It’s hard to remember a time in which economic malaise was so pervasive and credible solutions to basic economic problems so remote. In the affluent industrialized countries, the welfare state is atrophying, while the drawbacks of the neoliberal free-market-driven capitalism that claimed to supersede it are increasingly evident. Meanwhile, in low-income countries, the tried-and-tested model of development through export-led industrialization has run out of steam. And even where economies remain vibrant, prosperity comes at an ominous price. A variety of forces are combining to exacerbate divisions between high productivity urban areas and less productive communities and industries that neither contribute to innovation nor benefit from it. This dualism is widely seen as generating growing inequality, despair and receptivity to authoritarian populism. While none of this is really news, much of the conversation about the current malaise focuses on solutions that ignore the deep economic roots of the problem. Redistribution through taxes and transfers accepts the pernicious dual productive structure as a given. Investments in education and a stronger safety net seek to enhance the productivity of the workforce, but on their own offer little hope of integrating the dual economy.

Fiscal stimulus, for its part, may be useful in minimizing the gap between actual and potential output in the short run. But the dualism that poisons society cannot as a practical matter be addressed solely through demand management. By the same token, although the rise of market power has no doubt exacerbated inequality by allowing corporations to increase prices and drive wages below the competitive level, antitrust enforcement is at best a partial solution.

Glimmer of a Solution

What’s left? We offer what, on initial glance, may seem old wine in new bottles: direct government intervention in ailing sectors in collaboration with the most productive segments of the private sector. The idea is to create productive (and high paying) jobs by supporting firms and workers in efforts to expand the dynamic portion of the economy. Until someone thinks of a catchier phrase, we’re calling it a strategy to “build a good-jobs economy.”

Our definition of a good job isn’t all that ambitious. We have in mind stable on-thebooks employment that comes with core labor protections that most of us used to take for granted. But a good job also pays enough to ensure a middle-class living standard, with enough left over to educate the kids and save for retirement.

Again, there’s no magic here. Creating good jobs turns on increasing skill levels and firm competitiveness, while bolstering efforts at income redistribution and aggregate demand management. Public-private collaborations must be at the heart of the strategy because neither firms nor government has the knowledge or authority to act alone.

But, inspired by the flexible arrangements that have evolved in sectors where innovators must deal with considerable technological and market uncertainty, we do think we have something to add when it comes to the principles by which public and private actors can collaborate under today’s highly differentiated and rapidly changing conditions.

We’re not quite first to this party. For example, Michelle Miller-Adams and colleagues at the Upjohn Institute recommend differentiated strategies, tailored to individual localities, that combine skill development with programs to attract businesses. In view of the uncertainty about what works in general and the need to adapt broad solutions to locales, we focus here on a broad regime for generating good jobs, not on specific interventions.

Good Jobs as a Positive Externality

Before we can get to the “what” and “how” of our good jobs strategy, we should say a bit more about the “why.” We need to explain how good jobs generate social benefits far exceeding the gains to employees and employers – and thus why, left on their own, markets won’t produce enough of them.

When wages rise, firms try to economize on labor by adopting labor-saving technologies. The result is an undesirable trade-off between good jobs and the level of employment. Historically, this tension has been resolved by an economy-wide rise in productivity, which offset the costs of labor dislocation. For example, the mechanization of agriculture during the 19th and early 20th centuries (and parallel events in post-Mao China) created a surplus of labor in the countryside that was absorbed by higher-productivity industry.

Deindustrialization in affluent countries during the second half of the 20th century led to a similar, but more challenging, dilemma. Rapid labor productivity growth in manufacturing and the growth of import competition led to a loss of production jobs and a shift to service employment, where wages and conditions were often inferior. The benefits of these new technologies remain bottled in a limited number of sectors and localities, generating relatively small numbers of good jobs, while wage stagnation reigns elsewhere.

We do not view this simply as a problem of inequality and exclusion, but also as a problem of market failure that can be described in terms of externalities. The substitution of bad jobs for good, as firms use technology to minimize production costs, undermines the social structures that underpin economic prosperity, and thus can create enormous negative externalities.

This is not speculation. David Autor of MIT and colleagues studied communities across the U.S. to weigh the impact of competition from China. Localities under greatest stress experienced an increase in “idleness” among young men and a rise in male mortality due to drug and alcohol abuse, HIV/AIDS and homicide. Job loss also led to an increase in the number of unwed mothers, children in single-headed households and children living in poverty.

These impacts were compounded by the political consequences – notably, the rise of nativist populist movements. Autor and (other) colleagues found that places experiencing sharper increases in import competition were less likely to elect “moderate” legislators in 2010. Indeed, their quantitative research implies that Hillary Clinton would have obtained a majority in the Electoral College in 2016 if Chinese import penetration had been half what it actually was in the years 2002-2014.

