|
on Education |
|
Issue of 2026–11–09
twenty-one papers chosen by João Carlos Correia Leitão, Universidade da Beira Interior |
| By: | Cruces, Guillermo; Fernández Meijide, Diego; Galiani, Sebastián; Gálvez, Ramiro H.; Lombardi, María |
| Abstract: | Does generative artificial intelligence (AI) reinforce or reduce productivity differences across workers? Existing evidence largely studies AI within firms and occupations, where organizational selection compresses educational heterogeneity, leaving unclear whether AI narrows productivity gaps across individuals with different levels of education. A related question is whether the productivity gains from AI reflect productive use of the tool or mere delegation, fading once AI is unavailable, and whether the answer differs across education groups. We address both questions using a randomized online experiment outside firms, in which 1, 174 adults aged 2545 complete an incentivized, workplacestyle business problem-solving task with or without a generative-AI assistant, followed by a non-AI-assisted follow-up module. AI increases performance for all participants, with substantially larger gains for lower-education individuals. In the control group without AI, higher-education participants outperform lower-education participants by 0.548 standard deviations; with AI, this gap falls to 0.139 standard deviations, closing about three-quarters of the initial gap. Chat-log measures explain both why this gap narrows and why it does not disappear: lower-education participants successfully obtain substantial assistance from AI, while higher-education participants use the tool somewhat more effectively across several margins. The follow-up results showthat these gains are not purely driven by delegation; treated participants do not perform worse than controls once AI is removed, and lower-education participants retain part of their gain, although a sizable education gap re-emerges. Consistent with this interpretation, intensive use of the assistant produces strong task performance regardless of how much time and effort participants devote themselves, but follow-up performance is substantially higher only when intensive AI use is combined with sustained effort on the task. We interpret these findings as evidence that generative AI narrows effective productivity differences in task execution while underlying human-capital differences continue to shape unassisted performance and effective use of the tool. |
| Keywords: | Productivity;artificial intelligence;Education;human capital;Inequality |
| JEL: | J24 O33 |
| Date: | 2026–10 |
| URL: | https://d.repec.org/n?u=RePEc:idb:brikps:14819 |
| By: | Dang, Hai-Anh H.; Do, Minh N. N. |
| Abstract: | Generative AI can make schoolwork look better without making students better at it. AI tutors could reach families who cannot afford human tutors, but whether they narrow learning gaps depends on policy. Randomized trials in Ghana, Nigeria and Sierra Leone show that AI-supported programs can raise learning measured without AI. Yet they do not isolate what works, and a large trial in rural China found no gain from AI personalization over non-personalized practice. This essay suggests that education systems, in Africa and beyond, judge complete programs rather than chatbots: by what students learn without AI, who benefits, and how they compare with feasible alternatives at full cost. |
| Keywords: | AI tutoring, generative AI, learning, inequality, poorer countries |
| JEL: | I21 I24 I25 O15 O33 |
| Date: | 2026 |
| URL: | https://d.repec.org/n?u=RePEc:zbw:glodps:1844 |
| By: | Ciprian Domnisoru; Tuomas P. Heikkinen |
| Abstract: | Have concerns about the potential job displacement effects of artificial intelligence affected the education choices of students? We analyze administrative data on the university applications and course choices of Finnish students. We find, after the launch of ChatGPT 3.5 in November 2022, a persistent drop in applications to fields of study classified as more exposed to AI under several different measures. The drop is particularly large in computer science and other ICT fields. Examining the elective course choices of non-ICT students, we find declining enrollment in programming and algorithms courses, particularly among students who had not taken mathematics as part of their high school exit exam. At the same time, non-ICT students enrolled more in AI/ML courses, and this increase was relatively larger among men. |
| Keywords: | artificial intelligence, college major choice, higher education, unemployment |
| JEL: | I21 J24 J64 O33 |
| Date: | 2026 |
| URL: | https://d.repec.org/n?u=RePEc:ces:ceswps:_13105 |
| By: | Robert W. Fairlie; Saurabh Khanna; Prashant Loyalka; Gagandeep S. Sachdeva |
