Technology, Inequality, Market Structure, And Labor-Market Institutions

Learning Objectives¶

By the end of Week 5, students should be able to:

1. explain how technological change can alter who captures rents even when productivity rises;

2. connect task and skill inequality to labor share, superstar firms, intangible rents, employer concentration, and wage setting;

3. compare mechanization, electrification, computerization, robots, and AI as persistent but historically distinct labor-market shocks;

4. distinguish employment reallocation from welfare-improving reallocation across agriculture, manufacturing, formal services, and informal services;

5. evaluate how unions, bargaining, platform rules, and labor-market institutions mediate technology incidence;

6. use market-level, firm-level, historical, cross-country, and infrastructure data to design labor-focused studies of technology, inequality, and structural change.

The central market-level question is this: when technology changes what firms can produce and how workers are allocated, who captures the gains, which workers bear risk, and which institutions shape the distribution?

Opening Orientation¶

The earlier weeks tracked technology from tasks to labor demand, worker adjustment, and firm adoption. Week 5 asks what happens when those adjustments scale up to markets. Who captures the gains from new technology? Do productivity improvements translate into higher wages, lower prices, or larger rents? Does technology increase competition, or does it reinforce concentration and labor-market power? And when workers reallocate across sectors, do they move into better jobs, more precarious jobs, or simply different low-productivity activities?

This lecture treats technology as a market-level labor force. The same innovation can raise productivity while lowering labor share, increasing concentration, changing bargaining institutions, and reallocating workers across sectors with very different job quality. The point is not to decide whether technology is good or bad. The point is to identify the channels through which its gains and losses are distributed.

Bridge¶

Weeks 2 through 4 studied demand shocks, worker adjustment, and firm adoption. Week 5 aggregates these pieces. Technology no longer appears only as a firm or worker shock; it becomes a market-level force that changes concentration, labor share, structural change, and the institutional environment governing work.

The bridge from Week 4 is direct. If better-managed or larger firms adopt technology earlier, if high-skill workers capture more augmentation, if low-wage workers face monitoring or substitution, and if incumbents are trained unevenly, then firm adoption becomes a mechanism for economy-wide inequality. Week 5 asks how those firm-level choices aggregate into rents, labor-market power, worker welfare, and institutional response.

Field Core¶

From Adoption To Market-Level Incidence¶

A useful way to discipline the lecture is to treat market outcomes as the composition of three margins:

1. within-firm productivity and wage effects;

2. between-firm reallocation, where more productive or more technology-intensive firms gain market share;

3. between-sector reallocation, where labor moves across agriculture, manufacturing, formal services, informal services, and digital service channels.

This decomposition explains why micro evidence and macro evidence can differ. A technology may raise productivity within adopting firms but reduce aggregate labor share if adoption shifts demand toward firms with lower labor shares, higher markups, more intangible capital, or weaker rent-sharing norms Autor et al., 2020Karabarbounis, 2023.

where $y^{L}$ is a labor outcome such as labor share, wage premia, job quality, worker voice, or employment quality; $s_{fm}$ is firm market share within market $m$; and $\omega_{jm}$ is sectoral employment share. Equation (1) is not a full structural model. It is a teaching map: technology incidence can arise inside firms, through reallocation across firms, and through reallocation across sectors.

Technology, Inequality, Market Structure, and Labor Institutions¶

Technology, Task Inequality, And Rent Capture¶

The first inequality channel is task and skill demand. Technologies that substitute for routine tasks can lower demand for workers whose comparative advantage was in those tasks, while raising demand for workers who perform abstract, social, technical, problem-solving, maintenance, or coordination tasks. This is the familiar task-biased and skill-biased logic from earlier weeks, now read as a distributional mechanism Autor et al., 2003Goos et al., 2014.

The second channel is rent capture. Technology can generate quasi-rents through patents, proprietary data, software, organizational capital, brands, network effects, or scale economies. The labor question is not only whether rents exist, but who shares in them. Innovative firms may generate rents that flow to shareholders, executives, inventors, high-skill complements, or workers through rent sharing. Patent-rent evidence is useful because it shows that innovation gains can reach workers, but unevenly across firms and worker groups Kline et al., 2019Card et al., 2018.

The third channel is wage setting. If technology strengthens the position of a small number of employers, reduces worker outside options, or moves work into platform and contractor arrangements, then productivity gains can coexist with weaker wage growth for exposed workers. Employer concentration evidence is therefore part of the technology lecture when technology changes which firms grow and how workers search, match, bargain, or organize Azar et al., 2022Benmelech et al., 2022.