A recent paper by Ernesto Dal Bó and colleagues traces a similar dynamic in Sweden. Conservative initiatives to reduce social insurance and government transfer payments while simultaneously lowering taxes increased the income gap between those with steady jobs and those who were either unemployed or relied on temporary jobs. The main political beneficiary of the resulting populist backlash appears to have been the right-wing, anti-immigrant Sweden Democrats party. Parallel research implies that import competition from China partly explains the political realignment behind Brexit as well as the shift toward the radical right in the rest of the EU.

The Inadequacy of Standard Remedies

The conventional means for internalizing a positive externality is a subsidy, which would be a generalized employment subsidy in this case. But successful administration of such subsidies requires a solid sense of the size of the externality. Moreover, in the standard conception of externalities, technological opportunities are fixed, and the policymaker has a good sense of how employers and workers will react to the subsidy. The appropriate intervention consists of directly offsetting the externality and doing no more than that. But when there is deep uncertainty about how firms, workers and consumers will respond, learning about what does and doesn’t work to create good jobs becomes an integral part of the policy process.

Finally, as uncertainty increases, it becomes difficult to specify in advance not only the costs and benefits of intervention, but also its precise objectives. What is a good job, how many can be reasonably created, how technological and other firm-level choices influence job creation – these are necessarily local, contextual questions. They can be answered only through ongoing strategic interaction between public agencies and private firms. This process is alien to the familiar framework of intervention with taxes and subsidies (which economists are most comfortable with) that assumes goals and social benefits must be known in advance. But it is the hallmark of the new type of intervention to which we turn next.

Intervention Under Extreme Uncertainty

Consider first the circumstances in which contracting parties know enough to specify what each expects from the other. Precision is often unnecessary because the parties often contract repeatedly with each other, giving rise to trust and shared expectations. But when circumstances are changing in unpredictable ways, neither party can say exactly what will be feasible – let alone what the other should contribute to the joint effort.

What’s needed, then, is an agreement that establishes broad goals and a systematic way to evaluate achievement. As collaboration progresses, each party comes to rely increasingly on the capacities of the other. Trust and mutual reliance are the result of agreement to collaborate, not its precondition, just as the precise aims of cooperation are the outcome, not the starting point.

Regulation under extreme uncertainty arrives at a related solution from a different starting point. Under stable conditions, mitigation of externalities is mandated by legislation and given precise form. The upshot is a fixed set of limits on permissible behavior and a schedule of fines for exceeding them.

But in response to uncertainty – as seen in food safety, civil aviation and pharmaceuticals among many others – regulation must be subject to continuing exploration of solutions. Typically, the regulator establishes an ambitious open-ended goal: for example, “good water” as measured by minimal deviation from the pristine state, or a large but gradual reduction in vehicle emissions requiring technological change. The regulated entities are obligated to make plans to achieve the goals and to regularly report their results.

Penalties in this regime are not calculated to deter infractions of clear rules but to create incentives for openness from which standards will eventually be derived. Thus, penalties are imposed only for failure to report honestly or for persistent failure to achieve results comparable to those of peers. Though infrequent, those penalties can be daunting, often amounting to exclusion from the market or severe limits on autonomy.

This approach encourages investigation of new possibilities. If some actors are looking to raise the bar though innovation, competitors will be less willing to cling to the status quo. In an environment where the development of technology is uncertain precisely because of the surprising abundance of opportunities it affords, the search for innovation is likely to feed on itself. Indeed, the search will probably be most effective if it is collaborative, either because projects are interdisciplinary or because any one approach is likely to fail, and it thus is prudent to pool the risks.

“Notice and comment” – the one-time consultation of stakeholders required under U.S. law before the promulgation of rules – gives way to regular, organized exchanges as the parties seek to establish common expectations in the face of rapidly evolving knowledge. By making it risky to bet on the status quo and potentially rewarding to try to surpass it, this sort of regulation turns uncertainty from an obstacle to tough standards into a spur to find out how to realize them.

Two Examples of Success

We point to two instances of success for this approach. The first is the initiatives of the Defense Advanced Research Projects Agency (DARPA) and its offspring, the Advanced Research Projects Agency-Energy (ARPA-E) to jumpstart technological change. The second is regulation of pollution by the Irish dairy industry. Neither case is perfectly analogous to the good-jobs challenge. But the success of both is due to the emergence of common mechanisms of governance under uncertainty that could make the good-jobs strategy workable and accountable.

DARPA and ARPA-E

Created in 1958 in response to the Soviet Union’s launch of the Sputnik satellite, DARPA is often invoked as a reminder that the knowledge economy was not created solely by private actors responding to market signals. Among its iconic contributions: the computer network protocols underlying the internet, precursors to global positioning systems and fundamental tools for microprocessor design and fabrication. ARPA-E, created in 2009 to foster innovation in energy, is faithful to the original model.

ARPA-E’s overarching goal is to eliminate “white spaces” in the landscape of technical knowledge – missing capabilities, just beyond the frontier of current technical possibility that would clear the way to advances in an important domain.

