Investment effect of regional trade agreements: an analysis from the perspective of heterogeneous agreement provisions

The paper investigates whether and how the depth of provisions in China’s regional trade agreements (RTAs) affects inward foreign direct investment (FDI) stocks. While RTAs have proliferated globally in both number and scope—expanding from traditional border measures to behind-the-border disciplines—most existing research emphasizes trade effects, leaving investment impacts underexplored and with mixed findings. Given China’s substantial FDI attraction post-WTO accession and its active strategy to build a high-standard FTA network, the authors examine the effect of heterogeneous and deeper RTA provisions on inward FDI to China. They posit that deeper agreements may reduce trade and investment barriers, enhance legal certainty and investor protections, and improve the broader business environment, thereby promoting FDI. The study aims to fill gaps by focusing on investment effects, incorporating heterogeneity in provisions (total, core, WTO+, WTO-X, and legal enforceability), and assessing differential impacts by partner type (developed vs. developing; Belt and Road vs. non-Belt and Road).

The literature is reviewed in three strands: (1) Investment effects of RTA presence: Early studies (e.g., Yeyati et al., 2003; Büthe & Milner, 2008; Medvedev, 2012; Berger et al., 2013) generally found that RTAs/PTAs increase FDI, though many treated RTAs as binary dummies, ignoring clause heterogeneity. (2) Measuring RTA text depth: Horn et al. (2010) distinguished WTO+ (14) and WTO-X (38) provisions; Hofmann et al. (2017) identified 18 core provisions (14 WTO+ and 4 WTO-X) and built total and core depth indexes in the World Bank Deep Trade Agreements database; Dür et al. (2014) created the DESTA depth index considering enforceability. These efforts opened the RTAs’ “black box.” (3) Investment effects of depth: Evidence is mixed—some find deeper RTAs boost FDI (Büthe & Milner, 2014; Osnago et al., 2017; Kox & Rojas-Romagosa, 2020; Gamso & Grosse, 2021), while others find insignificant effects (Gounder et al., 2019; Di Ubaldo & Gasiorek, 2022). Few studies focus on China’s RTAs; a related exception is Li et al. (2016) on ACFTA. The authors highlight the need to assess heterogeneous provisions and China’s overall RTA experience on FDI.

The study employs a structural gravity framework for FDI inspired by Anderson et al. (2019, 2020) and operationalized following Kox and Rojas-Romagosa (2020). The dependent variable is China’s inward bilateral FDI stock from partner j in year t (IMF CDIS). The baseline specification regresses inverse-hyperbolic-sine-transformed FDI stocks on: (i) an RTA dummy (1 if a China–j RTA is in force), (ii) gravity controls—market size (log sum of China and partner GDP at 2010 prices), market size similarity, technological similarity (based on GDP per capita), bilateral distance, contiguity dummy, and common language dummy, and (iii) year fixed effects. To capture heterogeneity in agreement content, the authors replace the RTA dummy with depth measures: total depth (count of all WTO+ and WTO-X provisions), core depth (count of 14 WTO+ and 4 WTO-X core provisions: competition policy, IPR, investment, capital movement), WTO+ depth, and WTO-X depth, constructed using the World Bank RTA text depth database (Hofmann et al., 2017). Legal enforceability is incorporated by assigning provision scores 0–2 (not mentioned/not enforceable; mentioned but excluded from dispute settlement; fully enforceable) and building enforceability-weighted depth indices. As additional robustness, a DESTA-based depth index (destadepth) is constructed using Dür et al. (2014). Data: Bilateral inward FDI stocks (IMF CDIS) 2009–2019; RTA text depth (World Bank Deep Trade Agreements database up to 2015) with the authors coding post-2015 China–Korea, China–Australia, and China–Georgia FTAs; GDP and GDP per capita (World Bank WDI); gravity variables (CEPII GeoDist). Sample: China with 184 partner economies, 2009–2019. Certain recently effective FTAs (China–Maldives, China–Mauritius) are excluded due to timing. Estimation and robustness: - Baseline OLS with year fixed effects. - Enforceability-weighted depth indices. - Alternative DESTA depth measure. - Re-sampling by 2-, 3-, and 4-year intervals to address lagged RTA effects. - One-period lag of all time-varying regressors to mitigate reverse causality. - Dynamic panel System GMM including lagged dependent variable L.In(FDI). - Instrumental variables (2SLS) for potential endogeneity of RTA and depth indices, using third-party domino instruments: number and average depth of RTAs signed by China and partner j with third countries (sample restricted to 2009–2015 due to instrument data availability). Diagnostics include AR(1)/AR(2), Hansen J, and weak-IV/under-identification tests. Additionally, heterogeneity analyses split samples by Belt and Road vs. non-Belt and Road, and developed vs. developing countries. Provision-level analysis substitutes the RTA dummy with ten key-provision dummies: state trading enterprises, government procurement, TRIMs, competition policy, environmental protection, IPR protection, investment, labor rights, movement of capital, and industrial cooperation.