| Abstract: | Affirmative action programs are often criticized because of concerns that they result in lower worker productivity and efficiency losses. We study the relative productivity of workers benefiting from an aggressive affirmative action policy in a setting where hiring constraints are mandated upon employers and are typically likely to bind. In India, colleges are required to reserve approximately 50 percent of faculty hires for individuals from disadvantaged caste and social class groups. We collect and analyze data from a nationally representative sample of 50 engineering and technology colleges in India, some of which randomly assign students to classrooms. We find that reservation category faculty have lower levels of education, lower professorial ranks and fewer years of experience in academia than general category faculty who are not hired through reservations. Yet, even with lower qualifications, we do not detect meaningful differences in instructional quality between reservation category faculty and general category faculty across a wide range of measures that include course grades, follow-on course grades, standardized test scores, dropout, attendance, graduate school plans, and graduation. In fact, we find that, at least for immediate effects on course grades, students taught by reservation category faculty perform slightly better than students taught by general category faculty. We also estimate “teacher-like-me” effects of reservation category faculty on the relative course performance and longer-term outcomes of reservation category students and cannot reject the hypothesis of null effects. Furthermore, even in the face of potential discrimination and resentment against faculty hiring quotas, general category students perform slightly better in classrooms taught by reservation category faculty than general category faculty. Finally, estimates of differences between reservation category and general category faculty in several measures of research productivity and administrative service are small in magnitude and not statistically different from zero. The findings have implications for the heated debates over affirmative action programs found in many countries around the world and in India which is now the largest country in the world. |
| JEL: | I23 I24 J15 J78 |
| Date: | 2026–10 |
| URL: | https://d.repec.org/n?u=RePEc:nbr:nberwo:35963 |
| By: | Flávio Cunha; Qinyou Hu; Andrea Salvati; Kenneth I. Wolpin; Rui Zeng |
| Abstract: | We evaluate JumpStart, a parent-directed school-readiness program run by a Houston-area school district, in a randomized trial spanning three cohorts. Being offered the program raises scores on a curriculum-aligned assessment by 7.6 percentage points and parental investment by 0.26 standard deviations. A model of skill formation with endogenous inputs and program participation yields an average treatment effect of completing the program of about 9 percentage points, and attributes roughly a quarter of this effect to higher measured investment. Nonparametric bounds corroborate the average effect, and a mediation analysis that relaxes part of the model's structure supports the decomposition. |
| Keywords: | early childhood development, human capital development |
| JEL: | C93 I21 I24 I28 J13 J24 |
| Date: | 2026 |
| URL: | https://d.repec.org/n?u=RePEc:ces:ceswps:_13100 |
| By: | Dang, Hai-Anh; Behrman, Jere |
| Abstract: | Nearly half of 15-year-old students in OECD countries, and more in some poorer countries, use AI weekly for learning. Generative AI makes schoolwork help cheap for connected students, but cheaper help need not yield lasting learning. We link the economics of supplementary education with learning science to examine how effort, entry costs, public provision and assessment shape AI tutoring's benefits. Randomized trials find gains and losses varying by setting, use and outcome, while lasting, equitable gains remain uncertain. We distinguish program effects from generative AI's contribution, compare full costs with alternatives, and propose decision rules and a trial design. |
| Keywords: | artificial intelligence, tutoring, learning, educational inequality, public investment, randomized trials |
| JEL: | I21 I24 I28 H52 O33 |
| Date: | 2026 |
| URL: | https://d.repec.org/n?u=RePEc:zbw:glodps:1861 |
| By: | Dana J. Smith1, Huiyi Chen2, and Molly L. Brewer1 (College of the Holy Cross, James Madison University) |
