The middle-income trap and foreign direct investment: a mixed-methods approach centered on Mexico and South Korea

The paper addresses how foreign direct investment (FDI) affects the ability of middle-income (MI) countries to transition to high-income (HI) status, a challenge known as the middle-income trap (MIT). The MIT refers to countries remaining in MI status for extended periods (e.g., >28 years at lower MI or >14 years at upper MI, or >42 years in MI overall). As MI countries lose low-wage advantages but have not yet upgraded to high value–adding industries, industrial upgrading becomes central to escaping the MIT. Given the rise of global supply chains and large FDI flows into MI countries, the study investigates whether FDI facilitates or hinders the industrial upgrading needed to reach HI status. The research proposes two competing hypotheses: (1) higher FDI inflows are associated with a greater likelihood of remaining trapped; (2) higher FDI inflows are associated with a greater likelihood of escaping the trap.

The literature specific to FDI and the MIT is limited, though broader work on FDI and industrial upgrading is larger. From a World-Systems Theory perspective, several works argue that FDI can impede upgrading in MI countries. Zeng and Fang (2014) suggest globalization reinforces core–periphery dynamics: MNCs allocate low-margin activities to developing countries, retain high-profit functions at home, and thus limit local upgrading; FDI may also increase vulnerability and suppress local firm emergence. Anastasi (2023) similarly argues MNCs protect technology and limit spillovers to maintain quasi-monopolistic positions, impeding emergence of national champions; increasing FDI can depress local R&D spending (Azman-Saini et al., 2010; Anastasi, 2024a). Doner and Schneider (2016) and Henley (2018) discuss how heavy reliance on FDI in MI contexts can crowd out domestic capability formation, particularly in Southeast Asia. This strand motivates Hypothesis 1: higher FDI inflows (as % of GDP) increase the likelihood of being trapped in MI status. Conversely, other empirical studies find beneficial roles of FDI. CM et al. (2024), using BMA and GMM, report that FDI correlates positively with transitions from MI to HI and accelerates transition speed in Cox models. Tampakoudis et al. (2017) examine determinants of FDI inflows to MI countries (assuming FDI helps avoid the MIT), finding openness and growth attract FDI, while population growth deters it. Nguyen-Huu and Pham (2021) find FDI positively correlates with long-run GDP per capita growth (especially in upper MI), though short-run effects can be negative. This strand motivates Hypothesis 2: higher FDI inflows (as % of GDP) reduce the likelihood of being trapped in MI status.

Design: Mixed-methods approach combining (1) a comparative case study of the automotive sectors in Mexico and South Korea (1960s–2000s) and (2) quantitative logistic regressions. Comparative case study: Mexico and South Korea were selected for similar initial development levels in the early 1960s and comparable population sizes then. The analysis focuses on how differing FDI acceptance models affected industrial upgrading. Mexico’s policy allowed broad MNC entry (Ford, GM, Nissan) and numerous local assemblers, undermining economies of scale and leading to the failure of Automex (a potential national champion). South Korea initially restricted entry, promoted local content, and emphasized technology acquisition via licensing and selective partnerships. Hyundai pursued indigenous capability building (e.g., the Pony project), diversified foreign technology sources, and invested heavily in R&D, evolving into a global contender. Quantitative data: World Bank DataBank (World Development Indicators). Sample comprises eight countries: four breakout cases (South Korea, Uruguay, Greece, Japan) and four trapped cases (Mexico, Turkey, Malaysia, Brazil). Period: 1970–2000 (start constrained by data availability; end chosen to limit years when breakouts had already reached HI). Outcome (MITO): binary, coded 1 for countries that escaped the MIT and 0 otherwise; coded constant over time per country. Key predictor: FDI net inflows as % of GDP (FDI%G). Controls: Gross capital formation (% GDP, GCF), age dependency ratio (ADR), exports of goods and services (% GDP, E), manufacturing value added (% GDP, MVA). Descriptive statistics and correlations indicate no severe multicollinearity. Models: Cross-sectional logistic regressions using the panel of annual observations with country outcome fixed across years. Four specifications: Model 1 includes only FDI%G; Model 2 adds GCF and ADR; Model 3 adds E and MVA; Model 4 includes all controls. Robustness: Subsample analyses by region—East Asia (South Korea, Japan, Malaysia) and Latin America (Mexico, Brazil, Uruguay)—retest Model 1 and Model 2 structures. Estimation and fit metrics: Odds ratios (ORs) with 95% confidence intervals reported; model fit via McFadden and Nagelkerke pseudo-R² and AIC/BIC. Figures depict logistic curve and marginal effects of FDI%G.

