The dynamic relationship between trade openness, foreign direct investment, capital formation, and industrial economic growth in China: new evidence from ARDL bounds testing approach

The study investigates how trade openness (TRO), foreign direct investment (FDI), and capital formation (K) relate to industrial economic growth (IEG) in China over 1990–2021. Motivated by mixed international evidence on whether FDI and trade openness promote growth, and recognizing China's rapid growth alongside rising FDI inflows and trade integration, the paper analyzes whether and how these factors jointly drive China's industrial economy. It addresses gaps in prior China-focused research that often used aggregate or regional perspectives and overlooked industrial-sector dynamics. The authors posit two hypotheses: (1) Trade openness and FDI positively affect China's industrial economic growth; (2) Trade openness, FDI, and capital formation synergistically contribute to industrial economic growth in China, interacting with external spillovers such as inflation, labor, and technological innovation. To test these, the paper applies ARDL/ECM bounds testing to evaluate long- and short-run dynamics among TRO, FDI, K, and IEG, with controls for inflation, labor, and technological innovation.

The literature emphasizes scientific and technological progress as essential to sustainable growth, with international trade and FDI serving as key channels for knowledge diffusion and spillovers. Studies such as Mohammed and Ruslee (2015) and Kathuria (2001) argue FDI boosts domestic productivity and generates positive spillovers, including technology transfer and human capital gains. Liu et al. (2002) find mutual reinforcement among growth, trade, and FDI in China, but caution that ignoring interactions can produce negative spillovers. Blyde (2004) highlights trade’s role in technology diffusion. Other work observes mixed effects: some find trade openness and FDI promote growth and technological progress (Kalai and Zghidi, 2019), while others report insignificant or negative impacts depending on macroeconomic stability, government size, and capital openness (Sadni-Jallab and Gbakou, 2009; Ali and Mingque, 2018; Vikas, 2014). Several studies stress that trade alone is insufficient and must be integrated with domestic reforms and absorptive capacity. Overall, evidence is inconclusive, justifying further investigation for China’s industrial sector and the roles of trade, FDI, and capital formation within it.

Framework: The study extends a Cobb–Douglas production function to model industrial economic output (IEG) as a function of trade openness (TRO), foreign direct investment (FDI), capital formation (K), and controls (X) for inflation, labor, and technological innovation: IEG = A f(TRO, FDI, K, X). Following Ramirez (2000), FDI stock effects are emphasized. Econometric approach: Autoregressive Distributed Lag (ARDL) bounds testing is employed to assess long-run cointegration and short-run dynamics among IEG, TRO, FDI, K, and controls. The paper specifies ARDL models for each endogenous variable and corresponding error-correction models (ECM) to capture adjustment to equilibrium. Due to the relatively small sample, Narayan (2005) critical values are used for bounds testing. Lag lengths are selected via information criteria (SIC/AIC), with VAR-based lag diagnostics indicating p=1 as maximum in preliminary analysis. Data: Annual data for China from 1990–2021. Sources: World Bank (IEG: industrial value added % of GDP; FDI: % of GDP; K: capital formation % of GDP; L: labor force, % of total population; TRO: trade openness % of GDP; INF: CPI annual % change) and National Bureau of Statistics of China (TI: technological innovation proxied by patent applications, in ten-thousands). All variables are transformed into natural logarithms to stabilize variance and interpret elasticities. Pre-estimation tests: Unit root tests (ADF and Phillips–Perron) indicate variables are I(1) at conventional significance levels. Given I(0)/I(1) mixture suitability, ARDL bounds testing proceeds. Model selection uses SIC; diagnostic tests include normality, ARCH, RESET, serial correlation. Robustness and stability: CUSUM and CUSUMSQ tests are used to assess parameter stability. Results indicate stability for IEG and FDI models; some instability (CUSUMSQ exceeding 5% bands) for TRO and K models, likely associated with shocks around the 2008 global financial crisis affecting trade and capital formation. Model specifications: Reported long-run ARDL structures include ARDL(1,2,0,1,2,1,2) for IEG and ARDL(1,0,2,1,2,1,2) for FDI (with controls L, TI, INF), alongside simpler structures for TRO and K. Short-run ECM terms (ECM(-1)) capture speed of adjustment.

