion of market size and performance controls in the empirical models and cautions against treating African economies as a homogeneous investment environment.
Overall, the descriptive statistics reveal substantial cross-country and temporal variation in investment, trade integration, and economic performance. This heterogeneity provides a strong empirical rationale for the panel-data approach adopted in the subsequent analysis, which exploits within-country variation while controlling for time-invariant structural characteristics.
Figure 1 illustrates the evolution of average FDI inflows as a percentage of GDP across the sample economies over the period 2000–2024. The vertical dashed line marks the beginning of the post-2021 period. The figure highlights pronounced cyclical patterns, with episodes of expansion and contraction reflecting both global and country-specific factors. While variation across time is evident, the aggregate pattern does not suggest a clear structural break. Instead, it motivates the empirical strategy adopted in the paper, which examines whether the association between trade integration and FDI inflows differs across global contexts rather than focusing solely on aggregate trends.
While no uniform upward trend is observed over the entire period, the figure highlights fluctuations in FDI inflows around the early 2020s. Rather than indicating a generalized surge or contraction, the post-2021 period appears characterized by continued variability across countries, reflecting the broader uncertainty associated with global disruptions and geopolitical tensions.
Importantly, Fig. 1 does not point to a clear structural break in aggregate FDI inflows. Instead, it provides descriptive motivation for the empirical strategy adopted in the paper, which evaluates whether the association between trade integration and FDI inflows differs across global contexts. This approach is consistent with recent international business research emphasizing conditional relationships and mechanism-based analyses in periods of systemic change.
Empirical strategy
The empirical strategy relies on fixed-effects panel estimations to exploit within-country variation over time while controlling for time-invariant structural heterogeneity. Country fixed effects account for unobserved characteristics such as geography, institutional legacy, and long-standing development differences [8, 14]. Year fixed effects are included to capture common global shocks affecting all countries simultaneously, including commodity price fluctuations, financial conditions, and the COVID-19 pandemic.
The baseline specification is expressed as:
$$begin{aligned} {text{FDI}}_it , = & , beta_{1} {text{ TradeIntegration}}_it , + , beta_{2} , left( {{text{TradeIntegration}}_it , times {text{ Post}}2021_t} right) , & quad + , beta_{3} {text{ Capital_}}it , + , beta_{4} , ln {text{GDPpc}}_it , + , mu _i , + , lambda _t , + , varepsilon _it end{aligned}$$
where i denotes countries and t denotes years; μ_i and λ_t represent country and year fixed effects, respectively. Because year fixed effects absorb time-specific shocks common to all countries, the main effect of the Post2021 indicator is not separately identified. Accordingly, the empirical focus is on the interaction term, which captures whether the within-country association between trade integration and FDI inflows differs in the post-2021 period.
To explore the developmental implications of FDI, a second specification relates lagged FDI inflows to subsequent economic performance:
$$begin{aligned} {text{Growth}}_it , = , & alpha_{1} {text{ FDI}}_left( {it – 1} right) , + , alpha_{2} {text{ TradeIntegration}}_it , & quad + , alpha_{3} , ln {text{GDPpc}}_it , + , mu _i , + , lambda _t , + , u_it end{aligned}$$
Using lagged FDI mitigates simultaneity concerns and reflects the notion that the effects of foreign investment on growth may materialize with a temporal delay [5, 18].
Given the relatively small number of cross-sectional units (eight countries), conventional cluster-robust variance estimators may produce downward-biased standard errors. To address this issue, statistical inference for key coefficients is conducted using a wild cluster bootstrap procedure based on Rademacher weights. This approach is widely recommended for panels with a limited number of clusters and provides more reliable p values than standard cluster-robust estimators in small-sample settings. Bootstrap-adjusted p values are reported for the interaction term and other central coefficients.
All results are interpreted as conditional associations rather than causal effects.
Limitations and scope of inference
While the fixed-effects framework controls for time-invariant heterogeneity and common global shocks, the empirical strategy does not aim to establish strict causal identification. The post-2021 indicator captures a period characterized by multiple overlapping global and regional developments rather than a single identifiable policy intervention. Accordingly, the analysis evaluates whether observable macroeconomic relationships differ across global contexts, rather than attributing changes to specific causal mechanisms.
Furthermore, the limited number of countries constrains statistical power and calls for cautious interpretation of inference. The findings should therefore be understood as evidence regarding the stability or variation of the trade–FDI association within the sample, rather than definitive conclusions about structural transformation in the broader African context.
By focusing on within-country dynamics and conditional relationships, the empirical design aligns with recent calls in international business research to integrate macroeconomic context and systemic change into the analysis of investment patterns, while maintaining methodological transparency and interpretative discipline.
