The global landscape of RTAs is experiencing rapid growth in both quantity and the depth of their provisions. China, a major trading nation, has actively participated in this expansion, aiming to establish a global high-standard FTA network. While the primary goal of RTAs is reducing trade barriers, their increasingly detailed investment-related clauses suggest a significant influence on investment flows. Previous research on the economic effects of RTAs has largely focused on trade, with limited consensus on the impact of provision depth on bilateral investment. Some studies have found a positive relationship between RTA depth and FDI flows, while others found no significant effect. This study addresses this gap by focusing on China's experience, examining how the depth of its RTA provisions affects inward FDI stocks.

The literature review examines three key areas: the investment effect of RTA presence, the measurement of RTA text depth, and the investment effect of RTA provision depth. Studies on the investment effect of RTA presence have shown mixed results, with some finding positive impacts on FDI inflows, particularly when RTAs include free access rules for investment. However, these studies often treated RTAs as simple dummy variables, failing to account for the heterogeneity of provisions across different agreements. The measurement of RTA text depth has evolved, with methods like those by Horn et al. (2010) and Hofmann et al. (2017) categorizing provisions into WTO+ and WTO-X categories and creating depth indexes. The investment effect of RTA provision depth is also a subject of ongoing research, with studies producing varied results depending on methodologies and data. Some studies have confirmed that deeper agreements with strict investment and dispute settlement provisions increase FDI, but others found no significant effect or that the effect is dependent on specific agreements and the inclusion of bilateral investment agreements. The authors highlight the relative lack of research focusing on the effect of China's specific RTAs on its inward FDI.

The study utilizes a structural gravity model for FDI, adapting the model of Kox and Rojas-Romagosa (2020). The econometric specification is a panel data regression model (Equation 2) analyzing the impact of RTA provisions on China's inward FDI stocks. The dependent variable is the inward FDI stock from partner country j to China in period t. The independent variables include a dummy variable for RTAs, and indexes for total depth, core depth, WTO+ provisions depth, and WTO-X provisions depth. Several control variables are included: market size, market size similarity, technological similarity, geographical distance, common border, and common language. The inverse hyperbolic sine transformation is used to handle zero and negative values in the FDI stock data. Data on FDI stocks are from the IMF CDIS database, data on RTA provision depth are from the World Bank database (supplemented by the authors' own analysis of newer FTAs), and other data are from the World Bank WDI and CEPII-GeoDist databases. The sample includes China and 184 investment partners from 2009 to 2019. Robustness checks include replacing core explanatory variables with legally enforceable provisions, dividing the sample into different time intervals, using lagged time-varying variables to address potential endogeneity issues, employing a dynamic panel model with the System GMM method, and performing two-stage least squares (IV) estimation using the number of RTAs signed by both parties with third parties as instrumental variables.

The baseline regression results show that all four depth indexes (total depth, core depth, WTO+ provisions depth, and WTO-X provisions depth) have significantly positive effects on China's inward FDI stocks. This result is robust to various robustness checks, including those that consider the legal enforceability of provisions, different sample intervals, lagged variables to address endogeneity, dynamic panel models and IV estimation. Heterogeneity analysis reveals that the positive effect of RTA depth on FDI is greater for developing countries and Belt and Road countries than for developed countries and non-Belt and Road countries. An analysis of specific provisions reveals significant positive effects from provisions on state trading enterprises, government procurement, TRIMS, intellectual property rights protection, investment, labor rights, and industrial cooperation. Provisions on competition and environmental policies showed no significant impact.

The findings strongly suggest that deeper RTAs, particularly those with detailed provisions covering investment and related aspects, significantly boost FDI inflows to China. The greater impact on developing countries and Belt and Road countries aligns with the expectations of those countries having more to gain from reduced investment barriers and improved investment environments. The heterogeneity across different provisions underscores the importance of carefully selecting the specific clauses in FTA negotiations to maximize their impact on investment. The insignificant effects of competition and environmental policies might reflect the limited legal enforceability of these provisions in the agreements analyzed.

This paper provides robust evidence that increasing the depth of RTA provisions significantly enhances China's inward FDI stocks. The significant heterogeneity in the effects across partner country types and specific provisions suggests a need for targeted strategies in FTA negotiations. Future research could explore the long-term dynamics of these effects, investigate the impact of specific policy mechanisms within provisions, and further examine the interaction between RTA provisions and other factors influencing FDI.

While the study employs a robust methodology and comprehensive dataset, certain limitations should be acknowledged. The analysis relies on the available data on RTA provision depth, which may not capture all relevant aspects of the agreements. The study focuses on China's experience, and the findings might not be directly generalizable to other countries. Despite efforts to address endogeneity, some unobservable factors could still influence the results. Future research should consider expanding the dataset to include more recent agreements and potentially different measures of agreement depth.