The fourth channel is institutions. Unions, bargaining coverage, wage floors, occupational licensing, platform rules, algorithmic management rules, data rights, transparency obligations, and worker voice mechanisms can alter who captures the gains from a technology. Treating institutions as afterthoughts misses the main labor mechanism: technology changes the surplus, but institutions help decide the incidence.

Labor Share, Superstar Firms, And Intangible Rents¶

One central market-level question is whether technology raises the share of value added captured by capital, intangibles, or dominant firms rather than labor. The superstar-firm literature argues that technology and globalization can shift sales toward more productive, more capital- or intangible-intensive firms, thereby lowering aggregate labor share even if nothing within a representative firm looks mechanically anti-labor Autor et al., 2020.

This literature is useful for labor economists because it links technology to:

• labor share;

• firm wage premia;

• rent sharing;

• concentration;

• markups and intangible assets;

• and the distribution of good jobs.

The empirical object is not simply concentration. It is the joint movement of concentration, labor share, productivity, wages, and firm composition. A rise in superstar firms can be pro-productivity and still change who captures surplus. A labor-focused interpretation asks whether workers in expanding firms share in the rents, whether displaced workers can move into those firms, and whether labor-market frictions prevent reallocation from becoming worker welfare.

Market Structure, Labor-Market Power, And Institutions¶

Technology does not only change production. It can change the institutional environment in which wages are set. Platform technologies can reshape contracting boundaries, scheduling, rating systems, algorithmic control, and the allocation of risk. Intangible-heavy production may strengthen firms with scale advantages. Automation may weaken worker voice in some sectors while creating new organizing challenges in others.

A labor-economics way to state the problem is:

• who captures the rents created by technology;

• which workers are complement, substitute, or residual claimants;

• which institutions mediate rent sharing;

• and which workers are protected by bargaining, transparency, competition, or regulation?

Unions and bargaining institutions matter here because they can change the distribution of technology gains without changing the engineering frontier. Platform rules matter because they can convert a technical system into a labor-market governance system. Antitrust and labor-market competition policy matter when technology reallocates demand toward employers with stronger wage-setting power.

Historical Technology Waves: Persistence, Reversal, And Interaction¶

Current AI and automation research becomes more legible once it is compared to earlier technology waves. Mechanization, electrification, computerization, and robotics all changed task content, firm boundaries, production scale, and sectoral employment. They differed in speed, complementarity, diffusion, and persistence.

The comparison should ask five questions:

1. how long do technology-induced labor-market changes last?

2. when are they reversible?

3. when does a second technology wave offset an earlier one?

4. when do successive waves reinforce each other?

5. when is AI genuinely new rather than another task-reallocation episode?

Mechanization shifted human and animal effort toward machine power and reorganized production around capital-intensive methods. Electrification changed factory layout, workflow, and the timing of productivity gains. Computerization substituted for routine codifiable tasks while complementing abstract and analytical work. Robots made embodied automation measurable in modern manufacturing. AI extends the comparison to prediction, language, coding, decision support, monitoring, and task delegation Autor, 2015Goldin & Katz, 1998Atack et al., 2020Fiszbein et al., 2020Acemoglu & Restrepo, 2020.

Historical evidence matters because the labor consequences of technology often persist through complementary investments, worker cohorts, geography, institutions, and firm organization. But persistence is not inevitability. New tasks, expanding demand, worker retraining, institutional response, and later technologies can partially offset displacement. The right conclusion is not that AI is unprecedented or ordinary. The right conclusion is that AI must be located in a sequence of task reallocation, adoption costs, institutional mediation, and equilibrium adjustment Autor, 2015Acemoglu & Restrepo, 2019Acemoglu, 2019.

Historical, Global, and Structural-Change Extensions¶

Structural Change And Job Quality¶

Technology is also a structural-change force. In developing economies, technical change in agriculture can release labor into manufacturing or services. In high-income economies, automation, trade, and digital tools can accelerate reallocation out of manufacturing and toward services. But these transitions are not mechanically welfare-improving.

Labor economists should distinguish four objects:

1. employment reallocation, where workers move across sectors or firms;

2. productivity-enhancing reallocation, where labor moves toward higher-productivity uses;

3. job-quality reallocation, where workers move toward jobs with better wages, stability, voice, safety, schedules, and mobility;

4. welfare-improving reallocation, where workers are better off after accounting for earnings, amenities, risk, family constraints, location, and adjustment costs.

These objects can move in different directions. A worker who leaves manufacturing for services may remain employed but lose wage premia, schedule stability, bargaining coverage, or career ladders. A worker who leaves agriculture for informal services may move into a more urban labor market without entering a high-productivity or high-quality job. Conversely, formal services, health, logistics, business services, and digital knowledge services can absorb labor in ways that raise productivity and create better jobs if institutions and complementary investments support that transition McMillan & Rodrik, 2011Herrendorf et al., 2013Rodrik et al., 2024.

