Digital Financial Inclusion and Provincial Economic Growth in China: Evidence from a Two-Way Fixed-Effects Panel, 2011–2023

Abstract

This paper asks whether the rapid expansion of digital payment platforms such as Alipay and WeChat Pay is linked to provincial economic growth in China, and if so, through what channel. It builds a panel of all 31 provincial-level regions from 2011 to 2023 (n = 403), combining the Peking University Digital Financial Inclusion (DFI) Index with official provincial data on GDP, urbanization, patents, foreign direct investment, higher-education enrollment, and industrial structure. A pooled regression finds a strong positive relationship between DFI and GDP (β = 0.573, p < 0.001), but this weakens sharply once province and year fixed effects are added (β = 0.142, p = 0.117) and is confirmed insignificant by a wild cluster bootstrap (p = 0.187), suggesting the pooled result largely reflects a shared national trend rather than a province-specific effect. Testing technological innovation as a possible channel, DFI strongly predicts patent applications (β = 0.775, p < 0.001), and patents in turn predict GDP after controlling for DFI (β = 0.093, p = 0.023), with a significant indirect effect confirmed both analytically (Sobel p = 0.042) and by a percentile bootstrap (95% CI [0.007, 0.150]). This innovation channel is sensitive to further checks, however: it weakens once foreign direct investment, education, and industrial structure are added as controls (b-path p rises to 0.259), and a panel unit-root test cannot rule out non-stationarity in the patent series. An exploratory instrumental-variables estimate using each province's distance to Hangzhou finds a large, positive association between instrumented DFI and GDP, but with a first-stage F-statistic that is only borderline strong (10.8) and an exclusion restriction that cannot be verified, so it is offered as a supplementary check rather than a resolution of endogeneity. Overall, this paper finds no robust direct effect of digital financial inclusion on provincial GDP, but suggestive evidence of an innovation-related channel that does not fully survive additional controls. A separate test for moderation by urbanization is statistically unstable and is not treated as evidence.