- Cointegration: ARDL bounds tests show long-run cointegration among IEG, TRO, FDI, K, and controls, with F-statistics exceeding upper critical bounds (e.g., IEG: 7.295; FDI: 10.590; TRO: 6.445; K: 4.373). - Long-run effects (Table 7): - FDI → IEG: Positive and significant (1%); a 1% increase in FDI is associated with a 0.071% increase in IEG. - IEG → FDI: Positive and significant (1%); coefficient ~11.613, supporting a strong feedback effect. - IEG → TRO and K: Positive and significant (5%); industrialization promotes trade openness and capital formation. - TRO → IEG: Significant negative effect; contrary to the initial positive hypothesis, indicating potential adverse or insignificant long-run impact of openness on industrial growth in this sample. - Controls: Labor (L) and inflation (INF) positively affect IEG (both 1%), but negatively affect FDI (both 5%). INF positively affects TRO (10%). Technological innovation (TI) positively affects K (10%). - Short-run dynamics (Table 8): - ECM(-1) is negative and highly significant across models, with adjustment speeds of approximately 82.7% (IEG), 102.7% (FDI), 101.4% (TRO), and 69% (K), implying rapid reversion to equilibrium (about 1.2–1.4 years for IEG and K; within-year for FDI and TRO). - In the short run, TRO, FDI, and K positively affect IEG and vice versa; TRO and K positively affect FDI, while FDI negatively affects K (weakly, at 10%). FDI positively affects TRO; TRO negatively affects K (both not statistically significant). - Controls: In the short run, L, TI, and INF generally exert negative effects on IEG and FDI; INF positively affects TRO; L negatively affects K, while TI positively affects K. - Diagnostics and stability: Normality, ARCH, RESET, and serial correlation tests generally support model adequacy. CUSUM and CUSUMSQ indicate parameter stability for IEG and FDI models; TRO and K show some instability consistent with the 2008 crisis shock.

Findings corroborate a bidirectional (feedback) relationship between FDI and industrial economic growth, indicating that FDI inflows enhance industrialization through capital, technology, and productivity channels, while a growing industrial sector attracts further FDI. The positive long-run impact of industrialization on trade openness and capital formation supports the notion that industrial development deepens integration into global markets and stimulates investment. However, the significant negative long-run effect of trade openness on IEG suggests openness alone may not guarantee industrial growth benefits without adequate absorptive capacity, policy frameworks, or sectoral upgrading—consistent with evidence of mixed openness–growth relationships in the literature. Short-run results reinforce complementary roles among TRO, FDI, and K in supporting IEG, but also indicate potential crowding-out or competition effects of FDI on domestic capital formation. Control variables highlight that inflation can stimulate industrial activity and openness but may deter FDI, while technological innovation aids capital formation. Overall, the results address the research question by demonstrating robust long- and short-run linkages among TRO, FDI, K, and IEG, largely supporting Hypothesis 1 and partly supporting Hypothesis 2, with qualifications regarding the openness–IEG link in the long run.

The study integrates trade openness, FDI, capital formation, and industrial economic growth into an ARDL bounds-testing framework for China (1990–2021). It documents long-run cointegration and dynamic feedback among these variables. Key contributions include: (1) evidence that FDI significantly promotes industrial growth and vice versa; (2) industrialization fosters trade openness and capital formation; (3) trade openness exhibits a negative long-run association with industrial growth in this setting; and (4) nuanced roles of inflation, labor, and technological innovation, with TI supporting capital formation. Policy implications include expanding and facilitating FDI (streamlining foreign currency administration and settlement), enhancing trade openness via tax relief and infrastructure/trade services to attract investment and technology, and addressing regional imbalances in openness and FDI by linking industrial resources across regions and promoting industrial transfer to central/western areas. Strengthening capital market functioning can enhance resource allocation and sustain industrial development. Future research should refine models to capture regional heterogeneity within China, explore sectoral dynamics, and assess post-crisis structural shifts affecting trade and capital formation.

- The analysis uses national-level annual data and does not account for regional heterogeneity within China, which may bias results given significant east–west disparities in FDI and trade openness. - Sample size is limited (1990–2021), necessitating reliance on Narayan (2005) small-sample critical values and potentially limiting power. - Stability diagnostics indicate parameter instability for TRO and K models (CUSUMSQ), likely related to the 2008 global financial crisis, which may affect generalizability across subperiods. - Use of aggregate measures (e.g., FDI % of GDP, TRO % of GDP) may mask sectoral composition effects and differences between FDI stock vs. flow. - Potential measurement and endogeneity issues remain despite ARDL/ECM framework and inclusion of controls (L, INF, TI).