| Abstract: | Following Dobbs v. Jackson Women’s Health Organization, many U.S. states adopted new abortion restrictions. Using IPEDS data and a difference-in-differences design comparing institutions in states that adopted bans with institutions in states that did not, we examine student enrollment, institutional responses, and faculty hiring. We find do not find evidence of average changes in women’s shares of undergraduate applicants or entering students. However, declines in women’s applicant, graduate enrollment, and tenure-track hiring shares are evident at more selective institutions compared to less selective schools, consistent with greater responsiveness among groups with broader choice sets. Evidence of compensating admissions practices or financial aid responses on the institution side is limited. Overall, the clearest changes in gender composition are concentrated in more selective markets. |
| Keywords: | abortion policy; college choice; higher education; Dobbs; enrollment |
| JEL: | I23 I26 J16 K38 |
| Date: | 2026–10–14 |
| URL: | https://d.repec.org/n?u=RePEc:hcx:wpaper:2603 |
| By: | Eric Brunner; Shangyue Jiang; Stephen Ross |
| Abstract: | Beginning in the summer of 2021, the state of Connecticut launched the Learner Engagement and Attendance Program (LEAP) where home visits by school personnel were conducted with students identified as chronically absent from Connecticut’s 15 most disadvantaged school districts. Home visits were designed to connect students and their families directly with teachers to foster strong connections. We use student-level data from 2021–22 through 2023-24 to estimate event studies of the effect of LEAP on student attendance, chronic absenteeism, disciplinary incidents, and student achievement. Utilizing models that address selection into treatment timing over student observable and unobservable attributes, we find that students who received the LEAP intervention experienced approximately a 4 to 5 percentage point increase in attendance rates, a 13 to 15 percentage point decline in chronic absenteeism, and a 3 to 5 percentage point decline in the number of disciplinary incidents. Notably, initial event studies indicated increasing treatment effects over time, but after correcting for selection on unobservables treatment effects appeared constant over time. These effects were broad based, with similar impacts by gender, race and ethnicity, and free or reduced-price lunch status. These attendance and discipline results provide suggest that the LEAP intervention increased student noncognitive skills. |
| JEL: | I21 J13 |
| Date: | 2026–10 |
| URL: | https://d.repec.org/n?u=RePEc:nbr:nberwo:35961 |
| By: | Peter Arcidiacono; Cory Dowd; Matthew Lilley |
| Abstract: | There has been widespread concern about self-censorship on university campuses, and especially that conservative voices are being stifled. Using survey data from a private elite college, we show that self-censorship is not simply about being conservative: it is students with the most moderate views who self-censor the most. This pattern is associated with confidence and salience — students who are more confident in their views or who see an issue as especially important tend to hold more extreme views and to self-censor less. At the same time, conservatives self-censor at higher rates than liberals, and the same individual is more likely to self-censor on the views where they are relatively more conservative than on those where they are relatively more liberal; conservatives are also far more likely than liberals to cite concerns about social exclusion as a reason for self-censoring, and report self-censoring less outside the university community. Groups perceived as marginalized self-censor less than majority groups, and especially so on issues salient to them (e.g., Black students on BLM). |
| JEL: | D83 D91 I23 Z13 |
| Date: | 2026–10 |
| URL: | https://d.repec.org/n?u=RePEc:nbr:nberwo:35979 |
| By: | Zhang, Bo (Management School, Capital Normal University, Beijing, China); Wang, Yizhu (Management School, Capital Normal University, Beijing, China); Zhou, Peng (Cardiff Business School, Cardiff University, Cardiff, UK) |
| Abstract: | Some university students choose not to use artificial intelligence (AI) tools, yet research on students and AI has focused largely on adoption and has treated avoidance as maladaptive. That view cannot explain when avoiding AI might support, rather than undermine, student development. Drawing on transactional coping theory, threat-appraisal reasoning, and self-regulation theory, this study examined whether avoidant AI use is associated with career exploration and whether AI-shock awareness moderates that association linearly or curvilinearly. We hypothesized a positive main effect, a negative linear moderation, and a U-shaped curvilinear moderation. Survey data from 289 students at multiple Chinese universities (recruited via snowball sampling; 289 valid cases from 314 returned questionnaires) were analyzed with three-step