Foreign direct investment and the innovation performance of local enterprises

China has attracted substantial foreign direct investment (FDI), particularly after accession to the WTO, becoming the second-largest FDI recipient globally. While FDI is known to affect firm-level outcomes such as markups, productivity, and employment, its effect on local firms’ innovation is debated. Theory suggests two main channels: (1) positive spillovers via technology, management practices, demonstration and learning effects, and skilled labor mobility that can raise domestic firms’ R&D and innovation; and (2) competition effects that can either stimulate innovation through escape-competition incentives or dampen it via Schumpeterian erosion of rents, depending on market competition intensity. The research question is whether and how FDI affects the innovation performance of Chinese local manufacturing firms, and under what conditions and for which types of firms these effects arise. The study is important for understanding how openness and multinational presence shape domestic innovative capacity and for informing policies to maximize beneficial spillovers.

Prior empirical work offers mixed evidence. Several studies document positive links between FDI/foreign ownership and domestic innovation through spillovers (Lai, 1998; Girma et al., 2009; Cheung and Ping, 2004; Vahter, 2011; Sandu and Ciocanel, 2014; Crescenzi et al., 2015; Olabisi, 2017; Gorodnichenko et al., 2020). Others find insignificant or negative effects, or imitation rather than innovation (Buckley et al., 2002; Brambilla et al., 2009; García et al., 2013; Chen and Zhang, 2019). Discrepancies stem from differences in methods, data (often industry-level), and innovation measures (e.g., TFP, R&D, new product output), with TFP only weakly related to innovation outputs (Crépon et al., 1998). Spillovers are also conditional on domestic firms’ absorptive capacity and heterogeneity (e.g., ownership type), which shape their ability to learn from and compete with foreign entrants (Khachoo et al., 2018). This study addresses these gaps with firm-level microdata, direct innovation output measures, and explicit mechanisms and heterogeneity analyses.

Data: Annual Survey of Industrial Firms (ASIF), National Bureau of Statistics of China, covering SOEs and above-scale firms (turnover >5 million RMB), ~90% of manufacturing output, 2000–2007. Sample restricted to manufacturing; data cleaned by dropping missing key variables, firms with <8 employees, and accounting inconsistencies. To focus on local firms, enterprises with foreign capital >50% of registered capital are excluded. Patent data (1985–2013) from China Intellectual Property Office are matched for robustness checks.

Innovation measure (dependent variable): log of the output value of new products at the firm-year level (primary); robustness uses log of total patent applications.

Key regressor (industry-level FDI exposure): Following Javorcik (2004), Lu et al. (2017), Lin et al. (2020), construct FI_jt as an output-weighted measure of foreign capital penetration in 4-digit industries: aggregate across firms the product of each firm’s foreign-capital share and its output, divided by total industry output. Robustness constructs alternative FI using sales-weighting (FI_S) and value-added-weighting (FI_A).

Controls: Firm size (log sales), factor intensity KI (log capital-labor ratio; capital = net fixed assets deflated; labor = average employees), average wage (log total wages/employees), firm age (log years since establishment), government subsidies (subsidies/sales), and industry concentration HHI (4-digit).

Baseline specification: Panel regressions of firm innovation on FI with firm, year, and province fixed effects; alternative specifications include industry fixed effects. Standard errors clustered appropriately. Interpretation: coefficient on FI reflects association between industry-level foreign penetration and firm innovation output.

Endogeneity and robustness: Reverse causality is mitigated by industry-level FI vs. firm-level outcomes. Instrumental variables (2SLS) used to address potential omitted variables: (i) log number of foreign-funded enterprises at industry level; (ii) initial-period foreign investment (Ahsan, 2013). Additional robustness: alternative dependent variable (patents), alternative FI weights (sales, value added), and subsamples excluding pre-WTO period and retaining only post-2005 exchange-rate-reform years.

Mechanism tests:

• Spillover (mediation) via R&D: Estimate (i) effect of FI on firm R&D investment (log R&D); (ii) effect of R&D on innovation controlling for FI; assess mediation via changes in FI coefficient and Sobel test of b1*d2.
• Competition effect: Interact FI with HHI (industry concentration) to test whether the impact of FI varies with competition intensity; higher HHI indicates less competition.

