|
on Artificial Intelligence |
| By: | Shuo Zhang; Peter Kuhn |
| Abstract: | We use an algorithm audit of China's four largest job boards to measure the causal effect of a job seeker's gender on the jobs that are recommended to them, and to identify the algorithmic processes that generate those recommendations. Focusing on identical male and female worker profiles seeking jobs in the same industry-occupation cell, we find precisely estimated but modest amounts of gender bias: Jobs recommended to women pay 0.2 percent less, request 0.9 percent less experience, come from smaller firms, and contain .07 standard deviations more stereotypically female content such as requests for patience, carefulness, and beauty. The dominant driver of these gender gaps is content-based matching between posted job ads and the declared gender in new workers' resumes. 'Action-based' mechanisms -based on a worker's own actions or recruiters' reactions to their resume- contribute relatively little to the gaps we measure. |
| Keywords: | Recommender System, Algorithm, Gender, Job Platform |
| JEL: | C93 J71 J16 O33 M50 |
| Date: | 2025–11 |
| URL: | https://d.repec.org/n?u=RePEc:crm:wpaper:25108 |
| By: | Michelle Yin; Hoa Vu; Claudia Persico |
| Abstract: | A rapidly growing literature estimates AI's labor-market effects using large language models (LLMs) to self-assess occupational exposure. We demonstrate these measures are highly fragile. Replicating the dominant rubric with three frontier models on identical tasks, we find a 3.6-fold divergence in mean exposure with agreement as low as 57%. This measurement instability alters downstream empirical conclusions: in a difference-in-differences framework, individual-level coefficient magnitudes vary 2.4-fold across annotators, and county level estimates flip from a significant negative to an insignificant positive depending on annotators. We formalize this non-classical measurement error, highlighting the risks of treating evolving LLMs as static instruments. |
| JEL: | C81 J23 J24 O33 |
| Date: | 2026–04 |
| URL: | https://d.repec.org/n?u=RePEc:nbr:nberwo:35110 |
| By: | Eduardo Levy Yeyati |
| Abstract: | We study how the speed of AI adoption affects labor market outcomes during technological transitions. In a dynamic model where displaced routine workers enter a retraining pipeline with finite capacity, faster adoption compresses the displacement window without reducing total displacement, overwhelming the pipeline and generating permanent labor force exit through worker discouragement. The central result is that, even when two economies share the same long-run automation level, adoption speed alone determines transition welfare: faster adoption produces a larger discouraged stock, a steeper and more persistent decline in labor force participation, and a sustained compression of the labor share throughout the transition window. Non-routine employment and wages exhibit a crossing pattern — initially higher under fast adoption, then lower — so that faster adoption can simultaneously raise long-run wages for survivors while permanently reducing participation. Social welfare is strictly concave in adoption speed and maximized at an interior optimum below the market rate, because firms do not internalize the congestion externality they impose on the retraining queue, the irreversibility of permanent exit, or the wage depression borne by non-routine incumbents. The socially optimal speed and retraining capacity are complements: stronger institutions raise the optimal adoption speed. |
| Keywords: | Inteligencia Artificial, Innovación tecnológica, Mercado Laboral, Desarrollo de Habilidades, Artificial Intelligence, Technological innovation, Laboral Market, Skills Development |
| Date: | 2026–02 |
| URL: | https://d.repec.org/n?u=RePEc:udt:wpgobi:wp_gob_2026_01 |
| By: | Anders Humlum; Emilie Vestergaard |
| Abstract: | We study the early labor market impacts of AI chatbots by linking large-scale adoption surveys to administrative labor market records in Denmark. We document rapid currents: most employers in exposed occupations have adopted chatbot initiatives, workers report productivity benefits, and new AI-related tasks are widespread. Yet these currents have not broken the surface: using difference-in-differences, we estimate precise null effects on earnings and recorded hours at both the worker and workplace levels, ruling out effects larger than 2% two years after the launch of ChatGPT. What moves is the structure of work: employers absorb AI through task reorganization-including new tasks in content generation, AI oversight, and AI integration-and adopters transition into higher-paying occupations where AI chatbots are more relevant, though still too few to move average earnings. Technological change reshapes work well before it surfaces in earnings or hours. |
| Keywords: | Generative AI; Labor Markets |
| JEL: | J23 J24 J31 O33 |
| Date: | 2026–03 |
| URL: | https://d.repec.org/n?u=RePEc:crm:wpaper:26078 |