hierarchical regression and bias-corrected and accelerated bootstrap confidence intervals (5000 resamples). Avoidant AI use was positively associated with both environment exploration and self-exploration. The linear interaction was negative in sign, but unreliable for environment exploration and only marginally supported for self-exploration. The quadratic interaction was significant for both outcomes: the association held at low awareness for both outcomes and at high awareness for self-exploration, but not at moderate awareness. For environment exploration, the high-awareness rebound rested on a small subgroup and is treated as preliminary. These results are consistent with the view that the developmental consequences of avoiding AI depend on the self-regulatory cost of maintaining that stance, a mechanism we treat as inferred rather than directly measured, and they carry implications for career education and AI-literacy instruction in higher education. |
| Keywords: | technology avoidance; career exploration; AI-shock awareness; self-regulation; curvilinear moderation; higher education |
| Date: | 2026–11 |
| URL: | https://d.repec.org/n?u=RePEc:cdf:wpaper:2026/14 |
| By: | Drydakis, Nick |
| Abstract: | This study examines whether pluralistic teaching is associated with students' learning dispositions. The data are drawn from an economic inequality course delivered at an English university between 2016 and 2025. A new scale capturing pluralistic knowledge of economic inequality is developed around six principles: structural and relational ontology; epistemic pluralism and situated knowledge; intersectional and decolonial analysis; normative commitment to justice, equity and transformation; methodological diversity and grounded inquiry; and engaged scholarship and evidence-informed policymaking. The results show that, after completing the course, students reported higher levels of pluralistic knowledge of economic inequality and stronger learning dispositions relating to critical thinking, epistemic curiosity, civic engagement, self-efficacy and social justice. Pluralistic knowledge of economic inequality is also positively associated with these learning dispositions. In addition, both pluralistic knowledge and learning dispositions are positively associated with academic performance in course assessments. These findings are relevant to educators, curriculum designers and policymakers seeking to strengthen students' analytical, ethical and civic capacities in engaging with socio-economic phenomena. |
| Keywords: | Pluralism, Pluralistic teaching, Economics education, Economic inequality, Critical thinking, Civic engagement, Epistemic curiosity, Social justice orientation, Curriculum design, Grades |
| JEL: | A20 A23 B50 D63 I24 I28 Z13 |
| Date: | 2026 |
| URL: | https://d.repec.org/n?u=RePEc:zbw:glodps:1853 |
| By: | Roy, Karna; Zhang, Andrew |
| Abstract: | This study examines the relationship between school performance and housing prices in the Dayton Metropolitan Area. The analysis uses data from 40 geographic subunits across seven school districts, combining housing prices, Ohio Performance Index scores, and socioeconomic characteristics. A baseline regression is first used to estimate the association between school performance and housing prices while controlling for socioeconomic characteristics, followed by a district fixed-effects model that accounts for unobserved characteristics shared by geographic subunits within the same school district. The baseline model finds a positive and statistically significant association, with a one-point higher Performance Index associated with approximately 1.8% higher median home prices. However, after the district fixed effects are included, the Performance Index coefficient becomes negative and statistically insignificant. This indicates that the positive association does not remain within districts under this specification. These findings suggest that the observed relationship between school performance and housing prices may reflect broader differences across school districts rather than a consistent relationship within districts. |
| Date: | 2026–10–23 |
| URL: | https://d.repec.org/n?u=RePEc:osf:socarx:4s3nm_v1 |
| By: | Pamela Giustinelli; Edwin Leuven |
