CoChat
Sign Up
Cited by 406Labor market dynamics and wage inequalityEconomic Growth and ProductivityEconomic Theory and Policy

The DOI matches its Crossref record; 33 of the 34 references it cites were found in a registry or an index; none of the 9 sources its claims cite is checked for support yet.

The Race Between Machine and Man: Implications of Technology for Growth, Factor Shares and Employment

Open access
Acemoglu, Daron, Restrepo, Pascual·2016·SSRN Electronic Journal
DOI 10.3386/w22252
Open this paper in CoChat — free account

Abstract

We examine the concerns that new technologies will render labor redundant in a framework in which tasks previously performed by labor can be automated and new versions of existing tasks, in which labor has a comparative advantage, can be created.In a static version where capital is fixed and technology is exogenous, automation reduces employment and the labor share, and may even reduce wages, while the creation of new tasks has the opposite effects.Our full model endogenizes capital accumulation and the direction of research towards automation and the creation of new tasks.If the long-run rental rate of capital relative to the wage is sufficiently low, the long-run equilibrium involves automation of all tasks.Otherwise, there exists a stable balanced growth path in which the two types of innovations go hand-in-hand.Stability is a consequence of the fact that automation reduces the cost of producing using labor, and thus discourages further automation and encourages the creation of new tasks.In an extension with heterogeneous skills, we show that inequality increases during transitions driven both by faster automation and introduction of new tasks, and characterize the conditions under which inequality is increasing or stable in the long run.

Overview

AI snapshot

Written from the paper’s full text.

TL;DR

The paper develops a task-based model in which automation replaces labor in existing tasks while new labor-complementary tasks are created, showing that automation reduces employment and the labor share while new tasks raise them. With endogenous directed technology, balanced growth with both types of innovation is stable unless capital is sufficiently cheap relative to labor, in which case full automation can occur.

Key findings

  • In the static model, automation always reduces the labor share and employment and may reduce wages, while creation of new tasks increases wages, employment, and the labor share.
  • In the full model, if the long-run rental rate of capital is sufficiently low, the long-run equilibrium involves automation of all tasks; otherwise there is a stable balanced growth path where automation and new-task creation advance together.
  • Empirically, occupations with 10 percentage points more new job titles at the start of a decade grew 5.05% faster over the decade (SE = 1.29%); new job titles account for about 8.84% of the 17.5% employment growth from 1980–2007 in the baseline stacked specification, with long-difference estimates between 7.27% and 8.5%.
  • In 2000, about 70% of computer software developers held new job titles; occupations with 10 percentage points more new job titles hired workers with 0.35 more years of schooling initially, but after 30 years hired workers with 0.0411 fewer years of education (SE = 0.0176), and occupations with one additional year of schooling at the start reduced average years of schooling by 0.149 years after 30 years (SE = 0.021).

Methods

  • Theoretical framework: a task-based model with a unit measure of tasks, capital and labor allocated across tasks, automation expanding the set of tasks capital can perform, and creation of new tasks in which labor has comparative advantage; static model with fixed capital and exogenous technology, then dynamic model with capital accumulation and exogenous technology, then full model with endogenous directed technological change.
  • Empirical analysis: uses Lin (2011) data on new job titles within 330 consistently defined occupations, U.S. Censuses for 1980, 1990, 2000, and the American Community Survey for 2007, focusing on workers aged 16–64; regressions of decadal employment growth and changes in average years of schooling on the share of new job titles, with demographic controls and robust standard errors.
  • Main analysis: characterizes static equilibrium and comparative statics, balanced growth paths with exogenous and endogenous technology, stability conditions, and extensions for heterogeneous skills, creative destruction, and welfare with labor market frictions.