| By: | Guillermo Cruces; Diego Fernández Meijide; Sebastian Galiani; Ramiro H. Gálvez; María Lombardi |
| Abstract: | Does generative artificial intelligence (AI) reinforce or reduce productivity differences across workers? Existing evidence largely studies AI within firms and occupations, where organizationalselectioncompresseseducationalheterogeneity, leavingunclearwhetherAI narrows productivity gaps across individuals with substantially different levels of formal education. Weaddressthisquestionusingarandomizedonlineexperimentconductedoutside firms, in which1, 174 adults aged 25–45 with heterogeneous educational backgrounds complete an incentivized, workplace-style business problem-solving task. The task is a general (not domain-specific) exercise, and participants perform it either with or without access to a generative-AI assistant. Unlike prior work that studies heterogeneity within relatively homogeneous worker samples, our designtargets the between–education-group productivity gap as the primary estimand. We find that AI increases productivity for all participants, with substantially larger gains for lower-education individuals. In the absence of AIaccess, higher-education participants outperform lower-education participants by0.548standarddeviations; withAIaccess, thisgapfallsto0.139standarddeviations, implying that generative AI closes three-quarters of the initial productivity gap. We interpret this pattern as evidence that generative AI narrows effective productivity differences in task execution by relaxing constraints that are more binding for lower-education individuals, even though underlying skill differences remain, as reflected in persistent education gaps in task performance and in a follow-up exercise without AI assistance. |
| Keywords: | Productivity, artificial intelligence, education, human capital, inequality |
| JEL: | J24 O33 |
| Date: | 2026–03 |
| URL: | https://d.repec.org/n?u=RePEc:udt:wpgobi:wp_gob_2026_03 |
| By: | Lukas Freund; Lukas Mann |
| Abstract: | Who will gain and who will lose as AI automates tasks? While much of the discourse focuses on job displacement, we show that job transformation-a shift in the task content of jobs-creates large and heterogeneous earnings effects. We develop a quantitative, task-based model where occupations bundle multiple tasks and workers with heterogeneous portfolios of task-specific skills select into occupations by comparative advantage. Automation shifts the relative importance of tasks within each occupation, inducing wage effects that we characterize analytically. To quantify these effects, we measure the task content of jobs using natural language processing and estimate the distribution of task-specific skills. We construct projections of automation effects due to large language models (LLMs), exploiting a mapping between model tasks and automation exposure measures. Within highly exposed occupations, like office and administrative roles, workers specialized in information-processing tasks leave and suffer wage losses. By contrast, those specialized in customer-facing and coordination tasks stay and experience wage gains as work rebalances toward their strengths. Our findings challenge the common assumption that occupational automation exposure necessarily implies individual wage losses; and highlight that AI, through job transformation, may be disruptive even absent job displacement. |
| Keywords: | Artificial Intelligence, Labor Markets, Inequality, Skills, Tasks, Technological Change |
| JEL: | E24 J23 J24 J31 O33 |
| Date: | 2025–11 |
| URL: | https://d.repec.org/n?u=RePEc:crm:wpaper:25117 |
| By: | Matthias Fahn; Jin Li; Chang Sun |
| Abstract: | We study how AI affects compensation and job design when performance depends on workers' non-contractible effort. In a principal–agent model with limited liability, AI reduces effort costs but disproportionately lowers the cost of achieving satisfactory performance. This raises the incentive cost of sustaining high effort and can induce firms to replace high-wage, high-effort good jobs with low-wage, low-effort bad jobs, even when good jobs create more total surplus. As a result, AI can lower wages, reduce worker welfare, and even depress profits. If workers can adopt AI unilaterally, adoption occurs even when the resulting equilibrium harms both parties; when adoption requires worker cooperation, resistance is strongest where AI erodes rents embodied in good jobs. In a search-and-matching extension, endogenous outside options amplify these forces, reinforcing a bad-job economy and potentially reducing employment. |
| Keywords: | artificial intelligence, agency costs, job design, labor contracts, limited liability, incentives, search and matching |
| JEL: | D86 J41 O33 L23 |
| Date: | 2026 |
| URL: | https://d.repec.org/n?u=RePEc:ces:ceswps:_12612 |
| By: | Cassandra Merrit (Keough School of Global Affairs, University of Notre Dame); Jacob Dominski (Institute for Ethics and the Common Good, University of Notre Dame); Christopher Hoy (Melbourne Institute of Applied Economic and Social Research) |
