Whether it is fostering the green transition, rebuilding the middle class through good jobs, or reducing poverty in the developing world, engineering structural change is key. Meeting each of these objectives requires moving the economy’s resources – innovation, organizational capacity, entrepreneurship, capital and workers – to activities that are more productive and achieve social, environmental and developmental goals in the process. The strategy that connects all three domains is productivism, a paradigm that I describe here in detail.

Market fundamentalists would say structural transformation is a task better left to the operation of markets. Of course, they would readily acknowledge that some tweaking of market forces may be required. But the task of government policy, in their view, should be limited primarily to letting the markets do their job of allocating resources to their best – meaning most profitable – uses.

We have seen time and again that success requires something different: more government intervention than what market fundamentalists would want, but also better government intervention. The task in this essay is to make the case for governments’ role in structural change. It is also to distill what we learn from experience about how governments can be more effective in this role. Don’t let the label productivism turn you off. Call it sensible, pragmatic policymaking and you’ll have it exactly right.

The Visible Hand

In his 1980 TV series Free to Choose, Milton Friedman held up a pencil to illustrate the power of markets. It took thousands of people all over the world to make this pencil, Friedman said – to mine the graphite, cut the wood, assemble the components and distribute the final product all around the globe. No single central authority coordinated their actions; that feat was accomplished by the magic of free markets and the price system. It was Adam Smith’s famous invisible hand at work.

Forty-five years later, the pencil story serves a very different narrative – one that gives government policy a much more prominent place. Today China is the world’s leading producer of pencils. Yet China was hardly a natural destination for the industry. There are better sources of graphite in Mexico and South Korea. Forest reserves are more plentiful in Indonesia and Brazil. Germany and the United States had better technology when China’s industry got off the ground. China had lots of low-cost labor, but so did Bangladesh, Ethiopia and many other developing countries. Much of the credit belongs to the initiative and hard work of Chinese entrepreneurs and workers. But leaving out the Chinese government’s contribution would be like staging Hamlet without the prince of Denmark.

The initial investments in technology and labor training were made by China’s state-owned firms. The government then stimulated the industry by keeping wood artificially cheap, providing generous export subsidies and intervening in currency markets to enhance Chinese producers’ competitiveness on world markets. As in so many other branches of manufacturing, China’s government subsidized, protected and goaded its firms to ensure rapid industrialization.

Or consider orchids in Taiwan. The industry took off four decades ago thanks to concerted efforts by the Taiwanese government to diversify away from sugar. Sugar had traditionally held an important position in Taiwan’s countryside, both as an export commodity and as an employer for farmers. But it had fallen on hard times due to declining prices on world markets.

In many countries, the outcome might have been deteriorating incomes, rising indebtedness for farmers and a depressed rural sector. The Taiwanese government chose instead to mount a comprehensive investment drive to develop a world-class orchid industry. It paid for a genetics laboratory, quarantine site, shipping and packing areas, new roads, water and electrical hookups for privately owned greenhouses, and an exposition hall. It provided low-interest loans to help farmers build the greenhouses.

Supported by government extension services, large numbers of orchid growers, from micro enterprises to medium-sized ones, became part of the orchid cluster and supply chain. Today, Taiwan is the world’s third biggest exporter of orchids behind the Netherlands and Thailand.

Maybe it is just East Asian nations that are able to pull off these feats? Not really. Consider a case from Latin America. Fundación Chile is a nonprofit set up in 1975 that acts as a public venture capital fund. It served as an incubator for new technologies, adapting them to the Chilean context and then selling off the successful ones to the private sector.

In 1981, Fundación Chile acquired a small, local aquaculture company. Using Norwegian and Japanese salmon farming technology and through a process of learning by doing, it developed an entire supply chain from specialized feed to export logistics. The knowledge it acquired was disseminated freely to private firms, producing an explosion of salmon farming. Exports went from 300 tons to 24,000 tons per year by the 1990s, making Chile the second largest exporter of salmon after Norway.