Limitations

  • The paper states it is “a first step” and notes that the model imposes automation of tasks at the bottom, whereas in reality middling tasks may be automated; it also calls for incorporating technological barriers across industries and investigating other labor-complementary technologies.
  • The paper states that additional empirical evidence is needed on how automation impacts employment and wages and how incentives for automation and new-task creation respond to policies, factor prices, and supplies.
  • The text makes evident that the analysis relies on several simplifying assumptions, including Assumption 2 (homotheticity via η → 0 or ζ = 1), Assumption 4 (σ > ζ), and a small supply of scientists S for some stability results; the NBER working paper is also circulated for discussion and has not been peer-reviewed.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.3386/w22252

Cites

33 of 33 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Verification

Claims the paper makes

0 of 9 cited sources checked · 0 supported · 0 flagged

Show all 95 claims

  • It may state two separate claims as one. 75
  • The quote only partly states this claim. 11
  • Probably not a substantial claim. 8
86 of 86

  • The authors state this as an assumption.

    Labor has a strict comparative advantage in tasks with higher indices.

    “Assumption 1 γ(i) is strictly increasing”
    “Assumption 1 implies that labor has strict comparative advantage in tasks with a higher index”

  • The authors state this as an assumption.

    The demand for labor and capital is homothetic, which holds if either η approaches 0 or ζ equals 1.

    “Assumption 2 One of the following two conditions holds: (i) η → 0, or (ii) ζ = 1.”
    “These two special cases ensure that the demand for labor and capital is homothetic.”

  • The authors state this as an assumption.

    The capital stock is below a threshold K defined by R = W/γ(N), ensuring that new tasks are adopted immediately.

    “Assumption 3 Let K be such that R = W γ(N ) .We have K < K.”
    “This assumption ensures that R > W γ(N ) and consequently that new tasks will increase aggregate output and will be adopted immediately.”

  • The authors state this as an assumption.

    Skilled labor has a comparative advantage in new tasks.

    “we introduce heterogeneity in skills, and assume that skilled labor has a comparative advantage in new tasks, which we view as a natural assumption.”

  • The authors state this as an assumption.

    New tasks created at N immediately replace tasks at N-1 and are profitable to produce with labor.

    “we have simplified the exposition by imposing that new tasks created at N immediately replace tasks located at N -1, and it is therefore profitable to produce new tasks with labor”

  • The authors state this as an assumption.

    The representative household has dynamic preferences given by equation (14), with a positive discount rate rho.

    “We assume that the representative household's dynamic preferences are given by ∞ 0 e -ρt u(C(t), L(t))dt,(14) where u(C(t), L(t)) is as defined in equation (4) and ρ > 0 is the discount rate.”

  • The authors state this as an assumption.

    The comparative advantage schedule satisfies gamma(i) equals e to the power A i, with A positive.

    “Assumption 1 ′ γ(i) satisfies: γ(i) = e Ai with A > 0. (15”

  • The authors state this as an assumption.

    In the dynamic model before the full model, the path of technology represented by I(t) and N(t) is exogenous.

    “The path of technology, represented by {I(t), N (t)}, is exogenous, and we define”

  • The authors state this as an assumption.

    Only tasks above N minus one remain compatible with current technology and can be combined with tasks currently in use.

    “As N increases, as emphasized by equation (1) and in footnote 9, the set of feasible tasks shifts to the right, and only tasks above N -1 remain compatible with and can be combined with those currently in use.”

  • The authors state this as an assumption.

    In the full model, intellectual property rights to each intermediate are held by a technology monopolist that can produce at marginal cost mu psi; a competitive fringe can copy at marginal cost psi; mu is such that the unconstrained monopoly price exceeds psi; the equilibrium price is the limit price psi; and monopolists earn per-unit profit (1 minus mu) psi.

    “To endogenize technological change, we deviate from our earlier assumption of a perfectly competitive market for intermediates, and assume that (intellectual) property rights to each intermediate, q(i), are held by a technology monopolist which can produce it at the marginal cost µψ in terms of the final good, where µ ∈ (0, 1) and ψ > 0. We also assume that this technology can be copied by a”

  • The authors state this as an assumption.