| Abstract: | Artificial intelligence (AI) is frequently cast as a transformative technology that will raise productivity while displacing human work, yet organizational adoption remains uneven and aggregate effects are mixed. We examine whether middle managers contribute to this gap by acting as gatekeepers to AI adoption. In a pre-registered survey experiment of 2, 000 managers in the United States and United Kingdom, respondents were randomly assigned to view videos summarizing recent evidence on AI’s productivity benefits, its labor-displacing potential, or a placebo control. Exposure to information about labor displacement leads to a large reduction in intended AI adoption and advocacy (by 0.4–0.5 standard deviations) and a moderate reduction in staffing intentions (by 0.2 standard deviations). In contrast, information about productivity benefits has no significant average effect, although it increases advocacy among managers with low prior familiarity with AI. These findings indicate that middle managers’ responses to the information environment shape both technology adoption and employment intentions. Rather than inducing substitution away from labor and toward AI, information about AI’s labor-displacing potential leads managers to scale back both planned AI adoption and their staffing intentions. |
| Keywords: | Artificial Intelligence, Technology adoption, Managers, Gatekeepers, Productivity, Labor displacement |
| JEL: | J23 J24 O33 L2 M54 D83 |
| Date: | 2026–04 |
| URL: | https://d.repec.org/n?u=RePEc:iae:iaewps:wp2026n02 |
| By: | Tania Babina |
| Abstract: | This paper reviews firm-level data on artificial intelligence and the emerging evidence on AI's economic effects. It argues that measurement is central: different AI datasets capture different objects (including invention versus use, internal capability building versus outsourcing, and realized activity versus investor perceptions) and can therefore lead to different conclusions. The paper develops a framework for choosing among these measures and surveys available data sources on firm AI efforts. It synthesizes evidence on AI's effects on firm growth, valuation, productivity, risk, labor, competition, financial markets and applications. The paper concludes by suggesting some ideas for future research. |
| JEL: | G0 J0 M0 |
| Date: | 2026–04 |
| URL: | https://d.repec.org/n?u=RePEc:nbr:nberwo:35123 |
| By: | Fenella Carpena; Simon Galle |
| Abstract: | Motivated by the rise of artificial intelligence (AI), we set up a quantitative general-equilibrium model of the labor market impact of occupation-specific technical change. The highly tractable model crystallizes how three fundamental forces shape the impact of technical change on the occupational wage distribution: the input substitution elasticity, the final demand elasticity, and the labor supply (reallocation) elasticity. The difference between the former two elasticities determines whether machines and workers are gross complements, while the reallocation elasticity governs the magnitude of the distributional effects. We estimate the reallocation elasticities from group-by-occupation specialization changes, allowing for asymmetric reallocation and associated ripple effects on wages across occupations. After combining these estimates with externally disciplined demand and substitution parameters, as well as AI exposure measures, we shed light on the aggregate and distributional effects of occupation-specific advances in AI: wages in administrative services grow the least, ripple effects on less exposed occupations are substantial, AI modestly compresses the returns to education, and, on average, disproportionately benefits lower-income groups. |
| Keywords: | Technical change, labor reallocation, wage inequality, occupational heterogeneity, artificial intelligence, large language models |
| JEL: | J23 J24 E24 E25 |
| Date: | 2026–02 |
| URL: | https://d.repec.org/n?u=RePEc:crm:wpaper:26039 |
| By: | Francesca Miserocchi; Savannah Noray; Alice Wu |
| Abstract: | The advances of artificial intelligence (AI) are built on the groundwork laid by researchers. We study the labor market competition between academia and industry for AI researchers and its consequences for public knowledge production. Using data on 150, 000 computer science researchers, we document a major reallocation of AI talent toward top technology firms between 2005 and 2020. Publications at AI conferences predict transitions to top firms more strongly than to academia. Exploiting acceptance decisions at a leading AI conference, we compare accepted authors with similar rejected authors and find that a publication increases the probability of moving to a top firm by 2-6 percentage points in the next 1-3 years. Sorting to top firms is stronger for male researchers, whereas female students and postdocs are more likely to get tenure-track positions following a publication. Researchers who move to top firms subsequently publish fewer papers, resulting in approximately 1, 000 fewer AI papers and 2, 000 fewer papers in other computer science areas per year in the public domain. |
