Shujie Chang

Artificial intelligence (AI) is transforming productivity and reshaping labor markets. Yet, little is known about how people perceive the economic returns to AI skills and how these perceptions influence AI adoption. We investigate these questions using a randomized survey experiment conducted with U.S. participants. Drawing on data from the Economist showing an 8% AI premium for women and 0% for men among Norwegian job candidates, participants received information in one of three treatments: female premium only, male premium only, or both premiums. Our findings reveal four key patterns. First, participants exhibit a substantial upward bias in their initial beliefs, expecting premiums of 37.1% for women and 39.4% for men. Second, exposure to empirical data significantly lowers premium expectations in the U.S. market, and learning about men’s 0% premium reverses initial perception of male advantage among female participants. Third, the male-only treatment reduces female participants’ evaluation of AI’s personal benefit, and decreases both their willingness and actual use of AI tools. Male participants show little reaction. Last, presenting both premiums (highlighting a female relative advantage) yields more positive responses among female participants with stronger fairness concerns. These results reveal significant misperceptions about AI’s economic value and demonstrate how market information shapes belief updating and technology adoption along gender lines.

What causes the changing labor market power? We investigate how the rapid rise of the gig economy leads to the fall of the labor market power of Chinese manufacturing firms between 2007 and 2020. Using the staggered entry of online food delivery platforms into each city, we adopt a staggered difference-in-differences design and find that manufacturing firms in cities with the gig economy entry have experienced a significant reduction in their labor market power, as measured by the gap between the marginal revenue product of labor and wages. Our findings uncover the increasing wages of manufacturing firms and the bargaining power of workers as two primary mechanisms for the fall of labor market power.

The rapid expansion of the gig economy, fueled by internet and digital technology advancements, has not only transformed labor market formation but also exerted significant influence on social governance. On the one hand, gig economy platforms generate immediate employment opportunities that may mitigate criminal motivation through income enhancement. On the other hand, the prevalence of short-term labor contracts and increased workforce mobility may elevate public security risks. Leveraging the staggered entry of food delivery platforms across Chinese cities as a natural experiment, we employ a multi-period difference-in-differences model to identify the causal impact of gig economy penetration on crime rates. Empirical results demonstrate that platform entry reduced urban criminal incidence and prosecution rates by 7.1% and 7.0%, respectively, with particularly pronounced effects on property-related crimes. It reveals stronger impacts in eastern regions, areas with insufficient public security investment, and regions abundant in labor reserves. Furthermore, the rise of food delivery platforms can effectively mitigate the unemployment risks associated with robotic applications, contributing to a greater reduction in crime rates in cities with high levels of robotic penetration. Mechanism investigations indicate that platform expansion helps absorb migrant workers into the labor market, improve their income level, narrow relative income disparities, and enhance social trust, perceived fairness, and well-being—thereby collectively contributing to crime reduction.

We examine the relationship between cultural norms and tax aggressiveness by focusing on the unique role of the firm’s CEO, contrary to extant research on the topic. Our empirical evidence, from publicly traded firms in China during 2007–2020, suggests that China’s clan culture in CEO’s birthplace reduces the probability of aggressive tax planning, which is accordant to the prediction of our theoretical model. Using CEO turnover as a quasi-experiment, we employ the triple differences method to verify the causal effect. With respect to the mechanisms, we find clan culture norms could foster social trust among individuals, reducing CEOs’ willingness to engage in aggressive tax planning. Further, we exclude the competitive explanation that clan culture norms could restrict accessibility and ability of tax planning, including CEOs’ political connections, social ties, and financial expertise. Finally, our results suggest the persistence of cultural norms’ influence, which is complementary to the development of formal institutions, public attention, and marketization.