Mechanism Analysis and Response of Digital Financial Inclusion to Labor Economy based on ANN and Contribution Analysis

Abstract

Given the inclusiveness of digital inclusive finance (DFI) and its complex impact mechanism on the labor economy, this study uses the characteristics of adaptive and self-learning ability of artificial neural network (ANN) to simulate the process of delivering stimuli to nerve cells in the human brain through linear weighted summarization and functional mapping, and implement the optimization learning algorithm to adjust the weights in the network structure, thus completing the hierarchical analysis of index weight. At the same time, the neural network structure is used to approach the greatest extent and Garson algorithm is used for sensitivity analysis. We use data on the labor economy and digital financial inclusion in Heilongjiang, Jilin, and Liaoning provinces in China from 2011 to 2021 as a training dataset. The study found that (1) the indexes of DFI have different importance to the indexes of labor economy, among which the most important are the number and amount of insurance per capita and the proportion of the number and amount paid by digital technology, which have a normalized importance of 100 and 99.3%, further, R-square coverage is above 0.95, respectively, for labor economy indicators; (2) For different subdivided indicators, the indexes of DFI determine different significance. This study employs tools and policies related to DFI to address labor economy challenges, so as to promote the overall economic construction. This study studies the response and transmission mechanism of the concept of DFI to the labor economy, and explores the labor economy problems such as improving labor productivity and labor mismatch in the economy under its “inclusive” principle. Compared with the traditional weight analysis, it is closer to the real situation and has a stronger ability to fit the reality. In the future, the model could be rebased and measured against absolute indicators and a wider dataset could be adopted for extension to more areas.