The short answer is yes. This paper studies the potential impact of artificial intelligence (AI) on families' fertility choices. According to the literature, AI is increasingly automating and replacing tasks performed by people, while the population is usually assumed to be exogenous. Keeping the settings of AI and automation, the paper models an important task that AI cannot automate, namely, the "production of children", as the major part of home production. Adults allocate their time between market production and home production. The costs of "child production" include the time cost in terms of forgone wages, and the capital dilution cost in terms of capital per capita. When AI affects how many tasks will be replaced by automation, it also determines the marginal products and corresponding prices of labor and capital. Increased penetration of AI skews the income structure of households toward capital income, reducing the time cost and increasing the capital dilution cost. This effect is opposite to the neutrality of technological progress. More importantly, if parents have a relatively high preference for children or recognize the positive effects of population size on R&D, the deepening of AI technology can help to raise the fertility rate. This finding has surprising and profound implications: under certain conditions, AI may mitigate the trend of decline in fertility for most developed countries. In addition, the paper highlights the impact of the positive externality and potential risk caused by deep automation: the fertility rates in optimal allocation and competitive allocation may diverge as the use of AI spreads.

This paper studies a job-assignment model incorporating taste discrimination and statistical discrimination simultaneously. We argue that when taste discrimination of some employers can affect the human capital investment behavior of a particular group of workers, other employers without taste discrimination will take this effect into account when hiring. Statistical discrimination thus arises naturally and is influenced by the degree of taste discrimination. We demonstrate that even if employers have ex-ante homogeneous beliefs about different groups, this belief can be transformed into negative stereotypes by taste discrimination. When such externalities exist, affirmative-action policies need to persist in the long run, and once removed, inequality still occurs. Moreover, we show that the specific direction of impact on statistical discrimination depends on the relative strength of the two effects: (i) Compared to other groups, stricter standards lead to lower investment incentives. (ii) Compared to non-investors within the same group, stricter standards lead to better identification of investments.

The digital economy's functioning hinges upon the close collaboration of various core elements. Data, as a pivotal production factor, derives its value not in isolation but rather from its intricate connection to physical media, like data processors and storage devices. This paper introduces a comprehensive equilibrium model that unifies critical components of the digital economy, including computing power, storage, data, and algorithms, within an integrated analytical framework, forming a more general theory. In the general digital economic model, data accumulation is restricted by data storage capability, while the efficiency of data utilization depends on rival computing power. When data is shared and utilized non-rivalry across all production and innovation sectors, the volume of data generated, total computing power, and distribution of computing resources mutually influence each other, culminating in a market competition equilibrium. However, this equilibrium may give rise to potential issues, such as excessive investment in computing power, unequal distribution of computing resources, and insufficient innovation during market competition. Moreover, under different parameter settings, the market competitive equilibrium may experience inadequate data sharing or data misuse. In conclusion, this paper extends the model to include algorithm innovation and dynamic accumulation of storage and computing power. This comprehensive analysis provides valuable insights into the intricate dynamics of the digital economy, facilitating a deeper understanding of its core mechanisms and implications for policy making and industry practices.

We model a dynamic data economy with fully endogenous growth where agents generate data from consumption and share them with innovation and production firms. Different from other productive factors such as labor or capital, data are nonrival not only among firms but also in their uses across sectors, which affect both the level and growth of economic outputs. Despite the vertical nonrivalry, the innovation sector dominates the production sector in data usage and contribution to growth because (i) innovations are cumulative and benefit from data that are durable and dynamically nonrival; and (ii) innovations "desensitize'' raw data into knowledge when entering production, which allays consumers' privacy concerns. Data uses in both sectors interact in generating allocative distortions and an apparent substitutability due to labor's rival usage across sectors and complementarity with data. Consequently, growth rates diverge under a social planner and a decentralized equilibrium, which is novel in the literature and has policy implications. Specifically, consumers' failure to fully internalize knowledge spillover while bearing privacy costs, combined with firms' market power, leads to an underpricing of data and inefficient data supply, causing underemployment in the innovation sector and suboptimal long-run growth. Improving data usage efficiency is ineffective in mitigating the underutilization of data, but interventions in the data market and direct subsidies hold promises.