| Abstract: | College applicants in centralized systems rank programs before they know where they will be admitted and whether they will enroll and complete. We surveyed Norwegian applicants about these prospects for their first two choices after applying but before admissions and link their beliefs to administrative records. Comparing applicants just above and below admission cutoffs allows us to validate these beliefs with outcomes. Marginal applicants slightly overpredict enrollment if admitted to their first choice and underpredict it at their second, increasing between-program prediction error. While they substantially overpredict completion given enrollment for both programs, the error in the difference between programs is much smaller. In a model of applicants’ rankings, correcting completion beliefs for one program changes the predicted probability of ranking the current first choice above the second by 12-14 percentage points, whereas correcting both barely changes it. Our study sheds light on validation of subjective counterfactuals and implications of their correction. |
| Keywords: | subjective expectations, counterfactual outcomes, centralized college admissions, college enrollment, college completion, program rankings |
| JEL: | C21 C83 D84 I23 |
| Date: | 2026 |
| URL: | https://d.repec.org/n?u=RePEc:ces:ceswps:_13107 |
| By: | F. Cerina; E. Dienesch; A. Monge-Naranjo; A. Moro |
| Abstract: | Large cities gain abstract jobs and shed routine jobs faster than small ones. Whether this divergence arises because workers move across locations or because workers' labor market careers differ across locations matters for welfare and policy, and cross-sectional data cannot tell the two apart. Using the French administrative panel DADS–EDP (1984–2021), we decompose the change in each occupation's employment share, by city size, into within-city occupational switches, cross-city relocation, combined moves, and non-employment flows. The divergence is produced inside cities. We find that large cities move workers from routine to abstract jobs at twice the rate of small cities, and this flow more than accounts for the urban gap in abstract employment. The routine decline is absorbed by upgrading into abstract work, not by downgrading into manual services. This urban elevator predates the 1994 onset of French polarization and operates within occupations - holding worker characteristics and the occupation of origin fixed, staying in a large city raises the rate of upgrading by about 40%, for college and non-college workers alike. The evidence places the burden of explanation on agglomeration forces that act on occupational mobility within cities rather than on the sorting of heterogeneous workers across them. |
| Keywords: | Spatial polarization, Urban elevator, Worker flows, Spatial Sorting, Occupational mobility, Longitudinal data |
| JEL: | J62 J24 R11 R23 |
| Date: | 2026 |
| URL: | https://d.repec.org/n?u=RePEc:cns:cnscwp:202618 |
| By: | Nicky Nicholls (Department of Economics, University of Pretoria, Private Bag X20, Hatfield 0028, South Africa); Roula Inglesi-Lotz (Department of Economics, University of Pretoria, Private Bag X20, Hatfield 0028, South Africa); WiZelle Kritzinger (Department of Economics, University of Pretoria, Private Bag X20, Hatfield 0028, South Africa) |
| Abstract: | This paper examines whether structured near-peer mentoring improves the research performance of Honours students in economics. Using a randomised stage-rotation design at the University of Pretoria, 50 Honours students were assigned to a PhD-student mentor during either the research proposal stage (Semester 1) or the research paper stage (Semester 2), ensuring that every student was mentored while a comparison group was preserved at each stage. We estimate intention-to-treat (ITT) effects of assignment and, using random assignment as an instrument for self-reported contact, local average treatment effects (LATE) of actual engagement. Compliance was high, at 87 percent in Semester 1 and 100 percent in Semester 2. All point estimates are positive but statistically insignificant: assignment is associated with increases of 0.52 points (out of 10) in the proposal grade and 1.86 points (out of 100) in the research paper grade, with corresponding LATE estimates of 0.99 and 4.29 points. Among mentored students, more frequent meetings are associated with significantly higher proposal grades (1.26 points per additional frequency level, p = 0.024), but not with research paper grades. The findings suggest that engagement intensity, rather than assignment per se, is the more likely channel of any mentoring effect, and that generic research guidance is most valuable while the research question is still being formulated. The design offers a low-cost and ethically defensible framework for evaluating postgraduate research support. |
| Keywords: | Peer Mentoring, Near-Peer Mentoring, Postgraduate Supervision, Randomised Evaluation, Economics Education, Scholarship of Teaching and Learning |
| JEL: | A22 A23 I23 |
| Date: | 2026–10 |
| URL: | https://d.repec.org/n?u=RePEc:pre:wpaper:202639 |
