Enrico Moretti on Jobs, Cities, and Innovation
What this covers
Enrico Moretti of UC Berkeley discusses his research on how cities succeed or fail economically, speaking with Russ Roberts on EconTalk. The conversation centers on a simple but consequential claim: a city's prosperity depends not on charm or policy engineering, but on the education of its workforce and the strength of its traded sector—the companies competing nationally or globally rather than serving local customers. Moretti walks through three decades of American metropolitan change to show how innovation sector jobs, though only 10% of employment, generate a Local Multiplier Effect, with each high-tech position supporting roughly five additional jobs in the local service sector. He illustrates this pattern using Seattle and Albuquerque, once economically similar cities that diverged sharply after Microsoft's relocation to Seattle in 1979—an event driven by founders' personal preferences, not strategic planning.
Moretti argues that clusters emerge from Path Dependence and Agglomeration / Clustering rather than deliberate policy, and that efforts to copy successful cities often commit the error of Reverse Causation / Cause vs Effect. He disputes the "creative class" theory, offering Berlin—culturally vibrant but economically struggling—as a Counterexample Reasoning. The discussion addresses globalization and outsourcing, where Moretti takes a contrarian position: outsourcing strengthens demand for high-skill American workers. The conversation also covers less-educated workers' declining fortunes, whether due to technological change and Creative Destruction or measurement error in the CPI, with Roberts raising technical objections about Heterogeneous Inflation by Income and Bureau of Labor Statistics methodology. Moretti concludes by advocating for education and research investment as the surest path to shared prosperity.
Moretti argues that a city's economic fate is driven by its traded/innovation sector and the education of its workforce: attracting high-tech jobs generates a large local-service multiplier, while innovation clusters arise from serendipity and self-reinforcing agglomeration rather than deliberate policy.
- Innovation-sector jobs are only ~10% of employment but each one supports about 5 additional local-service jobs
- Clusters like Seattle emerge from idiosyncratic events (Microsoft's 1979 relocation) and feed on themselves, not from engineered policy
- The wages and prosperity of low-skill workers depend on the productivity of the local traded sector and on labor mobility across sectors and cities
High-tech job creation in traded sectors drives disproportionate local prosperity through wage spillovers to service workers.
- Each high-tech job created in a community supports about 5 additional jobs in that community's local service sector—both college-requiring (doctors, lawyers, teachers, nurses) and non-college (taxi drivers, bartenders, waiters, childcare workers); for example, Apple's ~13,000 Cupertino employees support nearly 70,000 additional metro-area jobs outside Apple and outside high tech.
“for each high tech job that you create in a community, 5 additional jobs are supported in the same community outside high tech in the local service sectors”
- Although the innovation sector is only about 10% of the labor market and roughly 70% of jobs are in the local service sector, it is the vitality, productivity, and innovation of the traded sector (high tech, manufacturing, agriculture) competing on national/international markets—not the local service sector—that is the key to a local community's prosperity.
“it's really the vitality, the productivity and the degree of innovation in the traded sector that is the key to the prosperity of the local community”
- Local service occupations (barbers, waiters, taxi drivers, teachers) experience little productivity growth because they require roughly the same labor per unit of service as decades ago, whereas the traded sector enjoys large productivity gains from new technology; because workers can move across sectors, the wage gains from traded-sector productivity raise wages economy-wide, not just within that sector.
“the local service sector because of the technology they are using to produce services, tends not to grow its productivity very fast. By contrast, the traded sector... is experiencing a big increase in productivity”
- Contrary to common belief, globalization and outsourcing increase demand for high-skilled US workers: outsourcing makes high-tech firms more competitive so they hire more domestically, and a larger global market disproportionately benefits innovation-sector firms because their costs are mostly fixed R&D with trivial variable costs, so expanding the market raises profits without raising costs.
“Globalization and outsourcing are actually good for labor demand, for the demand for high-skilled workers in our country.”
Educational attainment of the workforce predicts metro economic success far better than lifestyle amenities or policy engineering.
- The 'creative class' theory that amenities like farmers' markets and cultural vibrancy revitalize city economies lacks solid evidence of job effects; Berlin, a magnet for creative people and one of the world's coolest cities, has had one of Germany's highest unemployment rates and lowest per-capita incomes, surviving on its capital status and tourism rather than private-sector innovation—so making St. Louis or Detroit 'cooler' is unlikely to work.
“there is no solid evidence that suggests that those things have any effect on jobs”
- Attempts to imitate successful cities often confuse cause with effect: the University of Washington thrives now because of wealth created by Seattle's high-tech success (Gates and Allen donations), but in 1979 it was not better than Washington University in St. Louis, so the research university did not cause the cluster; ex-post rationalization of current advantages is pervasive and dangerous because it leads communities to waste money on policies unlikely to change their economies.
“That line of reasoning confuses cause with effect, effectively.”
- The average level of education in the workforce, especially the share of college-educated workers, has been the key predictor of communities' economic success over the past three decades; cities with many educated workers attract good employers and develop strong economies, and this is not a blue-state/red-state phenomenon.
“for the past three decades the average level of education in the workforce has been the key predictor of the economic success of communities”
- No major American innovation hub was directly and explicitly engineered by government policy; hubs develop from idiosyncratic factors—a local firm succeeds and attracts similar firms, creating a self-reinforcing cluster—so innovation clusters are an emergent phenomenon that cannot be easily steered or directed.