• Baseline effects: The RTA dummy is significantly positive, indicating RTAs increase China’s inward FDI stocks. All four depth measures—total depth, core depth, WTO+ depth, and WTO-X depth—are significantly positive; core depth has a larger effect than total depth, and WTO+ depth generally exceeds WTO-X depth in baseline OLS (Table 2). - Enforceability: Using legally enforceable depth indices, all depth measures remain significantly positive. With enforceability, WTO-X depth’s coefficient exceeds WTO+ depth, consistent with fewer but more consequential enforceable WTO-X provisions (Table 3). The DESTA-based depth index (destadepth) is also strongly positive and larger in magnitude, reflecting its focus on crucial, enforceable areas. - Timing and dynamics: Results are robust when sampling at 2-, 3-, and 4-year intervals, when lagging time-varying regressors by one period, and in dynamic System GMM where L.In(FDI) ≈ 0.91 and RTA/depth effects remain positive (Tables 4–6). - Endogeneity: IV-2SLS estimates (2009–2015) with domino instruments yield predominantly positive and significant depth effects and pass under-identification and weak-IV tests for most specifications (Table 7). - Heterogeneity by partner type: • Belt and Road countries: Larger positive coefficients for RTA and all depth indices versus non-Belt and Road sample; stronger promotion of FDI to China (Table 8). • Developing vs. developed: For developing countries, RTA and all depth indices are significantly positive with larger magnitudes. For developed countries, RTA, core depth, and WTO+ depth are positive and significant; total depth and WTO-X depth are not significant (Table 9). - Provision-level effects: Seven provisions exhibit strong positive associations with inward FDI—investment (largest), state trading enterprises, government procurement, TRIMs, IPR protection, labor rights, and industrial cooperation. Movement of capital has a smaller positive effect. Competition policy and environmental protection are insignificant, likely due to limited inclusion and enforceability in China’s RTAs (Table 10). - Controls: Market size and market-size similarity positively relate to FDI; technological similarity is negative; distance and lack of contiguity reduce FDI; common language increases FDI, consistent with gravity intuition.

The findings support the three hypotheses. First, deeper RTA provisions are associated with higher inward FDI, suggesting that beyond tariff reductions, inclusion of investment-related and behind-the-border disciplines reduces uncertainty, lowers setup and operating costs, and enhances investor confidence. Second, heterogeneity across partner types indicates that developing and Belt and Road partners are more responsive to deeper provisions, potentially reflecting closer development stages, greater sensitivity to reductions in investment barriers, and synergies with the Belt and Road initiative’s infrastructure and institutional improvements. Developed partners appear most responsive to core/WTO+ areas that directly affect market access, transparency, and standards. Third, specific provisions differ in their FDI impact: direct investment disciplines, IPR, government procurement access, TRIMs alignment, labor rights, and industrial cooperation are key drivers, while competition and environmental provisions (as included in China’s RTAs during the period) show limited observable impact, likely due to sparse coverage or weaker enforceability. Overall, the results indicate that the credibility and depth of legally enforceable provisions matter for investment responses, and tailoring RTAs’ content can strategically influence FDI patterns.

The study constructs multiple RTA depth indices (total, core, WTO+, WTO-X, and enforceability-weighted; plus DESTA-based) to assess how heterogeneity in China’s RTA provisions affects inward FDI stocks (2009–2019). It finds that deeper RTAs robustly promote inward FDI to China across numerous checks (lagging, resampling intervals, System GMM, IV-2SLS, and alternative depth measures). Heterogeneous impacts are stronger for developing and Belt and Road partners, and provision-level analysis shows the largest positive effects for investment, IPR, government procurement, TRIMs, labor rights, industrial cooperation, and state trading enterprises, with smaller or insignificant effects for movement of capital, competition policy, and environmental provisions. Policy implications include: expanding coverage and enforceability of high-impact provisions (investment, IPR, procurement, TRIMs, labor, industrial cooperation, STEs); prioritizing deeper RTAs with Belt and Road partners; and, to attract more developed-country FDI, aligning with higher-standard rules to enhance transparency and predictability. These insights can guide China’s construction of a global high-standard FTA network from an investment perspective.

• RTA text depth data from the World Bank database are available only up to 2015; the authors hand-coded post-2015 China–Korea, China–Australia, and China–Georgia FTAs, and excluded recently effective FTAs (e.g., China–Maldives, China–Mauritius), potentially limiting coverage. - The analysis focuses on China’s inward FDI over 2009–2019, which may constrain generalizability to other countries or periods. - While reverse causality and endogeneity are addressed via lagging, System GMM, and IVs, residual endogeneity or measurement error in depth (e.g., simple counts vs. quality nuances) may remain. - Provision-level insignificance for competition and environmental policies may reflect limited inclusion or enforceability in China’s RTAs during the sample, not necessarily a true absence of effect.