Abstract

Purpose – This paper aims to offer a new explanation for the persistent “limited participation puzzle” in the financial market, from a new perspective of the artificial intelligence shock in the labor market.

Design/methodology/approach – This paper proposes three hypotheses on the relationship between AI applications and financial market participation based on the theoretical background. To exam the hypotheses, the empirical analysis is conducted using data from the 2017, 2018 and 2021 waves of the Chinese General Social Survey. Furthermore, the paper conducts a series of robustness and endogeneity tests, including the use of different AI indicators, different regression models, placebo tests, penalized regression, and instrumental variable methods, as well as excluding the interference of peer effects, self-selection bias and COVID-19.

Findings - Results indicate that the more the household head’s job is impacted by AI, the lower the probability that the household invests in financial assets and the findings remain robust across a variety of tests. In addition, AI shock inhibits financial market participation not only by diminishing objective income but also by reducing subjective risk appetite at a deeper level. Besides, AI has more pronounced impacts on the investment preference of older, female, and less-educated people. Moreover, labor protection can alleviate the adverse effects. The study recommends that it is crucial to develop more targeted policies and enhance labor protections to mitigate the effects of AI shock.

Originality/value – This study provides significant research value for a deeper understanding of the mechanisms underlying limited financial market participation from the new perspective of technological disruption in the labor market and thus for providing important policy implications.

Keywords Artificial intelligence shock, Financial market participation, Labor market

人工智能冲击、风险偏好与金融市场参与：对“有限参与之谜”的新解释

摘要：

目标：本文基于劳动力市场中的人工智能冲击这一新视角为金融市场中的“有限参与之谜”提供一个新解释。

方法：本文提出了人工智能应用与金融市场参与之间关系的理论假设。为检验这些假设，本文利用了中国综合社会调查数据进行了实证研究。此外，本研究开展了一系列的稳健性与内生性检验，包括使用其他人工智能应用指标、其他回归模型、安慰剂检验、惩罚回归、工具变量法等，以及排除同群效应、自选择偏误与公共卫生事件干扰。

发现：研究结果表明，人工智能应用显著降低了微观家庭金融资产投资偏好。该结论通过了一系列稳健性与内生性检验。机制检验表明，人工智能应用通过降低劳动者的客观收入与主观风险偏好降低了微观家庭金融资产投资偏好。进一步分析发现，人工智能应用对年长、女性、低学历群体的家庭金融资产投资偏好的影响更大。劳动保护则可以在一定程度上缓解人工智能应用的负向投资偏好效应。本文结果为改善家庭金融资产投资不足现状提供了重要的政策启示。

创新性：本研究基于劳动力市场中的人工智能冲击这一新视角为更深刻理解微观金融市场参与的决策机理提供了重要的研究洞见，具有重要的政策含义。

关键词：人工智能冲击，金融市场参与，劳动力市场

作者：Chao Li、Wenyu Lao

文章来源：《International Journal of Emerging Markets》（28 May 2025）