Recent structural-change work also emphasizes that reallocation happens across firms and within firms. The rise of knowledge services, outsourcing, and business-service production changes where workers are counted and how labor inputs are organized. A manufacturing firm may outsource services, automate production, and expand headquarters or technical occupations at the same time. Measured sectoral change therefore needs firm-level and establishment-level discipline Ding et al., 2022Helm et al., 2023.

Global And Global South Evidence¶

The global picture differs sharply from the standard U.S. automation narrative. In many lower- and middle-income economies, the main issues are not just robot exposure or generative AI exposure. They are technology sophistication, diffusion constraints, management quality, informality, energy reliability, trade integration, and whether services can absorb labor productively.

Cross-establishment evidence shows large dispersion in technology sophistication, especially in developing economies, with implications for productivity and labor demand Cirera et al., 2024. This changes the interpretation of technology shocks. In a high-income setting, the question may be whether robots displace manufacturing workers or whether AI substitutes for clerical tasks. In a Global South setting, the first-order question may be whether firms can adopt productivity-enhancing technologies at all, whether workers can access complementary training, and whether labor moves into formal, productive services rather than informal absorption.

Outsourcing and offshoring add another layer. Digital technologies can relocate tasks, not only jobs. Routine business processes, customer support, coding, data labeling, content moderation, design, accounting, legal support, and other knowledge services can be reorganized across countries. AI may automate parts of these chains, augment workers in them, or create new demand for data, oversight, localization, and infrastructure. The same AI shock may therefore reduce one outsourced task while expanding another service channel.

AI also raises a frontier infrastructure question. Compute-intensive AI can shift electricity demand, data-center location, grid investment, construction employment, local tax bases, and demand for complementary technical labor. These are not generic energy-policy issues in this course. They are labor-relevant when infrastructure demand changes local employment, bargaining environments, spatial inequality, service demand, and the geography of technology adoption Liu & others, 2025International Energy Agency, 2026Stiglitz et al., 2021.

The empirical frontier is still unsettled. A careful labor paper should avoid claiming that AI infrastructure has already transformed labor markets everywhere. It should instead ask which regions receive investment, which workers are hired during construction and operation, which sectors face electricity scarcity or price changes, which countries specialize in digital services, and how those shifts interact with local institutions.

Research Lab¶

The Week 5 research lab follows Reproduce -> Diagnose -> Transfer.

The primary anchor is Autor, Dorn, Katz, Patterson, and Van Reenen’s market-level design linking labor share and superstar firms Autor et al., 2020. It is the cleanest anchor for this week because it connects technological change, concentration, market share reallocation, and labor-market incidence in one empirical frame.

The challenge and extension anchor is Ding, Fort, Michaels, Morrow, and Schott on structural change within versus across firms Ding et al., 2022. It pushes the lab from concentration and labor share toward structural change, service reallocation, and the organization of knowledge work. A second transfer option is Cirera, Comin, Cruz, Lee, and Martins on technology sophistication across establishments, which is useful when students want a development setting Cirera et al., 2024.

The lab should not claim to replicate official estimates unless the needed data and code are locally available. No confidential firm microdata, official Economic Census files, Compustat extracts, establishment technology surveys, or official replication packages are bundled in this repository. The teaching path uses deterministic synthetic data to practice the design logic.

Reproduce. Students construct a small industry-firm panel with market share, labor share, productivity, intangible intensity, and technology intensity. They reproduce the structure of a concentration and labor-share decomposition: aggregate labor share falls when market share reallocates toward low-labor-share, high-productivity, technology-intensive firms.

Diagnose. Students diagnose what the decomposition identifies. They ask whether technology is measured directly or inferred from firm outcomes, whether concentration is product-market or labor-market concentration, whether wage-setting power is visible, whether labor share captures worker welfare, and whether institutions alter incidence.

Transfer. Students transfer the logic to structural change and global evidence. They compare employment reallocation with job-quality reallocation across agriculture, manufacturing, formal services, informal services, and knowledge services. Then they add a frontier AI scenario in which data-center and electricity demand affect local labor markets and outsourced service channels.