Heterogeneity analyses: Split samples by (i) productivity (TFP estimated via Ackerberg-Caves-Frazer; high vs. low by median), (ii) factor intensity (capital- vs. labor-intensive by median KI), (iii) region (coastal vs. non-coastal provinces), (iv) exporting status (exporters vs. non-exporters), and (v) ownership (SOEs vs. non-SOEs). Re-estimate baseline within each subsample.

• Baseline effect: Across specifications with firm, year, and province fixed effects (and with/without HHI), FI has a positive, statistically significant effect on firm innovation. In a representative specification (Table 2, col. 3), FI coefficient = 0.5859 (p<0.01). Interpretation: a one percentage point increase in industry FDI penetration is associated with about a 0.59% increase in a firm’s new product output.
• Robustness:
  • IV estimates using industry foreign-firm counts and initial-period FDI as instruments yield positive, significant FI effects (e.g., 4.4345 and 2.3858; p<0.01), supporting a causal interpretation.
  • Using patents as the innovation outcome: FI remains positive and significant (Table 3, col. 3, 0.4074*, p<0.1).
  • Alternative FI weights (sales-based FI_S; value-added-based FI_A): effects remain positive and significant (Table 3, cols. 4–5).
  • Policy shocks: Results persist when excluding pre-WTO years and when restricting to post-2005 exchange-rate-reform years (Table 3, cols. 6–7).
• Mechanisms:
  • Spillover via R&D: FI increases firms’ R&D investment (Table 4, col. 1; positive, significant). R&D, in turn, raises innovation (Table 4, col. 2; RD coefficient positive, p<0.01). The FI coefficient attenuates after including RD, and Sobel test rejects no-mediation, indicating a significant mediation channel through firms’ R&D.
  • Competition channel: The interaction FI × HHI is significantly positive (Table 4, col. 3), indicating that FI’s innovation-boosting effect is stronger in less competitive (more concentrated) industries and weakens as competition intensifies, consistent with an escape-competition effect dominating at low competition and Schumpeterian effect at high competition.
• Heterogeneity (Tables 5–6):
  • Productivity: Significant positive effect for high-productivity firms (0.8665***), not significant for low-productivity firms.
  • Factor intensity: Significant for capital-intensive firms (0.7394***), not significant for labor-intensive firms.
  • Region: Significant in non-coastal regions (1.9107***), not significant in coastal regions.
  • Trade status: Significant for both non-exporters (0.4337***) and exporters (0.8943***), with a larger effect for exporters.
  • Ownership: Significant for non-SOEs (0.7096***), not significant for SOEs.

The findings directly address the research question by showing that FDI, measured at the industry level, is associated with higher firm-level innovation outputs among Chinese local manufacturers. This effect operates through increased R&D investment, consistent with positive spillovers from foreign entrants via demonstration, learning, and labor mobility. The moderating role of market competition aligns with theory: in less competitive industries, FDI-induced competition spurs innovation (escape-competition); as competition intensifies, the Schumpeterian erosion of rents dampens incentives, weakening FDI’s positive impact. The heterogeneous effects underscore the conditional nature of spillovers: firms with greater absorptive capacity (high productivity, capital-intensive, exporting, non-SOE) and firms in regions with less intense competition (non-coastal) benefit more. These results reconcile mixed evidence in the literature by highlighting measurement at the firm level and conditioning factors. Policy-wise, attracting FDI can bolster domestic innovation, but complementary policies that raise firms’ absorptive capacity and appropriately manage competitive pressures are crucial to maximize gains.

Using firm-level data from Chinese manufacturing (2000–2007) and direct innovation measures, the study finds that FDI significantly enhances local firms’ innovation performance. Mechanism analyses reveal that (i) FDI boosts firms’ R&D investment, which mediates the innovation effect; and (ii) the positive impact diminishes as market competition intensifies. Benefits are concentrated among high-productivity, capital-intensive, exporting, non-SOE firms and firms in non-coastal regions. These insights contribute micro-level evidence to the FDI–innovation nexus and inform policies to attract and channel FDI to maximize spillovers, improve the innovation environment, and tailor approaches by firm type and region. Future research should leverage more recent data and further examine dynamic and policy-specific contexts to provide timely guidance.

The primary limitation is data coverage: the ASIF-based sample spans 2000–2007, preventing analysis with more recent firm-level information. Although endogeneity concerns are mitigated with industry-level FI measures, fixed effects, IVs, and multiple robustness checks, unobserved time-varying shocks cannot be fully ruled out. Additionally, measures of innovation rely on new product output and patents, which may not capture all dimensions or quality of innovation.