The reality is that virtually all instances of productive transformation since the Industrial Revolution have been the result of combined public-private initiatives. This is as true for countries that are normally associated with free market ideology as for others. Chile has long been lauded as one of Latin America’s most successful economies, and as one of its most market-oriented. But the state has played a role in all its major exports.

The country’s largest copper company is state-owned; the forestry sector benefited from generous subsidies, including under the free-market-radical President Pinochet. The wine industry was promoted through supplier development and export credit programs funded by government agencies. Scratch any modern export success story, and more likely than not, you will find the hand of government hiding beneath.

The U.S. government has always played a significant role in R&D. During the second half of the 19th century, land grant colleges and agricultural extension services disseminated know-how and helped create the most productive agriculture in the world. U.S. manufacturing grew, caught up and eventually surpassed Britain behind high tariff walls. In the postwar period, government funding by the Small Business Investment Company played a significant role in launching Silicon Valley and laid the groundwork for the subsequent development of the private venture capital industry.

The U.S. Department of Defense, through its procurement and R&D programs, enabled all the critical innovations that would eventually constitute the digital revolution. Its Defense Advanced Research Projects Agency is responsible for the internet, GPS, flat-panel displays and the computer mouse, among other innovations.

“Stop,” I hear you say. “We get the message. Government intervention works!” If that’s what you are thinking, we are halfway – but only halfway – there. My point is more nuanced. Government policy does work, but not always. It sometimes fails massively. And if we want to apply similar policies to the new domains of services and green industries, we’d better think hard about both the successes and failures, and learn how to improve their practice.

What Solyndra Teaches Us

Here is a cautionary tale. Solyndra was a solar cell company founded in 2005 and one of the first to get funding under an expanded government loan-guarantee program. Then-president Obama was keen to develop green technologies. The government provided Solyndra with $535 million in loan guarantees to supplement $450 million raised from private investors.

For Obama, Solyndra was much more than a startup experimenting with a new technology. It was a company that exemplified the economic transformation he wanted to achieve. Obama personally extolled the company at a visit to its facility in Fremont, California, in May 2010. “Companies like Solyndra are leading the way toward a brighter and more prosperous future,” he declared.

But government also had a big role to play. It had to “create the conditions in which students can gain an education so they can work at Solyndra, and entrepreneurs can get financing so they can start a company, and new industries can take hold.”

Not unlike the Chinese government, the Obama administration hoped to accomplish multiple goals with the program. Stimulating demand and employment, spearheading new technologies, competing with China and benefiting the environment were all cited in selling the program to congressional interests and the broader public. “If we want to compete with other countries that are heavily subsidizing the industries of the future,” said President Obama, “we’ve got to make sure that our guys here in the United States of America at least have a shot.”

By August 2011, Solyndra had gone bankrupt. The company had made a gamble that did not pay off: the viability of its business plan depended on silicon prices remaining high. Its technology for producing photovoltaic cells relied on CIGS (copper indium gallium selenide) as the semiconducting material instead of silicon, which was vastly more common in the industry.

CIGS was cheaper than silicon but less efficient at converting solar energy. At the time this seemed a reasonable gamble, as silicon prices had been rising. However, after 2008, silicon prices tumbled precipitously, thanks to new capacity coming online in China. The company failed even though it had met its own technological and cost-reduction goals.

Its bankruptcy became a major source of embarrassment for the Obama administration. Solyndra’s offices were searched by FBI agents, and the company’s top executives were hauled before Congress (where they invoked the Fifth Amendment). This is what you get when you pick winners, critics scolded. The most damaging consequence may have been that it made it virtually impossible for the U.S. to expand the initiative and truly match China’s ambition in renewables – at least until the IRA [the Inflation Reduction Act of 2022 that committed heavy subsidies to the energy transition] more than a decade later.

The simplistic version of what went wrong in this case is that the government backed the wrong company (and the wrong technology). But this is the wrong lesson. It is the very nature of innovation that R&D and market outcomes are inherently uncertain. When venture capitalists invest in a variety of firms and technologies, they do not expect all their investments to succeed. All they hope is that enough of them succeed to pay for the ones that fail. In fact, the calculus of profits under uncertainty ensures that, under an optimal investment strategy, some of the projects will necessarily be failures. As Thomas Watson, the founder of IBM, is said to have advised his managers, “If you want to succeed, raise your error rate.”

