数字求职教练能降低失业率吗？来自法国的实验证据

Can a Digital Job Search CoachReduce Unemployment? Experimental Evidence from France Aïcha Ben DhiaBruno CréponEsther MbihLouise Paul-DelvauxBertille PicardVincent Pons Development EconomicsDevelopment Research GroupMay 2026 Policy Research Working Paper11376 Abstract This paper evaluates the impact of Bob Emploi, a digitalplatform designed to provide personalized job search adviceand coaching to the unemployed. The platform, developedby a nonprofit organization with access to France’s publicemployment agency data, aims to replicate traditionalcounseling services through automated tools. The exper-iment included 212,277 individuals, with 56.3 percentrandomly assigned to receive encouragement to use the plat-form. Although our intervention increased Bob Emploi’susage by 27 percentage points, the effects of the platformremained limited. Users made modest changes to their search methods, showed slightly higher engagement withstandard employment services, and felt more supportedin their search. However, the evaluation finds no impacton time spent searching, occupational scope, or job seekerwell-being. Most importantly, the platform did not improveany employment outcomes over an 18-month follow-upperiod, with precise null effects across all subgroups. Theseresults suggest that digital job search assistance platformsmay need to combine coaching with specific job recom-mendations to effectively improve job seekers’ labor marketoutcomes. The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about developmentissues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry thenames of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely thoseof the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank andits affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. Can a Digital Job Search Coach Reduce Unemployment? Experimental Evidence from France∗ A¨ıcha Ben Dhia†, Bruno Cr´epon‡, Esther Mbih§,Louise Paul-Delvaux¶, Bertille Picard‖,and Vincent Pons∗∗ 1.Introduction Labor market frictions generate substantial unemployment even when jobs are available and workers’ skills matchemployers’ needs.Beyond standard search frictions (McCall 1970; Mortensen 1970; Pissarides 2000), job seekersface specific barriers to accessing and processing labor market information and translating it into effective searchstrategies (Babcock et al. 2012; Cooper and Kuhn 2020). Those who just started looking for a job may be overop-timistic about their prospects (Spinnewijn 2015; Mueller, Spinnewijn, and Topa 2021), while those who have beenunemployed for a longer time may lose motivation and self-esteem (Krueger and Mueller 2011; Arena et al. 2023). Toaddress these challenges, Public Employment Services worldwide provide two key forms of assistance: personalizedcoaching on job search strategies and direct matching between workers and job opportunities.Although studieshave shown that these services are effective (J-PAL 2022), large caseloads often limit how much individual attentionPublic Employment Services caseworkers can provide to each job seeker. The rise of online job boards in the 1990s promised to reduce labor market frictions by improving accessto job opportunities (Autor 2009).However, their impact on employment initially remained limited (Kuhn andSkuterud 2004; Kuhn and Mansour 2014), suggesting potential gains from supplementing these platforms withservices traditionally provided by Public Employment Services caseworkers (Kircher 2020; Kircher 2022). Recentstudies have found positive impacts of digital tools that focus on job recommendation and matching. In particular,exposing job seekers to vacancies that broaden their job search criteria (Belot, Kircher, and Muller 2019; Altmannet al. 2022) or to AI-generated vacancy recommendations (Le Barbanchon, Hensvik, and Rathelot 2023) improvestheir employment probabilities. Similarly, directing them to firms that are predicted to hire increases their short-term job-finding rates (Behaghel et al. 2024). By contrast, we lack evidence regarding digital platforms’ potential toscale and automate the second dimension of unemployment assistance: personalized job search strategy coaching. This paper bridges this gap by evaluating the impact of “Bob Emploi,” a digital platform designed to act as adigital coach, helping job seekers formulate and implement effective search strategies. Developed in 2016 through apartnership between the French public employment agency (“Pˆole emploi”) and the non-profit organization BayesImpact, the platform leverages rich administrative data to replicate core employment services typically providedby casewor