    Developing a new intermediate that automates or replaces an existing task infringes the patent on the previous technology, so the new inventor must compensate the incumbent patent holder through a take-it-or-leave-it offer.

    “We assume that developing a new intermediate that automates or replaces an existing task is viewed as an infringement of the patent of the technology previously used to produce that task.Consequently, a firm must compensate the technology monopolist who owns the property rights over the production of the intermediate that it is replacing.We also assume that this compensation takes place with the”

  • The authors state this as an assumption.

    There is a fixed supply of S scientists allocated between automation and new task creation; a scientist in automation automates kappa_I tasks per unit time and earns W_SI, while a scientist in new task creation creates kappa_N tasks per unit time and earns W_SN.

    “There is a fixed supply of S scientists, which will be allocated to automation (S I (t) ≥ 0) or the creation of new tasks (S N (t) ≥ 0), S I (t) + S N (t) ≤ S. When a scientist is employed in automation, she automates κ I tasks per unit of time and receives a wage W S I (t).When she is employed in the creation of new tasks, she creates κ N new tasks per unit of time and receives a wage W S N (t).”

  • The authors state this as an assumption.

    Scientists differ in effort cost; the distribution of the difference in effort costs is smooth and increasing over a bounded support, the support is small enough that all scientists work, and a normalization sets G(0) equal to kappa_N divided by kappa_I plus kappa_N.

    “In particular, we assume that scientists differ in the cost of effort: when working in automation, scientist j incurs a cost of χ j I Y (t), and when working in the creation of new tasks, she incurs a cost of χ j N Y (t).24 Consequently, scientist j will work in automation if W S I (t)-W S N (t) Y (t) > χ j N -χ j I . We also assume that the distribution of χ i N -χ i I among scientists is given”

  • The authors state this as an assumption.

    Assumption 4: sigma is greater than zeta.

    “Assumption 4 σ > ζ.”

  • The authors state this as an assumption.

    In the equilibrium analyzed, n(t) exceeds the maximum of the two thresholds, so I* equals I and newly automated tasks immediately use capital.

    “To simplify the exposition, let us assume that in this equilibrium n(t) > max{n(ρ), n(ρ)}(ρ), so that I * (t) = I(t) and newly-automated tasks start being produced with capital immediately.”

  • The paper relies on this without support.

    Patent holders immediately accept take-it-or-leave-it offers and reject less generous ones.

    “In both cases, the patent-holders will immediately accept theses offers and reject less generous ones.”

  • The authors state this as an assumption.

    High-skill workers have a comparative advantage in new tasks.

    “The natural assumption that high-skill workers have a comparative advantage in new tasks receives support from the data.”

  • The authors state this as an assumption.

    There are two types of workers: low-skill with time-varying productivity γ_L(i,t) in task i, and high-skill with productivity γ_H(i).

    “we assume that there are two types of workers: low-skill workers with time-varying productivity γ L (i, t) in task i, and high-skill workers with productivity γ H (i).”

  • The authors state this as an assumption.

    The supply of high-skill labor is fixed at H and the supply of low-skill labor is fixed at L.

    “we focus on the economy with exogenous technology and assume that the supply of high-skill labor is fixed at H and the supply of low-skill labor is fixed at L.”

  • The authors state this as an assumption.

    New technologies do not infringe the patents of the products they replace.

    “we modify our baseline assumption on intellectual property rights and revert to the classical setup in the literature in which new technologies do not infringe the patents of the products that they replace”

  • The authors state this as an assumption.

    The function ι(n(t)) is nondecreasing.

    “Here, the function ι(n(t)) is included and assumed to be nondecreasing to capture the possibility that automating tasks closer to the frontier (defined as the highest indexed task available) may be more difficult.”

  • The authors state this as an assumption.

    There exists an upward-sloping quasi-labor supply schedule L_qs(ω) that constrains employment so that L ≤ L_qs(ω).

    “suppose that there exists an upward-sloping quasi-labor supply schedule, L qs (ω), which constrains the level of employment, so that L ≤ L qs (ω).”