| By: | Pietro Geuna (PSE - Paris School of Economics - UP1 - Université Paris 1 Panthéon-Sorbonne - ENS-PSL - École normale supérieure - Paris - PSL - Université Paris Sciences et Lettres - EHESS - École des hautes études en sciences sociales - CNRS - Centre National de la Recherche Scientifique - INRAE - Institut National de Recherche pour l’Agriculture, l’Alimentation et l’Environnement - ENPC - École nationale des ponts et chaussées - IP Paris - Institut Polytechnique de Paris, IPP - Institut des politiques publiques, EHESS - École des hautes études en sciences sociales) |
| Abstract: | Outside options play a central role in models of wage determination, yet the employment opportunities available to individual workers are not directly observable. I develop an individual-level measure of the value of alternative employment, rooted in a discrete choice model of labor supply estimated from observed job-to-job transitions using neural networks to account flexibly for differences in workers' preferences and their compatibility with potential employers. I validate the measure in linked employeremployee data from the Veneto region of Italy, showing that workers with higher measured outside options recover more quickly from mass layoffs. Improvements in outside options lead to greater job-to-job mobility and higher wages within ongoing employment spells: a one standard deviation increase raises wages by approximately 1.4%. An instrumental variable design based on large-firm openings confirms the causal reading of these results. The findings replicate in Portuguese linked employer-employee data, where increases in employers' value added also pass through more strongly to the wages of workers with better outside options. Taken together, these findings support wage bargaining as an empirically relevant mechanism of wage determination. |
| Keywords: | Outside options, wage bargaining, employer mobility, monopsony, discrete choice, neural networks |
| Date: | 2026–10 |
| URL: | https://d.repec.org/n?u=RePEc:hal:ipppap:halshs-05845509 |
| By: | Johan Fourie |
| Abstract: | Writing produces two things: the manuscript published now and the capability its author will hold next year. Artificial intelligence can raise the first while lowering the second, and which happens depends on the task the author delegates. The relevant comparison is between workflows rather than between using and refusing AI. This article separates writing into its tasks and applies a standard tool of economics to each, synthesising existing models into five predictions. Cheaper drafting lets an author test more versions of an argument, and the largest gains should go to researchers writing in a second language, whose language costs are unrelated to their evidence. Manuscript studies, including new evidence from economics working papers, fit this pattern, although they measure prose and output rather than argument quality. Polishing prose with a model while delegating the first formulation of a problem can leave a better text and a weaker author, and it makes the generated draft an anchor and a default. Cheap fluency makes polished prose a weaker guide to the underlying research. The evidence is uneven, and I note where it is thin. As models improve, the bottleneck moves, and so does the division of labour between author and machine. |
| Keywords: | artificial intelligence, scholarly writing, human capital, language barriers, anchoring, judgement |
| JEL: | D83 I23 J24 O33 |
| Date: | 2026 |
| URL: | https://d.repec.org/n?u=RePEc:sza:wpaper:wpapers399 |
| By: | Simpson, Scott |
| Abstract: | This article presents a reproducible public-data methodology for constructing institutional peer groups in higher education. Using Integrated Postsecondary Education Data System (IPEDS) data, we develop a two-stage k-prototypes clustering workflow that accommodates both numerical and categorical institutional features. The first stage identifies a broad institutional comparison set across all IPEDS institutions, while the second stage reclusters institutions within the focal institution’s global cluster to identify a more focused local peer group. Using St. Bonaventure University as a case example, the procedure identified a global comparison cluster of 1, 132 institutions and a local peer cluster of 247 institutions. We further use a random forest classifier and shallow decision-tree surrogate to summarize the institutional characteristics that distinguish the resulting local clusters. The framework provides a transparent and adaptable approach for institutional peer-group construction using publicly available data and open-source analytical tools. |
| Date: | 2026–10–27 |
| URL: | https://d.repec.org/n?u=RePEc:osf:edarxi:phnaw_v1 |
| By: | Gethin, Amory Lionel |
