Mariana Mazzucato and colleagues have laid out a clear vision for a new role for government in the economy, and I agree with their principal argument. I see support for it in the literature on innovation systems, in the empirical data on successful past innovations, and in the work of my graduate students wrestling with how to perform policy analysis. I would like to expand on the role innovation, in particular, should play in a mission-oriented response to deep social problems. Three essential themes stand out in the developing body of policy advice for stimulating more innovation, faster.

Innovation has a role to play in a mission-oriented response to deep social problems.

First, expectations are central to stimulating innovation. A mission orientation can help to reduce risk and stimulate investment by aligning expectations about future conditions. This “guidance of the search” is one of the basic functions in the innovation systems literature. Moore’s Law, the doubling of computer processing power every two years, has provided exactly this sort of guidance for decades. It makes the future more predictable and has thus enabled significant and stable investment and innovation in computational technologies—not only in computer hardware, but also in software, as well as devices that rely on both. Such guidance can also be provided by the state, as Japan’s Ministry of International Trade and Industry (MITI) did so successfully for decades and Germany’s Energiewende is currently doing for climate change.

Second, policy experimentation is essential. A “lead-and-learn approach,” as Mazzucato and colleagues put it, is exactly right. Setting a mission requires bold leadership, but following through requires learning and iterative experimentation. New types of policies need to be designed, evaluated, and improved—continuously. Social priorities will evolve, and new trade-offs will emerge. As a result, policies will need to adapt to continuously shifting priorities among multiple social goals.

Experimentation also has a role to play in policy diffusion. Much as technologies go through a sometimes-lengthy process of adoption, so too new types of policies need to be implemented in one jurisdiction, observed, evaluated, and then adopted by others. In renewable energy we have seen this phenomenon play out as U.S. states sequentially adopt renewable portfolio standards; the same is visible in countries adopting feed-in tariffs and net-zero climate goals. Louis Brandeis’s laboratories of democracy are very much in play for twenty-first century industrial policy. In the same way that technologies can be de-risked through experimentation by early adopters, so too can policies. Barriers to greater policy experimentation arise not only from elusive creativity in policy design, but also from bureaucratic incentives to evaluate the benefits of new policies over easily quantifiable metrics, in the near term, and often for a narrow set of objectives. Performance management can make policies more efficient, but adaptive management requires space for learning from novel policies, some of which will have unintended adverse consequences. Careful policy evaluation and adjustment is essential, and we need much more of it.

Third, the empirical support for how successful innovations have emerged in the past makes clear that a comprehensive and systemic approach to supporting mission-oriented innovation will be most likely to succeed. This is now so clear in part due to Mazzucato’s earlier book, The Entrepreneurial State (2011). Touch screens, packet switching, and microprocessors all enjoyed direct government support, which extended well beyond traditional research and development (R&D), but crucially included support for demonstrations, early commercialization, and public procurement. We have seen the same in energy with the development of wind power and electric vehicles.

Mission-oriented industrial policy can easily fall into the trap of focusing on national rather than global priorities, but innovation nationalism will slow innovation.

I have seen evidence of successful industrial policy most clearly in my own research on the development of solar energy and how it became so inexpensive. A first observation is that no single country persisted in developing solar photovoltaics. Each leader relinquished its lead within a few years. The United States, Japan, Germany, Australia, and China each made a distinct contribution, reflecting the unique characteristics of their national innovation systems—their education infrastructure, industrial base, and culture. The United States funded R&D and directly purchased solar panels, stimulating an early industry. California experimented with policy through a series of fixed price standard contracts. Meanwhile, Japan’s MITI coordinated multinational corporations like Sharp and Panasonic to share R&D, leading them to include solar in niche consumer applications, such as watches and calculators. MITI also launched the world’s first residential solar subsidy program. Germany, through a process of policy diffusion, combined Japan’s subsidy program and U.S. fixed price contracts and made them bigger and longer term to create a massive demand-pull signal to the world. This scale led equipment suppliers to build machines specifically for making solar panels. China met German demand through scrappy entrepreneurs, most of whom trained in Australia. Later the central government supported these firms through tens of billions in loans, but only after the industry legitimized itself by selling panels to Germany and going public on Wall Street. Ultimately, solar became cheap thanks to many forms of knowledge diffusion around the world: sometimes as ideas in scientific publications, sometimes as tacit knowledge in the heads of internationally mobile individuals, and crucially as knowledge embodied in production equipment that could be used anywhere.

For all these reasons, the success of a mission-oriented approach will hinge on at least four key principles.


1. Be wary of innovation nationalism

Mission-oriented industrial policy can easily fall into the trap of focusing on national rather than global priorities, but innovation nationalism will slow innovation. There are reasons for supporting national manufacturing, making supply chains more local, and perhaps improving resilience to disruptions. But these types of onshoring efforts need to be made with a clear-eyed view that innovation nationalism will slow down innovation. In areas like climate, we need the opposite; we need acceleration and innovation nationalism gets in the way.


2. Implement a broad set of policies

Modern industrial policy involves a much more comprehensive approach to using the capabilities of governments than the industrial policy of the past. Instead of a low-cost government loan to a single firm, we should expect a mix of approaches: competitive solicitation for funding demonstration projects, government procurement of novel technologies, and preferential treatment in markets for technologies with promising long-term attributes. This approach will involve many policies, some of which will overlap, interact, and even be redundant and partially misaligned. Some of this redundancy can be worked out with iterative experimentation. But much of it is helpful in making investor expectations of the commitment to the mission robust—that is, resilient to elections, technological setbacks, and other unanticipated events.


3. Embrace multiple goals

A mission will need to address multiple goals—for example, not just economic growth, but also distributional impacts, inclusion, employment, political durability, and support for communities in transition. These goals also must move well beyond simply remedying isolated sources of market failure. As Mazzucato and colleagues make clear, market failure is just a point of departure; innovative, comprehensive solutions must be far more ambitious and move beyond incrementalism.


4. Anticipate risks

It is essential to get serious upfront about managing the risks that will emerge from more comprehensive state intervention. These risks are real: they include rent seeking, regulatory capture, picking winners, information access, crowding out, and bureaucratic risk aversion. But if these issues are clearly and forthrightly anticipated, they can be managed. A general remedy is boosting the talent in government, equipping government agencies with sufficient quality data to make decisions, and empowering them to take diversifiable risks.

A mission-oriented economy is an exciting concept because it has the potential to allow us to make progress on a set of stalled social problems. To make good on that potential, we must adopt a different policy mindset—one that places innovation at the center and embraces iterative policy experimentation as the means to implement the mission.