It is no different when new technologies are supported by the government. The failure rate at DARPA, probably the world’s most successful innovation agency, is as high as 85-90 percent. At Fundación Chile, the four most successful investments have more than paid for all the flops.

Similarly, the Department of Energy, which issued Solyndra’s loan guarantee, had backed a variety of green technology projects. The true test of the government’s success is whether the social return to the overall portfolio is high enough – higher than the government’s cost of borrowing.

I am not sure whether the DOE ever undertook such a calculation, but we do know that some of the private investments it backed were very successful. In fact, around the same time that Solyndra received its loan guarantee, the DOE also issued a $465 million loan to Tesla to build an all-electric plug-in vehicle. The financial crisis of 2008-9 had left Tesla in dire financial straits, and the loan was critical to the company’s survival. It certainly was a risky investment. But the rest is history, as they say.

Tesla would have another brush with bankruptcy in later years, but the company grew to be not only the world’s premier EV manufacturer but also its most valuable auto company. We can thank the same government agency that financed Solyndra for enabling this outcome.

If there are lessons from Solyndra, they are about government failures of a different kind. First and foremost, the government was never upfront about the experimental and risky nature of the technologies it supported. There was no public messaging about the need to prepare for disappointments or to evaluate the outcomes. Worse, the Obama administration publicly showcased and invested political capital in a single firm, Solyndra, before success was assured.

The government cannot consistently pick winners, but it can stop backing losers. The worst aspects of the Solyndra debacle could have been avoided if there had been closer scrutiny of the company’s progress, or lack thereof. One of the hallmarks of successful innovation programs is that the relevant government agency sets intermediate targets and clear milestones to determine whether projects should continue to receive support or be written off.

At ARPA-E (modeled after DARPA, but for advanced energy technologies), award recipients are required to participate in periodic reviews to assess the work performed and determine whether technical milestones are being achieved. ARPA-E staff members rate progress using a traffic light system: red for projects that miss a critical milestone and are at risk of failing; yellow for projects that miss a milestone but are expected to recover; and green for projects that are on track. Red ratings lead to intensified oversight and possible termination.

The DOE loan guarantee to Solyndra was not structured in a manner that would have provided similar monitoring. The drop in silicon prices, which should have raised some red flags, was overlooked. And as Solyndra’s financial difficulties mounted, it seems that DOE officials justified the losses by arguing that this was common in all startups. The DOE never responded to repeated requests from the Office of Management and Budget to answer specific questions relating to Solyndra’s finances.

The final mistake the administration made was to let itself be wooed politically. Solyndra spent nearly $2 million on lobbying from 2008 to its bankruptcy in 2011. The principal private investor in the firm was a fundraiser for Obama, who had at least one discussion on Solyndra with White House staff in then- Vice-President Biden’s office. Regardless of whether political connections played a role in the quick approval of the loan and its aftermath, this was a bad look.

Solyndra holds important lessons on how to conduct industrial policy, especially in a democracy. First, an ability to pick winners is neither a prerequisite nor even a determinant of the success of productive transformation programs. The failure of an individual investment is not on its own a black mark against such programs. The appropriate metric is the performance of the entire portfolio of projects.

Second, it is important to cut losses when individual initiatives appear not to be working. This in turn requires clear and measurable yardsticks for progress and continuous monitoring. Having multiple goals – innovation, employment, national security – may make it politically easier to sell industrial policy, but it also makes it more difficult to discern whether the program is on track.

Finally, the practice of industrial policy must be insulated from lobbying and rent-seeking. Politics does have a role: it is inevitable, and necessary, that the overarching goals of productivist policies will be shaped by politics. But the process by which projects are selected and supported should not be subverted by politically connected firms pulling strings.

