Dan Breznitz is right to question what struggling regions stand to gain from the Silicon Valley model of regional development. Estimates from the Brookings Institution indicate that just 5 metro areas accounted for 9 out of 10 new high-tech jobs between 2005 and 2017. Efforts to replicate this model elsewhere have typically failed because localities chase the outputs of an innovative region—large tech firms and high-growth startups—rather than cultivating the inputs: highly skilled individuals and technical infrastructure. As a result, tech companies and the people they aim to recruit mostly continue to flock to the places where the inputs for innovation already exist, and struggling regions remain stuck. For an alternative, Breznitz cites thriving places from China to Italy where firms innovate in more incremental ways than the likes of venture-backed unicorns.

What often distinguishes the successful regions are the actors driving local economic development. An ecosystem’s activities—and its effectiveness—depend on who leads, and how.

What can places like Cobalt and Cleveland learn from these examples? Breznitz paints in broad strokes, pointing to the value of fostering an “innovation ecosystem.” Exactly what this means is unclear, but presumably it includes a constellation of organizations operating outside the free market dedicated to promoting economic development: government agencies, industry associations and private foundations, and universities and community colleges.

These organizations are certainly important. What this prescription misses, however, is that many regions in the United States already have the trappings of an innovation ecosystem. The past several decades have seen a proliferation of nonprofit institutions dedicated to innovation and economic development. In St. Louis, there is the Cortex Innovation Community; in Kansas City, it’s InnovateKC; in Milwaukee, there’s MKE Tech. While eager to support innovation, these organizations do not always succeed: if you build it, innovators will not necessarily come. What often distinguishes the successful regions are the actors driving local economic development. An ecosystem’s activities—and its effectiveness—depend on who leads, and how.

In recent work at MIT’s Initiative for Knowledge and Innovation in Manufacturing, Dan Traficonte and I study innovation ecosystems in U.S. metro areas with a legacy of manufacturing, examining how some regions have remained strong even as many others have declined under pressure from globalization and technological change.

The ecosystem organizations in these regions fall into two categories. The first includes organizations focused on serving particular firms, including economic development organizations focused on attracting new businesses with tax incentives, industry associations that derive their revenue from membership fees, and organizations participating in the U.S. government’s Manufacturing Extension Partnership (MEP) that perform consulting projects for small and medium enterprise manufacturers. The second category features organizations that focus instead on serving the broader community—from government agencies that aim to advance regional economic development to universities and community colleges dedicated to basic research and education and private foundations with a community mission.

The challenge for many regions is that firm-focused organizations drive the economic development agenda. When these groups dominate, there are significant risks.

One risk is that the region’s economic fate becomes too closely tied to particular firms and industries. In Rochester, New York, for example, three employers—Kodak, Xerox, and Bausch and Lomb—dominated the local economy for decades. Estimates suggest that at one point more than half of the local workforce was employed at one of these companies. The dominance of these companies in turn shaped the type of ecosystem organizations that emerged in Rochester. A state-sponsored Center for Advanced Technology focused on optics, a joint area of interest for the region’s main employers. The companies also funded research and training in optics at local universities and community colleges. Kodak even gave the University of Rochester company stock, which made its endowment one of the largest in the country.

The clustering of regional economic activity around a few firms in focused industries appeared successful—until the region’s leading firms fell on hard times. In the 2000s, Kodak and Xerox stock suffered, and the university’s endowment shrunk. Both companies reduced their local workforces dramatically, and Kodak filed for bankruptcy in 2012. Jobs in Rochester stagnated. Although the innovation ecosystem was not responsible for Kodak’s dominance, the ecosystem organizations that could have served the broader public interest did not invest in diversifying the local economy.

A second risk of leaning too heavily on firm-focused organizations is that regions miss opportunities to support innovation in unexpected domains. Focusing investment efforts on “target industries” can blind leaders to long-term opportunities. For some policymakers, this might be an acceptable tradeoff; the benefits of focusing on some industry domains might outweigh the costs. But ecosystem organizations are not always good at predicting the highest-potential areas for innovation. Often target industries are simply sectors where the region has excelled in the past and seem to have potential to grow in the future. If current innovative regions had adopted this approach, the firms and industries that now drive their ecosystem might never have emerged.

Regions that lean too heavily on firm-focused organizations risk missing opportunities to support innovation in unexpected domains.

In Raleigh-Durham, North Carolina, for example, the original push to develop Research Triangle Park was industry agnostic. The legacy of the region was agricultural; the universities, government agencies, and real estate interests that emerged to support a new space for R&D could not have imagined that their park would be home to biotechnology innovation. Likewise, in Albany, New York, organizations could not have predicted that the region would become the hub for nanotechnology innovation it is today. Indeed, the campus that is now home to the state university’s nanotech research was originally focused on atmospheric science research. It turned out the university was flexible enough to identify and empower a pioneering lab in nanotechnology, which became the seed for billions of dollars in investment.

One might infer from these examples that community- rather than firm-focused ecosystems are key. But a university-dominated ecosystem without input from industry can also face big risks, such as providing skills and generating research detached from the demands of the local economy. Instead, Traficonte and I have found that in high-growth manufacturing regions, effective ecosystem institutions find ways to marry the firm-focused interests of industry groups with the community mission of some universities and community colleges.

Take Ames, Iowa, where the Center for Industrial Research and Service is part of both Iowa State University and the national MEP network. The organization draws on the research and technological expertise of the university to help regional employers identify opportunities for technology upgrading and growth. The center’s emphasis on new technologies differentiates it from other MEPs, which are separate from universities. Or take Tulsa, Oklahoma, where Tulsa Tech enrolls thousands of students per year in its company training programs. Trainees are matched with a company before Tulsa Tech provides tailored training that prepares them to enter a full-time job with the right skills. Although companies in the Tulsa region turn to Tulsa Tech to hire skilled workers, the organization is not exclusively firm-focused. It is part of Oklahoma’s state-funded Career Tech system, which funds Tulsa Tech and other centers through property taxes. As a result, Tulsa Tech has a public mandate to focus on finding career opportunities for its students, as well as supporting the growth of local employers to expand the property tax base.

What do places like Ames and Tulsa offer Cobalt and Cleveland? They suggest a road to organizing and balancing a region’s investments in innovation and economic development. On the one hand, investments too focused on particular industries and capabilities risk crowding out potential growth areas; on the other hand, supporting public goods alone can leave regions detached from market demands. The work of innovation policy—and regional economic development more broadly—should be to balance these goals.