| Abstract: | This paper surveys the evidence on the role of education in reducing poverty. Education can reduce absolute poverty by increasing productivity, relative poverty by lowering inequality, and multidimensional poverty by improving dimensions of well-being beyond income. Existing research indicates that it contributes to all three. Returns to schooling are large, especially for policies enabling children from disadvantaged backgrounds to acquire more education, and typically exceed the costs of education provision. Education also benefits low-wage workers through labor market spillovers: reducing the share of low-skilled workers in the economy raises their relative wage. The expansion of education has played a major role in reducing extreme poverty and can account for over half of income gains among the world’s poorest individuals since 1980. The non-monetary benefits of schooling are also large and may imply total returns that are twice as high as monetary gains. Substantial educational inequalities persist nonetheless, with important implications for global equity and the trajectory of poverty in the 21st century. |
| Date: | 2026–10–28 |
| URL: | https://d.repec.org/n?u=RePEc:wbk:wbrwps:11491 |
| By: | Hall, Tessa |
| Abstract: | This paper estimates the firm productivity returns to degree field, institution quality, and degree level, and compares them with wage returns. I link a LinkedIn panel from Revelio Labs to Orbis financial accounts for over 20, 000 UK firms over 2015?2024. Within firms, a bachelor's degree raises a worker's productivity by around 20%; a top-20 institution adds a further 10% and a postgraduate qualification 6%. The returns differ by organisational role. A bachelor's degree raises productivity for managers and non-managers alike. Specialist credentials - a STEM field, an elite institution - raise productivity for non-managers but not for managers, whose premium loads instead on postgraduate qualifications, above all MBAs. Managers also contribute around 26% more than otherwise identical non-managers, a premium that education does not explain. Averaged over the workforce, wage premia are close to productivity premia for each type of degree. However, the average conceals two wedges of opposite sign: STEM-trained non-managers are paid below their relative productivity, and degree-holding managers above it. Finally, for a subset of firms matched to the World Management Survey, workforce education accounts for around a quarter the well-documented correlation between management practices and productivity. |
| Keywords: | human capital;firm productivity;education returns;management |
| JEL: | J24 J31 D24 M11 |
| Date: | 2026–10–23 |
| URL: | https://d.repec.org/n?u=RePEc:ehl:lserod:141789 |
| By: | Riccardo Zanardelli |
| Abstract: | Organizations pairing a human and a machine on a task decide, often by default, which executes each instance and what happens when both do. We account for what this attains under the squared rule, with performance distributed over instances to capture generalization difficulty. A gross stake lies between the skills' mean and best per instance. It splits at the best single skill into a systematic part, known from separate assessments and secured by commitment, and a residual, the expected value of perfect information on which skill each instance favors, known only from paired observation and split further by classes. Skills set the stake; workflows decide the take, via two channels. Routing directs work, weighting or selecting responses; with binary truth it is capped by the best skill per instance. Reading responses as joint evidence is not routing; what they jointly reveal caps it. Synergy lets skills interact before responding; bounded by no skill either way, its lift over independent responses must clear a break-even and can fall below one. As algorithms, arrangements form a ladder stating what each reads, holds and records, and a procedure whose checks keep earlier gains when class means are known and lose little on finite records given a margin. The theory extends to many skills, each worth what it adds where it beats all others held, and to competing skills, where at parity a single failure's allowed cost decides. In a public chest-radiograph study with 227 radiologists, they are near parity with the model and the stake is mostly residual; assisted reading, no better on average originally, returned less than averaging both responses; recalibrating the model alone beats any deployable arrangement of raw responses. Organizations can learn from their records what is at stake, which arrangement collects it and what evidence each step needs; pricing costs and errors is future work. |
| Date: | 2026–10 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2610.16863 |