  • The authors state this as an assumption.

    The quasi-labor supply schedule L_qs(ω) is increasing in ω, has elasticity ε_L>0, and is binding in the sense that W/C > ν'(L).

    “Suppose also that the quasi-labor supply schedule L qs (ω) is increasing in ω, has an elasticity ε L > 0, and is binding in the sense that W C > ν ′ (L).”

  • The authors state this as an assumption.

    Labor market frictions also constrain the social planner's choices.

    “assuming that labor market frictions also constrain the social planner's choices”

  • The authors state this as an assumption.

    Assumptions 1', 2', and 4 hold.

    “Suppose that Assumptions 1 ′ , 2 ′ and 4.”

  • The authors state this as an assumption.

    Assumption 2' holds, allowing the proof of Proposition 5 to be given in a more general case.

    “We prove the proposition in the more general case in which Assumption 2' holds.”

  • The authors state this as an assumption.

    Assumption 2'' holds: one of three conditions is satisfied, namely σ-ζ tends to zero, or ζ tends to one, or η tends to zero, which guarantees homothetic factor demand.

    “Assumption 2 ′′ : One of the following three conditions holds: 1. σ -ζ → 0, or 2. ζ → 1, or 3. η → 0.”

  • The authors state this as an assumption.

    Assumption 4 holds, imposing σ > ζ and ensuring that capital and labor are complements to intermediates.

    “We show that this property holds under Assumption 4, which imposes that σ > ζ, and ensures that capital and labor are complements to intermediates.”

  • The authors state this as an assumption.

    Assumption 2 holds, so that X equals η/(1-η) times F(k, L; n).

    “(When Assumption 2 holds, we have X = η 1-η F (k, L; n).”

  • The authors state this as an assumption.

    Assumptions 1, 2, and 3 hold, and I* = I > I, for the welfare analysis in Proposition 9.

    “Consider the static economy and suppose that Assumptions 1, 2 and 3 hold, and that I * = I > I.”

  • The authors state this as an assumption.

    Newly created tasks are higher-productivity versions of existing tasks.

    “We view newly-created tasks as higher productivity versions of existing tasks.”

  • The authors state this as an assumption.

    Task i is not compatible and will not be used together with tasks i′ < i−1.

    “We are also assuming that task i is not compatible and will not be used together with tasks i ′ < i -1”

  • The authors state this as an assumption.

    Assumption 2′: one of the following holds: (i) η→0, or (ii) ζ=1, or (iii) γ(N−1)/γ(N) 2+2σ+η > |σ−ζ|.

    “Assumption 2 ′ One of the following three conditions holds: (i) η → 0, or (ii) ζ = 1, or (iii) γ(N -1) γ(N ) 2+2σ+η > |σ -ζ|.”

  • The authors state this as an assumption.

    In the efficiency-wage formulation, u(i) measures how untrustworthy worker i is, and firms observe this information.

    “In this formulation, u(i) measures how untrustworthy worker i is, and we assume that this information is observed by firms.”

  • The authors state this as an assumption.

    In the alternative case, workers draw a new value of u(i) at each point in time.

    “To achieve that, we assume that workers draw a new value of u(i) at each point in time.”
35 of 35

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

Each contradiction between two passages listed here was confirmed by an independent check.

  • Reported range excludes two table estimates

    Minor

    The stated range for Panel B's contribution estimates excludes the estimates of 0.16 and 2.17 shown in its last two columns.

  • New tasks are said to lower a measure they increase

    Minor

    The paper defines n as N minus I and explicitly says that introducing new tasks increases n. Its later description instead says that creating new tasks leads to a lower n.

2 of 2

Citations

Cites

Flagged first, then by citations

Show all 34 references

1 of the 34 references couldn’t be matched to a known paper.

