Review of "RMB Internationalization and the Export Trade Effects of China's FDI in ASEAN" (Jin Dan & Zhang Yufu, 2022)

Academic fraud report · Geng Detector

Summary

This report assesses Jin Dan and Zhang Yufu's 2022 article in Financial Development Research (DOI: 10.19647/j.cnki.37-1462/f.2022.12.003) on the export effects of China's outward FDI in ASEAN amid RMB internationalization. The overall verdict is "highly suspicious" (orange). Four main issues are flagged. First, Tables 10 and 11 appear to have swapped labels: Table 10 (titled by investment-motive classification) lists reserve-holding sub-samples, while Table 11 (titled by reserve-holding classification) lists motive-based sub-samples, with identical full-sample coefficients. Second, Table 5 reports a GMM Hansen J statistic of 9.39/9.44 without degrees of freedom or p-values, yet the text declares the null of no over-identification is accepted — an unverifiable claim. Third, although the authors claim 400 quarterly observations (10 countries × 10 years × 4 quarters), the core OFDI stock variable is annual, allegedly converted via a quadratic interpolation that artificially inflates degrees of freedom. Fourth, the threshold parameter is inconsistently denoted as ξ, γ, and λ across three nearby sections, suggesting patchworking. Confidence is moderate; no definitive fabrication is proven without raw data.

Verdict

Highly suspicious (orange). The paper shows multiple signs of rushed assembly and weak methodological reporting, though outright data fabrication is not independently confirmed.

Key findings

Swapped table labels (Tables 10 & 11, pp. 31–32): Table 10's title refers to investment-motive grouping but the panel split is by RMB reserve holding; Table 11's title refers to reserve-holding grouping but the split is by efficiency- vs. market-seeking motive. Full-sample regression statistics are identical.
GMM reporting gaps (Table 5, p. 30): Hansen J statistics of 9.39 and 9.44 are reported without degrees of freedom or p-values, yet the text asserts acceptance of the null hypothesis of no over-identification.
Artificial sample inflation (Section IV.ii, p. 28): Authors claim 400 quarterly observations, but the core OFDI stock is annual; an unspecified quadratic interpolation ("二次函数——与和相配") is used to generate pseudo-quarterly values, which can manufacture spurious autocorrelation.
Inconsistent threshold notation (pp. 27–28): The threshold is denoted ξ in equations (2)–(3), γ in the variable description, and λ in Table 7 (value 0.58), suggesting patchwork composition.

Evidence highlights

Identical full-sample coefficients, t-values, R², and N across Tables 10 and 11.
Hansen J values of 9.39 and 9.44 reported without associated p-values or degrees of freedom.
400 observations stated for a panel that, given annual OFDI data, would require interpolation of the core regressor.
Threshold symbol shifts across three consecutive sections: ξ → γ → λ (0.58).

Notes

DOI: 10.19647/j.cnki.37-1462/f.2022.12.003
Reported turnaround: received 23 Sep, revised 9 Nov, published in Issue 12 of 2022 — unusually fast.
Follow-up requests: raw annual data, interpolation code, original Stata/R logs, and a published corrigendum for the table swap.
Limitations: AI-assisted review; final determination of misconduct requires institutional investigation. The swapped table headers could in principle stem from copy-editing rather than author intent, though the identical numbers underneath strengthen the suspicion of templating.

This report is AI-assisted and intended for academic discussion only. Misidentification and omission are possible. Official inquiries should govern any final judgment.