This Bonus Session expands on our in-class GMI discussion on NAFTA and explores the long-term decline in U.S. manufacturing jobs, drawing from research by Acemoglu and Restrepo's 2022 paper, Tasks, Automation, and the Rise in U.S. Wage Inequality. While trade and globalization played a role, the research shows that automation is the primary force behind both the drop in manufacturing employment and the rise in wage inequality since 1980. Its effects have been broad, sustained, and nationwide, particularly displacing routine-intensive blue-collar jobs, even as total manufacturing output continued to grow.

1. Manufacturing Employment Decline

• In 1950, ~1 in 3 U.S. workers had a manufacturing job.
• By 1980: ~20%; by 2016: under 10%; today: closer to 8%.
• The decline occurred despite steady or rising output—indicating productivity gains through automation, not offshoring, were the main factor.

2. Automation’s Impact on Jobs and Wages

• Automation accounts for an estimated 50–70% of changes in the U.S. wage structure from 1980–2016.
• Workers without a high school diploma saw wages fall ~8.8%, with industrial robots alone displacing between 400,000–700,000 jobs from 1990–2015.
• Job loss was concentrated among those performing routine, repetitive tasks, traditionally the foundation of middle-class, blue-collar employment.

3. Trade vs. Automation

• NAFTA and China’s WTO accession caused localized disruptions, but the national employment impact was smaller.
• NAFTA job losses estimated at 200k–800k.
• Increased import competition from China linked to ~2 million lost jobs, but regionally concentrated.
• Automation alone accounts for 3.3–5 million lost jobs, across the entire country.

4. Echoes in the Age of AI
The same task-displacing patterns seen in manufacturing may now emerge in white-collar sectors:

• Generative AI poses risks to routine cognitive tasks (e.g., summarizing reports, drafting standard communications).
• Like factory automation, GenAI may reshape job roles rather than eliminate whole professions.
• The potential outcome: wage polarization, where high-skill workers benefit while middle-tier roles are eroded.

5. Key Differences with GenAI
Despite parallels, AI adoption is moving faster and may be more flexible:

• Potential for task complementarity rather than substitution.
• White-collar workers may adapt more quickly due to higher education levels and geographic mobility.
• Still, the risk of “technological hollowing”—a shrinking middle class—remains.

6. Institutions Matter
Following the arguments in Acemoglu and Robinson's book, Why Nations Fail (which we will discuss in Module 5), a shrinking middle class has repercussions for the long-term economic growth rate in the US, as innovation requires wide-scale participation in economic activities. If "inclusive" institutions do not exist that both allow (via access to education, capital markets, etc.) and encourage (via ensuring new businesses can compete, supporting both small- and large- scale innovation, etc.) widespread economic innovation, long-term growth will decline.

While technological hollowing is possible in white collar sectors with the growth of genAI, it is not inevitable. Recognizing that there are always tradeoffs, there are historical policy precedents in the US that could guide this technological change towards more broad-based growth than what was experienced as automation transformed the US manufacturing sector:

• Workforce investment (e.g., GI Bill)
• Wage subsidies and expanded Earned Income Tax Credit
• Public R&D support (e.g., DARPA)
• Place-based development (e.g., Empowerment Zones)
• Balanced labor standards (e.g., Fair Labor Standards Act)