Getting Productive Transformation Policies Right

In short, some of the critics’ concerns about government involvement in structural change do carry weight. Even when they are well intentioned, governments are not omniscient, and they make mistakes. Sometimes short-term political calculations override concerns over the public interest. These considerations do not undermine the case for productivist policies, but do highlight the need to be careful when designing and implementing them. The real question is not whether these policies should be carried out but how. Basic economics and the broad experience around the world provide some helpful answers.

Let’s begin with the economics. Markets are generally very effective at directing resources to areas where their contribution to economic well-being is high. When consumers value certain things highly, their willingness to pay is reflected in markets in the form of high prices and prospective profits. This incentivizes entrepreneurs to supply the goods and services in high demand. When goods and services are no longer in high demand, their prices fall, telling investors and producers to look elsewhere. This beautiful system can work to maximize a society’s productive potential as if there were an invisible hand guiding the allocation of their labor, capital, natural resources – and ingenuity.

One important criticism of markets is that they do not ensure distributional equity, even when they allocate resources to their most productive uses. For one thing, wealthier consumers get a disproportionate say in how resources are allocated because their preferences shape market demand. One dollar, one vote! More importantly, those who have more to contribute to the economy, whether through hard work or sheer luck, skill or inheritance, get bigger paychecks. These distributional outcomes may violate our sense of social justice.

Market enthusiasts generally do not disagree that free markets can produce too much inequality. They would argue, however, that intervening in markets for goods, services, labor, or capital is never the best response. If inequality has to be tackled – a big if for libertarians – it would be better to do so by redistributing a limited amount of purchasing power so those who start with limited resources get a leg up. This could take the form of vouchers for education, for example, or a universal basic income.

The welfare state paradigm, while less enamored of markets, essentially takes this idea one big step further. It prescribes broad access to education, health care and social insurance, either through public provisioning or through an extensive system of social transfers.

What concerns us here is a more fundamental shortcoming of markets: the failure to allocate resources efficiently. When markets fail in this fashion, the structure of economic activity is distorted and does not maximize overall productive potential. This problem goes to the heart of a market system because it calls the invisible hand theorem into question.

These are the kinds of problems where productivist policies come into their own. The immediate objective is to target and correct such inefficiencies. Typically, they also serve broader goals, such as the climate, the middle class and poverty reduction. But they do so by fixing markets directly rather than redistributing resources or ensuring broad access to public services.

There are three circumstances in which markets fail to do their primary job of allocating resources well. First, many economic activities produce “externalities” – positive or negative – that markets do not price in the decisions of firms or consumers. Environmental externalities, whether local or global (as in the case of climate change), are the best known negative externalities.

On the other hand, technological innovations typically produce positive externalities. When firms learn how to produce solar cells more efficiently, for example, other firms can also reduce their costs by copying the techniques or poaching the workers and managers who are adept at using them. A third type of externality, which is less well recognized but is central to this book, is good-job externalities.

When a large employer in a small town shuts down, the economic and social costs can go significantly beyond the wage losses incurred by the workers. Similarly, creating middle-class jobs where good jobs have become very scarce creates benefits that extend beyond newly hired workers if it helps revitalize the community. In the absence of government intervention, economic activities that generate negative externalities are overproduced, and those that generate positive externalities are underproduced.

Coordination failures are the second category of market malfunction. These typically occur in the presence of significant scale economies, when getting a new economic activity or technology off the ground requires complementary investments side by side and along the supply chain. Each investment may be unprofitable on its own.

For example, there may not be high enough demand for electric vehicles in the absence of a network of fast charging stations. And producing EVs may be too costly if cheap electric batteries, a key input, are not available. At the same time, it makes little sense to invest in fast-charging stations or batteries if there isn’t a large enough fleet of EVs already being produced. Creating training facilities for specialized technical skills will not be profitable unless there are firms that will employ the graduates. Those firms, in turn, will not exist unless they already have access to trained personnel in the first place.

In such circumstances, profitable clusters of new activities may never exist in the absence of some visible hand coordinating the activities of diverse actors. It is often the government that supplies that visible hand.

Third, many industrial and service activities require particular types of public inputs specialized to the needs of certain sectors, but not so distinct that it would make sense for each firm to procure them on their own. Workforce development, infrastructure, technical knowledge, regulations and standards that are specific to a sector are examples. Government has a role in providing these kinds of inputs as well.