Methods Box¶

Methods and Data Map¶

Reading Ladder And References¶

Frontier Questions and Reading Ladder¶

Core framing¶

• Autor et al., 2020

• Karabarbounis, 2023

• Acemoglu & Restrepo, 2019

Structural change and job quality¶

• McMillan & Rodrik, 2011

• Herrendorf et al., 2013

• Ding et al., 2022

• Helm et al., 2023

• Rodrik et al., 2024

Global / development / frontier¶

• Cirera et al., 2024

• Stiglitz et al., 2021

• Liu & others, 2025

• International Energy Agency, 2026

• Prytkova & Petit, 2025

Frontier questions¶

1. When do successive technology waves cancel earlier labor-displacing effects, and when do they reinforce them?

2. Does AI deepen superstar-firm dynamics or create a distinct infrastructure bottleneck through energy and compute?

3. When workers leave manufacturing, what predicts whether they move into good service jobs versus low-quality services?

4. How should labor economists think about outsourced digital work and data-labor supply in the Global South?

5. Which institutions prevent productivity gains from translating into worker welfare gains?

Core market-level framing. Start with Autor, Dorn, Katz, Patterson, and Van Reenen on superstar firms and labor share, then use Karabarbounis for broader labor-share interpretation Autor et al., 2020Karabarbounis, 2023.

Task inequality and rent sharing. Use Autor, Levy, and Murnane and Goos, Manning, and Salomons for task and polarization mechanisms Autor et al., 2003Goos et al., 2014. Use Kline, Petkova, Williams, and Zidar and Card, Cardoso, Heining, and Kline to connect innovation rents and firm wage premia to worker outcomes Kline et al., 2019Card et al., 2018.

Labor-market power and institutions. Read Azar, Marinescu, and Steinbaum and Benmelech, Bergman, and Kim as labor-market concentration anchors Azar et al., 2022Benmelech et al., 2022. Then ask how unions, bargaining coverage, platform governance, wage floors, and transparency rules mediate technology incidence.

Historical comparison. Use Autor and Acemoglu-Restrepo to connect current AI to automation, new tasks, and reinstatement Autor, 2015Acemoglu & Restrepo, 2019. Use mechanization and electrification evidence to see why complementary investment and organizational change shape persistence Atack et al., 2020Fiszbein et al., 2020.

Structural change and job quality. Read McMillan and Rodrik, Herrendorf, Rogerson, and Valentinyi, Ding et al., Helm, Kugler, and Schonberg, and Rodrik, Sandhu, and Suri to separate employment reallocation, productivity-enhancing reallocation, and labor-absorbing services McMillan & Rodrik, 2011Herrendorf et al., 2013Ding et al., 2022Helm et al., 2023Rodrik et al., 2024.

Global and frontier AI. Use Cirera et al. for technology sophistication and diffusion, Stiglitz and Rodrik for AI and globalization, the World Bank source for early generative-AI labor demand measurement, and IEA’s energy-AI work as a frontier infrastructure source Cirera et al., 2024Stiglitz et al., 2021Liu & others, 2025International Energy Agency, 2026.

Exercises And Discussion Prompts¶

1. Under what conditions can technology raise productivity while reducing labor share?

2. How would you distinguish a superstar-firm mechanism from a generic market-power mechanism?

3. Give an example where patent or intangible rents are partly shared with workers, and another where they mostly are not.

4. Why is structural change not automatically welfare-improving for workers?

5. How can two technology waves reinforce each other instead of offsetting each other?

6. Design a study that distinguishes employment reallocation from job-quality reallocation after a technology shock.

7. How can unions, platform rules, or wage floors change the incidence of the same technology?

8. What is the labor-economics case for studying AI-related electricity and data-center demand?

9. How could outsourcing and offshoring make AI incidence differ between high-income and developing-country labor markets?

Reproducibility And Code Lab Note¶

The Week 5 code lab lives at labs/05-technology-inequality-market-structure-and-labor-market-institutions/. It is a bounded synthetic teaching path that implements the Week 5 Reproduce -> Diagnose -> Transfer workflow. The smoke test constructs a firm-industry concentration and labor-share panel, writes a decomposition inspired by the superstar-firm design, diagnoses market-level incidence and institutional mediation, and transfers the logic to structural change, Global South technology diffusion, outsourcing, and AI infrastructure questions. It requires no proprietary firm microdata, official replication files, establishment surveys, external downloads, or confidential linked firm-worker data.

Slide Companion Note¶

The Week 5 slide deck lives at slides/week5/05-technology-inequality-market-structure-and-labor-market-institutions.tex. The deck does not duplicate the chapter. It defines the market-level question, shows the incidence decomposition, covers rent capture, labor share, superstar firms, labor-market power, historical technology waves, structural change and job quality, global and Global South evidence, frontier AI infrastructure questions, methods and data, and the Reproduce -> Diagnose -> Transfer lab design.

Bridge Forward¶

Week 6 closes the course by turning the sequence into a research architecture. Week 5 supplies the market-level capstone: how technological change reshapes inequality, market structure, institutions, structural change, and worker welfare once firm and worker adjustments are aggregated. The final week asks students to convert that logic into research designs with clear exposure measures, mechanisms, counterfactuals, and welfare objects.