Comparative case study: - Mexico: Allowing multiple MNCs and many assemblers prevented economies of scale; lobbying by MNCs (notably Ford) influenced policy away from nurturing a national champion (Automex), shifting focus toward export/balance-of-payments objectives. Production rose but value-added content and domestic capability lagged; over time, Mexico specialized in labor-intensive assembly with declining wages and domestic value added in autos. - South Korea: Restrictive entry, strong local content requirements, technology licensing, and later indigenous R&D enabled Hyundai to develop original models (the Pony), expand exports, and become globally competitive. The sector upgraded technologically and structurally, with Korea becoming a top global producer and exporter. Quantitative results (logistic regressions): - Model 1 (FDI%G only): OR = 0.36 (95% CI: 0.24–0.51); higher FDI%G is associated with lower odds of escaping the MIT. Fit: McFadden R² ≈ 0.16; Nagelkerke R² ≈ 0.20. - Model 2 (FDI%G + GCF + ADR): FDI%G OR = 0.42 (95% CI: 0.27–0.60), confirming the negative association. ADR shows a significant negative association with escaping (OR ≈ 0.81; 95% CI: 0.76–0.86). GCF not significant. Improved fit (lower AIC/BIC; higher pseudo-R²). - Model 3 (FDI%G + E + MVA): FDI%G OR = 0.12 (95% CI: 0.05–0.24). Exports (E) positively associated with escaping: OR = 1.07 (95% CI: 1.04–1.11). MVA not significant. - Model 4 (all controls): FDI%G OR = 0.22 (95% CI: 0.08–0.45), maintaining a strong negative association with escaping; model exhibits good fit (AIC ≈ 118.9; BIC ≈ 138.3; McFadden ≈ 0.55; Nagelkerke ≈ 0.70). Other controls generally weak or inconsistent: ADR sometimes significant; GCF and MVA not significant; E loses significance once all controls are included. Robustness (subsamples): - East Asia tests: FDI%G remains significant and negative (e.g., OR ≈ 0.10–0.16). - Latin America tests: FDI%G significant and negative with smaller magnitude (e.g., OR ≈ 0.33–0.49).

Findings consistently indicate that higher FDI inflows (as % of GDP) reduce the likelihood of escaping the middle-income trap. This supports Hypothesis 1 and leads to rejection of its null; Hypothesis 2 is not supported. The results align with World-Systems Theory perspectives that global capital flows can entrench peripheral roles by situating low-margin activities in MI countries and retaining high-value functions in core economies. The comparative cases corroborate the statistical results: in Mexico, deep MNC penetration hindered the rise of a national champion and limited upgrading, while South Korea’s selective engagement with foreign technology and emphasis on indigenous capability fostered upgrading and successful export integration. The mechanism suggested is that large FDI presence can suppress the development of domestic firms capable of high value–adding activities—through both competitive and political channels. Model fit indices (lower AIC/BIC; higher pseudo-R² in enriched models) indicate the robustness of the relationship, and subsample analyses confirm its persistence across regions. The study’s implications challenge work that finds uniformly positive roles for FDI and suggest a more contingent, policy-mediated relationship between FDI and structural transformation.

The study contributes by integrating qualitative and quantitative evidence to assess how FDI relates to escaping the middle-income trap. The Mexico–South Korea comparison shows that FDI-heavy, MNC-dominated strategies can impede national champion formation and upgrading, whereas selective technology acquisition and indigenous capability building can succeed. Logistic regressions on eight countries (1970–2000) consistently show a strong negative association between FDI inflows and the odds of escaping the MIT, with results robust across specifications and regional subsamples. Policy recommendations include exercising caution in accepting FDI within strategic high value–adding sectors; prioritizing technology purchases and licensing; and using joint ventures designed to ensure technology transfer. Future research should identify conditions under which FDI can support upgrading (e.g., market size, bargaining power, JV design, IP regimes) and refine measurement and identification strategies to parse mechanisms.

- External validity and scope: While the qualitative analysis focuses on one sector (automotive) and two countries, the quantitative sample includes only eight countries with outcomes coded constant over time (1970–2000), limiting generalizability. - Measurement and design: The outcome (MITO) is time-invariant at the country level across multiple years, which may overweight countries with longer time series. Potential inconsistencies between tabulated coefficients and narrative interpretations of some controls suggest caution. - Mechanisms: The study infers, but does not directly identify, mechanisms by which FDI affects upgrading (e.g., crowding out national champions, limiting spillovers). Cases such as China’s JV-led upgrading indicate heterogeneous effects requiring further investigation. - Sectoral heterogeneity: Effects may differ across industries and types of FDI (greenfield vs. M&A; assembly vs. R&D-intensive activities), which are not disaggregated here.