One benefit of articulating these rationales explicitly is that they clarify the type of government policy that is called for. In the case of externalities, taxes or subsidies that are directly targeted at the source of the externality are generally the best response. For technological or good-job externalities, this means subsidizing the types of investments that produce those externalities. Subsidies for R&D, for solar cell or advanced semiconductor facilities, and for investments by firms that will create jobs that would otherwise be unavailable are some examples. Subsidies may take different forms, such as grants, tax incentives, and cheap loans or loan guarantees.

The other two circumstances require different kinds of government policy. Coordination failures can be addressed at little fiscal cost, in principle, by bringing upstream and downstream investors, potential cluster members or the different stakeholders together around the table.

Government guarantees that do not entail budgetary outlays can serve a similar function in certain settings. For example, until the Asian financial crisis (1997-98), South Korean governments provided informal bailout guarantees to their conglomerates if they invested in priority areas. Since those investments generally proved successful, the guarantees were not called and the government did not incur any fiscal cost. The risk with such guarantees is that they may spur investments that are too risky along with those that are jointly profitable.

Customized public inputs typically do require government resources, but these must entail the provision of specific public services rather than financial incentives. If a firm is deterred from investing in a community or a developing nation because of a lack of specialized skills in the workforce or poor transport, providing those inputs is the best way to overcome the obstacle. Subsidies could serve as an inducement as well, but they may not be as effective or may miss the mark entirely.

These considerations are important because economists and policy practitioners both tend to put excessive weight on subsidies when they consider productivist policies. Their goals are often better served with other kinds of remedies. This point has been made forcefully by Tim Bartik, an economist with the W. E. Upjohn Institute for Employment Research.

Bartik has found that business services, ranging from customized training to entrepreneurship courses, generally are much more effective than subsidies at creating jobs 

in distressed communities. Yet the resources devoted to these programs are tiny, around $3 billion a year. By contrast, even before the industrial policy programs of the Biden administration, state and local governments spent around $50 billion annually on cash incentives and tax breaks for businesses. The magnitude of these subsidies has grown tremendously with the CHIPS Act [subsidies for manufacturing advanced digital processors] and IRA programs. It is a reasonable bet that a reallocation of resources from subsidies to customized public inputs would enhance prospects for local job creation.

The point is relevant to developing countries as well. Their governments often complain they do not have the fiscal resources to compete with China or advanced economies when it comes to wooing companies with subsidies. But often what’s required may be something different – better coordination of government services, say, or specific regulatory changes.

A useful illustration comes from Peru. Piero Ghezzi, the country’s minister of production during 2014-16, decided that he would run industrial policy differently. He set up a series of discussions (mesas ejecutivas) with groups of producers, with the objective of developing a common understanding on the most important bottlenecks that prevented productivity gains and the best ways to remove them.

When business leaders sit together with government ministers, the conversation typically focuses on generic complaints about taxes, red tape and lack of competitiveness. Ghezzi wanted a different conversation, focused on problems specific to each sector. He warned from the outset that subsidies were off the table.

The remedies discussed were divided into “your problems” and “my problems” – things firms could do on their own and things government should help them do better. From these conversations came a series of policy initiatives targeted at constraints identified in the process. In forestry, for example, the government amended legislation to facilitate the marketing of timber, simplified procedures for land concessions, established a new technology center to transfer innovations and facilitated the provision of long-term loans from the national development bank.

Overcoming Informational Limitations

Piero Ghezzi recognized from the outset that he was nowhere near the omniscient policymaker that conventional accounts of industrial policy posit. He knew there were problems that prevented productive upgrading. But he didn’t know exactly what those problems were. He couldn’t simply design an industrial policy scheme and implement it. He needed to engage the firms in problem discovery. He had to keep the conversation going, monitor outcomes and change course as required.

