The Commoditization of Labor (w35815)
A · Strong evidence NBER working paper by Masao Fukui, Emi Nakamura and Jón Steinsson, prominent macroeconomists, with a fully stated model, a calibration to US firm-size and wage data and checkable hiring-cost evidence; held at A because the headline effects are model outputs and the direct evidence rests on a convenience survey and industry ratios.
Summary based on the full text·Original: about 60 minutes to read
In short
NBER Working Paper 35815 (September 2026) by Masao Fukui (Boston University), Emi Nakamura and Jón Steinsson (UC Berkeley). When firms standardise tasks, output depends less on who does the job and more on the firm's technology. Firms become less picky about whom they hire, vacancies fill faster, and the firm's outside option improves. Output per worker rises while wages fall, so wage markdowns grow. In a calibrated model, extra standardization by the largest firms reproduces most of the collapse of the large-firm wage premium since the 1980s, which standard monopsony models predict should have risen.
Key numbers
- about 8%Fall in relative wages at megafirms in the model once they raise standardization
- more than 25%Rise in output per worker at megafirms in the same model experiment
- nearly 40%Fall in hiring costs at megafirms in the model, driven by lower hiring standards
- almost 60%Wage premium of megafirms over small firms in the 1980s data, down to about 20% in the 2010s (Bloom et al. 2018)
- 4.3 percentage pointsRise in the employment share of megafirms (10,000+ staff) from 1980 to 2023, the calibration target
As reported by the source.
English·Web page·opens nber.org
What it reports
- Standardization lowers the match quality a firm will accept (Proposition 1). For high-productivity firms it raises output per worker, lowers the average wage and lowers the firm's labor share (Proposition 2).
- The same firms fill vacancies faster and grow without raising pay. Workers find jobs more easily, but the jobs pay less: the authors call them "bad jobs" ex post.
- Raising firm productivity and raising standardization have opposite wage effects. Higher productivity raises wages; higher standardization lowers them.
- If all firms standardise equally, the aggregate labor share does not change: workers can also quit and walk to an equally standardised rival. The aggregate labor share falls only when the firms with the strongest outside options also fill vacancies fastest.
- High-productivity firms have the strongest incentive to invest in standardization, and the resulting equilibrium is efficient in the sense of maximising aggregate consumption.
- Calibrated experiment: megafirms raise standardization from 0.5 to about 0.63, their relative wages fall about 8%, output per worker rises more than 25%, hiring costs fall nearly 40%. The modelled megafirm wage premium drops from more than 50% to about 30%, against a fall from almost 60% to about 20% in the data.
- A rival explanation, wider productivity dispersion across firms, matched to the same 4.3-point rise in megafirm employment share, pushes the large-firm wage premium up, the opposite of what happened.
- SHRM benchmarking data (2022 and 2025): median cost per hire falls steadily with firm size; at firms with more than 5,000 staff it is roughly one third of the level at 1-99 staff firms. OES and JOLTS estimates put cost per hire at $105 in accommodation and food services and $250 in retail trade.
How they got it
A continuous-time directed search model with wage posting, durable jobs and fixed job creation, where a share s of each job's tasks is standardised and output equals z^s times worker match quality^(1-s). Analytical results come with proofs in the appendices; robustness versions cover random search with Nash bargaining, endogenous firm entry and an endogenous number of positions. The quantitative model is calibrated to the 1980 Business Dynamics Statistics firm-size distribution and the Bloom et al. (2018) firm-size wage premium for 1980-1986, then given a convex standardization cost fitted to the 1980-2023 rise in megafirm employment share. Hiring-cost evidence comes from two SHRM member surveys and an OES-JOLTS ratio of HR costs to hires by industry.
Limits
- The 8%, 25% and 40% effects are outputs of a calibrated model, not estimates from firm or worker data. The calibration fits only the megafirm employment share change; the authors note that letting more firms standardise overstates concentration.
- The SHRM data are a convenience sample of paying members and not representative of US firms; the cost-per-hire link is cross-sectional.
- Results depend on durable jobs and costly entry. The authors state they do not hold in a standard free-entry setting where jobs disappear when a worker leaves; with imperfectly durable vacancies the main results flip.
- Standardization is not measured directly in any firm; it is inferred from size and wage patterns.
- The paper does not discuss AI. It treats ICT-driven task simplification since 1980 (warehouse handhelds, scripted customer service, fast-food routines); applying it to AI is the reader's inference.
- Working paper, not yet peer reviewed.
Why it matters for companies
The paper separates two routes to cheaper labor that most AI business cases blur. Automation removes the worker from a task. Standardization keeps the worker but removes the need for that particular worker, and the paper shows that this alone can raise output per head while cutting pay and hiring cost. AI copilots, scripted flows and guided workflows that turn expert judgement into steps are standardization in this sense. For a single large employer the model says the gain is real and comes partly out of wages, through lower hiring standards and easier replacement. It also says the advantage fades if every competitor standardises the same way. HR leaders redesigning roles around AI should expect lower time-to-fill and training cost, track what happens to pay, tenure and the value of experience in those roles, and decide deliberately whether they are building interchangeable jobs or keeping the judgement-heavy tasks that make staff hard to replace.
Summary based on the full text. Read the original for the full argument and data.Read the original on nber.org
- Tags
- Future of work, Wages