| Keywords: | Sorting, Productivity Signals, Labor Market Concentration, Innovation |
| JEL: | J23 J24 O31 |
| Date: | 2026–04 |
| URL: | https://d.repec.org/n?u=RePEc:crm:wpaper:26106 |
| By: | Giacomo Battiston; Federico Boffa; Eugenio Levi; Alberto Parmigiani; Steven Stillman |
| Abstract: | Technological disruptions often generates political conflict. Artificial intelligence (AI) is widely expected to transform labor markets and economic systems, yet it has not become a strongly polarizing political issue in advanced democracies. This paper investigates why, by fielding a preregistered survey experiment with 11, 418 respondents in the United States, Germany and Italy. We examine factual knowledge on AI and automation, beliefs over its economic effects, demand for policy intervention and signatures of online petitions on Change.org. We document limited knowledge, widespread pessimism on their labor-market impact, substantial demand for government intervention and considerable potential for political mobilization, pointing to an unmet demand for policy responses. We then test the mobilization power of competing political narratives on the economic effects of AI and automation. Overall, across countries and institutional contexts, politicizing AI shifts policy preferences in the expected directions but reduces engagement in political mobilization. In addition, it decreases support for the extreme petitions, thereby reducing polarization. These findings suggest that emerging technologies characterized by high uncertainty and large distributive effects may not follow the historical pattern of polarization associated with past economic shocks. Our results rationalize politicians' hesitation towards increasing the salience of AI and automation. |
| Keywords: | Artificial Intelligence; Automation; Political Polarization |
| JEL: | O33 P16 |
| Date: | 2026–03 |
| URL: | https://d.repec.org/n?u=RePEc:crm:wpaper:26063 |
| By: | Ngunza Maniata, Kevin; Pinshi, Christian P. |
| Abstract: | This article provides a structured review of the rapidly growing literature on the implications of artificial intelligence (AI) for central banking and proposes an organizing taxonomy of the main macro-financial and institutional channels discussed in recent contributions. As a general-purpose technology, AI has the potential to influence productivity growth, price-setting behavior, inflation dynamics, and the transmission of monetary policy, while its diffusion in the financial system may introduce new sources of systemic risk through algorithmic coordination, model opacity, and cyber vulnerabilities. The article synthesizes recent contributions from international financial institutions and academic research to organize these channels within a coherent macro-financial framework. It also reviews documented applications of AI within central banks, including macroeconomic forecasting and nowcasting, supervisory technology, payments oversight, and internal information processing. Attention is given to the emerging literature on African central banks and other emerging market economies, where AI offers opportunities to alleviate data constraints but also raises challenges related to skills, infrastructure, and governance. By organizing and critically assessing existing evidence, the article clarifies why AI constitutes a structural issue for central banking and identifies key areas for future research on monetary and financial stability. |
| Keywords: | Artificial intelligence; Central banks; Monetary policy; Financial stability; Emerging markets. |
| JEL: | E31 E52 E58 G01 G28 O33 O55 |
| Date: | 2026–03–05 |
| URL: | https://d.repec.org/n?u=RePEc:pra:mprapa:128352 |
| By: | Luca Fornaro; Martin Wolf |
| Abstract: | We provide a macroeconomic framework to study monetary and fiscal policies for AI. Advances in AI expand firms’ ability to automate production. While higher automation boosts productivity and potential output, it also reduces workers’ share of income. Since workers have a high propensity to consume, advances in AI may depress aggregate demand and lead to a slump. Expansionary monetary policy can convert an AI slump into an AI boom, but in doing so it faces two challenges. In the short run, AI worsens the inflation-employment trade off faced by the central bank. In the medium run, monetary policy may be constrained by the zero lower bound, since weak demand lowers the natural rate. Employment subsidies and cuts in labor taxes can usefully complement monetary policy, by reducing firms’ cost of labor and inflation, as well as supporting workers’ income and aggregate demand. |
| Keywords: | monetary policy, automation, AI, inflation, liquidity traps, endogenous productivity, wages, artificial intelligence |
| JEL: | E32 E43 E52 O31 O42 |
| Date: | 2026–04 |
| URL: | https://d.repec.org/n?u=RePEc:upf:upfgen:1943 |