This might sound obvious, but in fact it is not what most analysts think of when they discuss successful industrial policies. Ask an economist or a technocrat what kind of policy setup maximizes efficacy, and you are likely to hear about the need to commit to a fixed set (or schedule) of policies, to keep the private sector at arm’s length, and to apply strict penalties when firms fail to deliver. Ask them why East Asia’s industrial policies appear to have worked better than those elsewhere, and they will explain that governments there followed these strictures.

This vision of industrial policy runs into trouble when there is rampant uncertainty about the nature of the underlying problem and the efficacy of alternative remedies. It is perhaps not surprising that it does not correspond well with actual East Asian practice either, conventional wisdom notwithstanding.

In a study of Brazil, India and South Korea, the sociologist Peter Evans found that the distinguishing feature of South Korean industrial policy was what Evans called “embedded autonomy.” Yes, government bureaucrats enjoyed relative autonomy from the private sector in that they could formulate broad policy objectives they thought were in the national interest and follow through with implementation unimpeded by businesses. But they also exhibited embeddedness, meaning they were engaged in ongoing communication and collaboration with the private sector.

We might worry that close relationships with private firms could render the government prone to capture. (I tell my students to make sure they do not confuse “embedded in” with “in bed with”!) But Evans argued these links were essential to ensure that governments could get access to the information needed to design workable policies, adjust to changing circumstances and prod firms along new technological trajectories. The difference with India and Brazil, Evans explained, was less the actual policies employed and more the manner in which the relationship with the private sector was managed.

Chinese industrial policy exhibits many of these elements of embeddedness. The architects of Chinese green industrial policies, write Professor Elizabeth Thurbon (University of New South Wales) and her co-authors, “behaved less like ‘top-down commanders’ (as authoritarian environmentalism would have it) and more like the collaborative ‘catalysts’ characteristic of traditional developmental states.” They argue this mode of government-business collaboration was critical to the success of their policies. Given the size of the Chinese economy, national policymakers invest significant effort to coordinate with local governments, to combine national resources with local knowledge.

In EVs, for example, the national government selected demonstration cities, which received priority in accessing financial incentives. In return, demonstration cities were expected to put in place complementary policies and raise their own resources. Cities then engaged in close collaboration with local companies and other stakeholders. Early results would be scrutinized by central government officials and experts, policies would be revised and disseminated accordingly, and the programs would be expanded to other regions. Municipal governments also often acted as venture capitalists, undertaking analyses of market and technological conditions before making investment decisions.

Liuzhou City, which achieved very rapid EV take-up, offers a particularly interesting example. Here, the municipal government worked closely with the local EV manufacturer, starting from the development phase. The local firm developed EV models that were specifically designed for the city’s transport and parking systems. At the same time, the city government introduced a variety of incentives, such as purchase subsidies, reserved parking and rapid deployment of charging infrastructure.

At the national level, the central government sought to institutionalize its collaboration with the private sector by establishing China EV100 in 2014. The group’s members included domestic and foreign manufacturers all along the supply chain, as well as high-level government officials and academics. The association was used as a forum for setting broad goals, coordinating the introduction of technologies, generating ideas about new policies and obtaining feedback from the private sector.

There is evidence from the U.S. that subsidy programs combining quantitative criteria and conditionality with flexibility and collaboration can work quite well. An example is the California Competes Tax Credit program. An initial list of awardees is selected through a strict formula that quantifies projected benefits. Administrators then negotiate with firms to finalize the list of recipients. These discussions produce a schedule of incremental employment, wages and investment targets, which the government monitors annually. Firms that do not stick to their commitments can risk losing their tax credits. But prospective applicants are told that administrators will do their best to work with them to prevent them falling into breach. A careful study has found that the program is effective in generating employment, especially in services.

Experimentalist Governance

An economist, the old joke goes, is someone who sees something work in practice and asks whether it can work in theory, too. The supreme theorist of the collaborative approach discussed in this chapter is Chuck Sabel, a political scientist and legal scholar at Columbia University. He has studied how governments make and implement policy in a wide range of settings where there exists high uncertainty about the effectiveness of policies and future technological trajectories, including public schools, environmental regulation, industrial diversification and social services. Along with Jonathan Zeitlin and other co-authors, he has distilled the lessons into a model of policymaking he calls experimentalist governance.

