Foreign direct investment, total factor productivity, and economic growth: evidence in middle-income countries

The paper situates sustainable economic growth within SDG 8 and argues that growth expands employment, incomes, and fiscal capacity for investment in technology, infrastructure, and human capital. Given mixed prior evidence on how foreign direct investment (FDI) and total factor productivity (TFP) affect growth, the study focuses on middle-income countries, which are diverse in economic structures and are often in transition, facing competitiveness and sustainability challenges. Two main research questions are posed: (1) Does the interaction between FDI and TFP positively impact GDP in middle-income countries? (2) Are prior findings from developed or low-income contexts robust in middle-income countries? Using World Bank data for 90 middle-income countries (1990–2020) and dynamic system GMM to address endogeneity, the study also conducts robustness tests using alternative growth proxies and pre-/post-financial crisis subsamples. The authors preview findings that FDI, TFP, and their interaction positively relate to growth, supporting economic growth and industrialization theories while not supporting labor market dynamics theories.

The literature review is structured around theories and empirical evidence. Economic growth theories emphasize GDP/GNP growth as core indicators, expecting positive impacts of both FDI and TFP on GDP through capital deepening, technology transfer, and efficiency gains. Industrialization theories argue that transitioning from agriculture to industry enhances productivity, technology, human capital, and TFP, thereby promoting growth, and also attracting FDI that further supports industrial expansion. Labor market dynamics theories present mixed effects of FDI, noting potential adverse impacts if FDI displaces local labor, depresses wages, or fails to transfer technology effectively. Empirically, studies find both positive and negative FDI–growth relationships in middle-income settings; similarly, TFP is often growth-enhancing but may falter under conditions like the Solow productivity paradox or misallocation. Prior work also suggests FDI–TFP complementarities via technology transfer and HR development, though concentration of FDI may harm sectoral diversity and TFP. Based on mixed evidence, the authors propose: Hypothesis 1: FDI positively impacts economic growth in middle-income countries. Hypothesis 2: TFP positively affects economic growth in middle-income countries. Hypothesis 3: The interaction between FDI and TFP positively affects economic growth in middle-income countries.

Data: Using the World Bank’s 2010 classification, the study compiles an unbalanced panel for 90 middle-income countries from 1990–2020. Variables are drawn from World Development Indicators and TFP is constructed via a Cobb–Douglas production function (Y = TFP × K^α × L^β) following prior studies. Observations with missing data are excluded; all variables are winsorized at the 5th and 95th percentiles to mitigate outliers, yielding 2714 observations (GMM estimations use 2534 due to lags). Variables: Dependent variable: economic growth (GDP). Main independent variables: FDI (net inflows as % of GDP) and TFP. Controls include agriculture share (AGRI), gross capital formation (GCF), government spending (GOV), imports (IMP), and inflation (INF). Multicollinearity is assessed via VIFs (< 5 for all variables). Models: Four baseline panel specifications are estimated: (1) GDP_it = α + β1 FDI_it + β2 CONTROL_it + α_i + α_t + ε_it; (2) GDP_it = α + β1 TFP_it + β2 CONTROL_it + α_i + α_t + ε_it; (3) GDP_it = α + β1 FDI_it + β2 TFP_it + β3 CONTROL_it + α_i + α_t + ε_it; (4) GDP_it = α + β1 FDI_it + β2 TFP_it + β3 (FDI_it×TFP_it) + β4 CONTROL_it + α_i + α_t + ε_it. Estimation strategy: Initial OLS/REM/FEM estimates are guided by Hausman and redundant fixed effects tests; Wald tests indicate heteroskedasticity. Endogeneity is assessed using a Durbin–Wu–Hausman framework with residual inclusion tests, revealing endogeneity in several regressors (FDI, TFP, GCF, INF). The preferred approach is dynamic system GMM (Arellano–Bond; Blundell–Bond) with cross-section fixed effects, using lagged endogenous variables (FDI, TFP, GCF, INF) as instruments. Diagnostic checks include AR(1)/AR(2) serial correlation tests and Hansen J-tests for overidentification. Robustness checks: (i) Alternative growth proxies: ln(GNP) and ln(GDP per capita). (ii) Subsamples pre- (1990–2007) and post- (2008–2020) the global financial crisis, estimated with GMM. Instrument ranks and diagnostic statistics are reported for all GMM models.