“If you look at the history of America's great innovation hubs, they haven't found one that was directly, explicitly engineered by an explicit policy on the part of the government.”
Structural forces—globalization, automation, and technology—have permanently reduced median worker wages and blue-collar employment despite manufacturing productivity gains.
- Over the past three decades, the acceleration of globalization and a variety of technological innovations have affected different workers' productivity differently, strengthening some metropolitan labor markets while weakening others, resulting in a redistribution of wealth, jobs, and population across American metro areas.
“These two changes are global in scope but they have profoundly different effects on the local economy of different metropolitan areas of the United States.”
- For the first time in American history, the median worker—someone with a high school degree—has seen wages decline relative to the previous generation; the average wage for a male full-time high-school-educated worker is about 20% lower than in 1980, reflecting globalization and technological progress reducing demand for his skills.
“The average wage for a male, full time worker with a high school education today is about 20% lower relative to 1980.”
- American manufacturing is highly productive—producing roughly twice as much value as 30 years ago—but employs fewer and fewer workers because machines have substituted for labor; for example, a GM worker who made about 7 cars a year in the 1950s now makes about 28, meaning 70% fewer workers are needed for the same output.
“today we are producing, in terms of value, twice as much as we were producing 30 years ago in the American manufacturing sector. So, American manufacturers are doing fine. But they are using fewer and fewer workers.”
- Subsidies for manufacturing and protectionist limits on imports are counterproductive and unlikely to slow the decline in blue-collar jobs, because that decline is driven by structural forces; even the surviving manufacturing jobs are increasingly engineering jobs rather than standard blue-collar work—the US lost an average of 360,000 manufacturing blue-collar jobs per year for 30 years while the number of engineers in manufacturing doubled.
“I am skeptical of these policies. I think they would be counterproductive. And most importantly, I doubt that they would slow the decline in blue collar jobs.”
Communities that fail to reinvent their economic base into new technologies lose competitive advantage and decline over time.
- No technology stays on the frontier forever, so a community's survival depends on reinventing itself into new technologies; Detroit was the Silicon Valley of its era but became complacent and failed to leverage its ecosystem into something new when auto demand collapsed, whereas Silicon Valley repeatedly reinvented itself from hardware to software to Internet to biotech/nanotech.
“The technological frontier keeps evolving, and there's no technology that stays on the frontier forever. The secret for a community is to reinvent itself into new things.”
- Seattle's transformation from a declining 1970s lumber/manufacturing economy into a vibrant high-tech hub traces to Microsoft's 1979 relocation from Albuquerque to Seattle—a 15-employee company moving because founders Gates and Allen wanted to be near family—which seeded the cluster; afterward Seattle's wages and college-graduate share diverged sharply upward from Albuquerque's, which had been comparable in 1979.
“in 1979, Bill Gates and Paul Allen, the founders, decided that they wanted to be close to their families--who were in Seattle. So they relocated the company back to Seattle.”
- When Jeff Bezos founded Amazon in 1995, he placed it in Seattle—despite being born in Albuquerque and having no personal tie to Seattle—because by then, 15 years after Microsoft's move, Seattle had become a magnet for innovative companies, demonstrating that an early success indirectly creates an ecosystem that draws further firms.
“When he decided to found Amazon, he places it in Seattle because by 1995, so 15 years after Bill Gates' regional move, Seattle has become a magnet for this type of innovative companies.”
- The modern American manufacturing model separates innovation from physical production: Apple's iPhone/iPad were designed, engineered, and marketed by highly skilled American workers in Cupertino, while components and assembly happened in Asia; computer and semiconductor manufacturing employment peaked 25 and 12 years ago respectively and declined since, even as highly skilled employment in those industries rose.
“that iPhone or iPad was designed, engineered, marketed from Cupertino, here in the Valley, by highly skilled American workers. But its physical components... were by and large not made in America.”
Low-skill workers benefit less from productivity gains and face higher barriers to relocating to prosperous cities.
- Sectors heavily exposed to outsourcing nonetheless created many US jobs: employment in software grew much faster than the rest of the labor market over the past 20 years, and Internet employment grew about 200 times faster than employment in the rest of the labor force.
“If you look at employment in the Internet, it has increased 200 times faster than employment in the rest of the labor force.”
- There is effectively no single inflation rate; low-income households have experienced lower inflation over the past 20 years than high-income households because they spend more on cheap globally-traded consumer goods (e.g., from China/Vietnam, bought at Wal-Mart), while high-income households spend more on personal services that are less exposed to global competition and rise faster in price.
“Households with low income, they tend to experience less inflation over the past 20 years than households with high income.”
- The wage spillover from rising traded-sector productivity to non-traded workers is larger the closer the skills used in the traded sector are to those used in the non-traded sector; a low-skill worker who cannot perform traded-sector jobs benefits less from traded-sector productivity gains.
“The effect is larger, the closer the skills that are used in the traded sector to the skills that are used in the non-traded sector.”
- Although the economic returns to geographic mobility are high given large and persistent wage differences across cities, workers differ greatly in their ability to relocate; less-mobile workers, who struggle to leave declining clusters for thriving cities, increasingly pay for this in lost jobs, careers, and incomes.
“there are vast differences in the ability of relocating and moving to different cities among American workers... this is costing the less mobile increasingly in terms of jobs, careers, and incomes”