The traditional framework of government intervention that economists work with makes several key assumptions. First, that the policymaker has clear, well-defined objectives, such as physical investment or exports in a sector. Second, uncertainty is low dimensional. The government may lack precise information about, say, firms’ production costs but is otherwise well informed about the consequences of its actions. Relatedly, the economic and technological environment is stable.

Finally, economists assume there is little value in direct communication between private actors and the policymaker: because firms have the incentive to be strategic, the only way in which useful information can be elicited from them is by observing their actions.

Experimentalist governance applies in settings where uncertainty is pervasive and background conditions are inherently unpredictable. In such settings, policies cannot be designed and implemented without interacting with private agents. The problem of strategic behavior by firms is real, but it is only one feature of this public-private interaction. Firms also benefit from close interaction, and they have an incentive to build a reputation in what is an ongoing, iterative relationship with government agencies.

Experimentalist governance has four elements, linked in an “iterative cycle.” First, the policymaker and the principal stakeholders establish broad, provisional goals and determine the metrics for gauging progress. An example might be increasing the number of good jobs in a region or upgrading the productivity of informal firms in a particular service sector. Second, the executing agents – firms, municipal governments, innovators, civil society groups, public service providers, frontline workers – are given broad discretion along with financial/institutional support to achieve these goals. Third, these agents provide periodic reports and participate in informal peer reviews where results are compared across experiments.

Where progress is unsatisfactory, either the agents take credibly corrective steps or the experiment is abandoned. The conditionality that government imposes on agents is soft rather than hard, in the sense that agents are merely expected to show a good-faith effort to meet their commitments rather than to adhere to strict performance criteria. Finally, the objectives of the program are revised and disseminated to a broader circle of agents. And the cycle repeats.

From the perspective of experimentalist governance, what matters most to the effectiveness of productive development policies is not the policy instruments or sectors selected, but the government’s ability to navigate these four steps effectively. A government evaluates its policy framework not by asking, which tax breaks or subsidies are we using, which sectors have we identified, what is the budget for productive upgrading? The more important questions are: do we have the process in place whereby policymakers engage with the private sector on obstacles and opportunities? Do we have the organizational capacity to monitor progress on the ground and respond to the needs that these conversations are helping identify? Can we coordinate the requisite policies across institutional silos within the government?

A nagging question in all discussions of productive development policies is whether government agencies have the capacity to develop and implement the required policies. Experimentalist governance does not require a great deal of state capacity, at least to begin with. Rather than presuming they can discipline firms through explicit penalties or other forms of hard conditionality, it relies on firms’ own self-interest to engage in collective problem- solving. Nor does it depend on mutual trust between state and private actors. The assumption is that trust, along with general state capacity, will grow in the collaboration.

The experimentalist governance schema above captures the broad outlines of how DARPA/ARPA innovation programs operate in the U.S. The similarity with China’s EV promotion policies is also obvious.

Final Thoughts

Productivism prioritizes both the green transition and the broad dissemination of economic opportunity across the economy. It differs from neoliberalism in giving the government an important role in directing structural change and technological innovation to achieve these goals. It places significantly less faith in the ability of markets and large corporations to serve these objectives on their own. It emphasizes the real economy over finance, production over consumption, and revitalizing local communities over globalization.

Productivism also departs from the welfare state. It emphasizes that redistribution, social insurance and macroeconomic management are not enough. A truly inclusive economy, one that gives people dignity and social recognition as productive members of society, also requires intervention on the supply side to create good jobs for everyone. And productivism diverges from both its predecessors by favoring collaborative, experimental solutions over technocratic ones.

Productivism tackles inequalities where they are created. It intervenes at the source – in employment, production and innovation – instead of after the fact through income redistribution.

If you want to lift someone from poverty, the old adage goes, teach them how to fish instead of giving them a fish. The redistributive approach is akin to handing out fish, while pre-distribution policies, such as education, amount to teaching people how to fish. The productivist approach, on the other hand, makes sure there are enough fish in the pond in the first place.