- Descriptive statistics: Average GDP growth is 4.05%; mean FDI is 3.80% of GDP (SD 6.43); average TFP equals 1.12. Correlations are modest (all < 0.55); VIFs indicate no multicollinearity issues (max 1.59). - Main GMM results (with cross-section fixed effects): • Dynamic persistence: GDP(-1) positive and significant across specifications (e.g., 0.105–0.148). • FDI positively affects growth. In the full model, a 1% increase in FDI associates with a 0.093 increase in GDP growth (interpreted as 9.3%). • TFP strongly and positively relates to growth; in the full model, a 1-point rise in TFP associates with a 16.378 percentage-point increase in growth. • Interaction FDI×TFP is positive and significant (coefficient ≈ 0.168), indicating complementary effects on growth. • Controls: AGRI (≈ 0.752) and GCF (≈ 0.241) positively relate to growth; GOV (≈ −0.296), IMP (≈ −0.050), and INF (≈ −0.006) are negatively associated with growth. • Diagnostics: AR(2) tests are insignificant (no second-order autocorrelation); Hansen J-tests are not rejected across models (no overidentification issues). - Robustness (alternative proxies): Using ln(GNP) and ln(GDP per capita) confirms positive effects of FDI and TFP. The FDI×TFP interaction remains positive for GDP and GDP per capita, but is negative for ln(GNP). - Robustness (pre-/post-crisis): Before the crisis (1990–2007), FDI negatively relates to growth while FDI×TFP is positive; after the crisis (2008–2020), FDI is strongly positive and FDI×TFP turns negative. TFP remains positive in both periods. Controls generally retain expected signs. - The results support Hypotheses 1 and 2 and, in the main specification, Hypothesis 3; interaction effects vary by proxy and period.

The findings indicate that in middle-income countries, FDI contributes to growth by expanding markets, augmenting capital, scaling production, and facilitating access to advanced technologies, aligning with economic growth and industrialization theories. TFP enhances growth through improved labor quality, technology transfer, and production process innovations. The positive FDI×TFP interaction suggests complementarities: FDI can catalyze technology diffusion and human capital development, strengthening productivity gains. However, heterogeneity appears by period and proxy. Prior to the global financial crisis, FDI’s concentration in select sectors may have dampened overall growth despite positive interaction with TFP, while in the post-crisis period, FDI became growth-enhancing but its interaction with TFP turned negative, possibly reflecting weaker technology transfers, sectoral concentration, or crisis-induced management and absorption challenges. Control variables corroborate established channels: agriculture and investment bolster growth, whereas excessive government spending, import dependence, and inflation hinder it. Overall, the results substantiate the importance of both external capital and domestic productivity capacity, and they partially contradict labor market dynamics theories that predict adverse employment and wage effects from FDI dominating growth outcomes.

The study examines how FDI, TFP, and their interaction affect economic growth across 90 middle-income countries (1990–2020) using dynamic system GMM to address heteroskedasticity and endogeneity. It finds that FDI and TFP each positively affect growth and that their interaction is growth-enhancing in the main specification, supporting economic growth and industrialization theories and not supporting labor market dynamics theories. Robustness checks using alternative growth measures confirm the main effects, with some variation in the interaction term. Pre-/post-crisis analyses show FDI’s positive effect is stronger post-crisis, while the FDI×TFP interaction was positive pre-crisis and negative post-crisis. The paper contributes by explicitly modeling and testing the FDI–TFP interaction in a broad middle-income sample and by providing policy-relevant evidence on conditions under which external investment and productivity improvements translate into growth.

- Methodological: Dynamic system GMM cannot distinguish short- from long-run effects; ARDL-type approaches could complement future analyses. - Scope: The sample is restricted to 90 middle-income countries based on the 2010 World Bank classification; results may not generalize to low- or high-income countries. - Future research: Compare across income groups; employ ARDL or related methods to parse short- and long-run dynamics; further explore mechanisms behind the time-varying sign of the FDI×TFP interaction and sectoral heterogeneity.