Research works printed but not linked: 16

  1. Akst, Daniel (2013) "What Can We Learn From Past Anxiety Over Automation?" Wilson Quarterly.
  2. Chandler, Alfred D. (1977) The Visible Hand: The Managerial Revolution in American Business, Harvard University Press: Cambridge, MA.
  3. Jones, Charles I. (2005) "The Shape of Production Functions and the Direction of Technical Change," The Quarterly Journal of Economics, 120(2): 517-549.
  4. Karabarbounis, Loukas and Brent Neiman (2014) "The Global Decline of the Labor Share," The Quarterly Journal of Economics, 129(1): 61-103.
  5. Kiley, Michael T. (1999) "The Supply of Skilled Labor and Skill-Biased Technological Progress," The Economic Journal, 109(458): 708-724.
  6. Landes, David (1969) The Unbound Prometheus, Cambridge University Press: New York.
  7. Lin, Jeffrey (2011) "Technological Adaptation, Cities, and New Work" Review of Economics and Statistics, 93(2): 554-574.
  8. Mokyr, Joel (1990) The Lever of Riches: Technological Creativity and Economic Progress, Oxford University Press: New York.
  9. Nelson, Richard R. and Edmund S. Phelps (1966) "Investment in Humans, Technological Diffusion, and Economic Growth," The American Economic Review, 56(1): 69-75.
  10. Rodriguez-Clare, Andres (2010) "Offshoring in a Ricardian World," American Economic Journal: Macroeconomics, 2: 227-258.
  11. Roy, A. D. (1951) "Some Thoughts on the Distribution of Earnings," Oxford Economic Papers, 3(2): 135-146.
  12. Schultz, Theodore (1975) "The Value of the Ability to Deal with Disequalibria." Journal of Economic Literature. 13:827-846.
  13. The New York Times (1983) "Machines vs. Workers," by Charlotte Curtis, February 8th edition.
  14. Thoenig, Mathias, and Thierry Verdier (2003) "A Theory of Defensive Skill-Biased Innovation and Globalization," American Economic Review, 93(3): 709-728.
  15. Zeira, Joseph (1998) "Workers, Machines, and Economic Growth," Quarterly Journal of Economics, 113(4): 1091-1117.
  16. Zeira, Joseph (2006) "Machines as Engines of Growth," Dynamics, Economic Growth, and International Trade Conference Papers. No. 011059.

Cited by

Sorted by citations

Items 301–400 of 405

Related

See all

Metadata

Authors
Acemoglu, D., & Restrepo, P.
Affiliations
Boston University (US)
Venue
SSRN Electronic Journal
Publisher
National Bureau of Economic Research
Published
2016-05
ISSN
1556-5068 (ISSN-L)
DOI
10.3386/w22252
OpenAlex
W3122373805
Cited by
406
References
34
Citations by year
2026: 8 · 2025: 58 · 2024: 19 · 2023: 23 · 2022: 51 · 2021: 50 · 2020: 32 · 2019: 48 · 2018: 75 · 2017: 39 · 2016: 2
Type
posted content
Language
en
Open access
Locations
  • SSRN Electronic Journal (publishedVersion)
  • National Bureau of Economic Research (publishedVersion)
  • RePEc: Research Papers in Economics (submittedVersion)
Funders
  • Lynde and Harry Bradley Foundation
  • Yale University
  • Yeshiva University
  • University of Cambridge
  • University of Oxford
Topics
  • Social Sciences › Economics, Econometrics and Finance › Economics and Econometrics › Labor market dynamics and wage inequality
  • Social Sciences › Economics, Econometrics and Finance › Economics and Econometrics › Economic Growth and Productivity
  • Social Sciences › Economics, Econometrics and Finance › General Economics, Econometrics and Finance › Economic Theory and Policy
Keywords
automation, job displacement, task creation, labor share, employment, capital accumulation, balanced growth, endogenous innovation, wage effects, skill heterogeneity, income inequality, technological change

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

A language model drew the claims from the paper on 2026-10-07, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).

A second model labelled each claim on 2026-10-07 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review