A program might, for example, aim to support investigation of novel battery concepts with the potential to allow 24/7 electricity delivery from wind or sun energy alone. At every stage in the organization of research, ARPA-E treats goals as corrigible in the light of experience. As with the contracts among innovating parties discussed above, precise goals are a result of a search rather than fixed from the first.

ARPA-E’s program directors are hired largely on the basis of their promise in giving direction to an emergent area of investigation. Once program goals have been framed, the director does a “deep dive” on past efforts and then tests the practicality of the emerging research area in technical workshops involving leading experts in relevant fields. If the research plan passes review, a project is formally created.

Proposals for research within the projects are developed in the same manner, with goals open to recurrent revision. Winning researchers negotiate project milestones with agency staff. Missed milestones can touch off intervention to get the project back on track. In short, the agency rejects the model of handsoff, bet-on-the-person-not-the-project administration in favor of ongoing collaborative review and adjustment.

ARPA-E is too new to permit any evaluation of its long-term impact. But the available evidence does strongly suggest the agency is using its information-generating regime effectively to make choices. What’s more, ARPA-E is already generating a higher rate of patenting than other research-related branches of the Energy Department.

Irish Dairy Farming

Regulation, and especially environmental regulation, differs from ARPA-E’s contractual governance of research in two ways. First, agreements between the ARPA-E and award recipients are fully consensual. By contrast, many regulated enterprises would prefer no public constraints, and some actively resist their imposition. Second, ARPA-E faces uncertainty that arises from manifest limits of knowledge. Environmental regulation encounters such frontier uncertainty as well. But the uncertainty often arises from the way that known factors combine to produce unforeseeable results.

Environmental problems typically must be addressed place by place – they are more often “white dots” than white spaces, and filling in one is little help in filling in another. In this regard, environmental regulation can serve as a partial model for regulation of the good-jobs externality. In both, a central task is to induce the local actors to cooperate, enabling them to benefit from the pooled experience.

The conviction that environmentally sustainable dairying could be a modern engine of growth came late to Ireland. Through much of the 20th century, farming was dominated by small, low-productivity holdings with limited export opportunities. But the EU’s Common Agricultural Policy together with the imposition of EU milk quotas prompted some consolidation, yielding more efficient, more specialized dairy farms. Ireland, which accounts for less than 1 percent of global milk output, now supplies almost 15 percent of the world’s infant formula and, all told, exports 90 percent of its dairy output.

The success of Irish dairy farming turns on the availability of home-grown grass feed, which is both cheap and yields superiorquality milk solids. The potential catch: pasturing must be environmentally sustainable.

Farms that fail EU mandates to limit nitrate pollution can be fined or disqualified from the EU single-farm payment. Countries that fail to meet national limits must submit a plan for improvement to secure delays from draconian sanctions.

The EU’s Water Framework Directive of 2000 has, in contrast, extremely broad objectives: “good water,” defined as minimal deviation from pristine bodies of water of the same type. The basic unit of management is the river basin or catchment, the contiguous territory that drains into the sea at a single river mouth, estuary or delta.

Implementation of both the nitrate and water quality directives has proved frustratingly difficult. Adherence to “good practices” in agriculture has often failed to produce improvements in nitrate levels, while lack of stakeholder participation and continuing revision of the intentionally open-ended goals have been major stumbling blocks in applying the water quality standard.

Compliance failures triggered research to improve understanding and control of pollution flows. One key finding is that variations in soil and the underlying geology are, in combination, so influential in the absorption and drainage of nutrients that general rules of nutrient management – let alone plans based on them – will regularly fail. The research programs, linked with similar ones in other member states, have borne other fruit, too. They’ve helped to generate a web of institutions that is coming to function as a system of local governance of water quality, greatly expanding public participation in the process.

Where Good Jobs Fit In

Simplifying greatly, the uncertainty faced in the ARPA-E case is principally about technological feasibility. The uncertainty in the dairy case derives largely from the need to customize solutions for individual localities. A good-jobs program faces uncertainties of both kinds.

Creating or preserving good jobs in a particular place often depends on adapting to technologies that are so new to a given locale that they must almost be reinvented to be mastered. Here ARPA-E’s experience with active project management is directly relevant. But fostering good jobs depends at least as much on solving place-specific problems. Here, new forms of collaboration developed in Irish pollution control come into their own.

An immediate question, then, has to do with the timing, scale and scope of the obligations (and penalty defaults) that are to be imposed. If there is a genuine good-jobs externality, there is no reason in principle that the obligation to join the effort shouldn’t apply immediately to all firms. But as just noted, those obligations are inherently broad, open-ended and, at least initially, ill-defined.

Unless we assume extraordinary consensus in favor of addressing the jobs externality, we can’t imagine regulators imposing the mandate across broad swaths of firms. And if we can’t imagine that, still less can we imagine imposing penalty defaults for persistent failure to make good-faith efforts to comply.

It is easier to envision using carrots to encourage participation, offering extensive customized support services benefits to volunteers. The framework goals, continuous monitoring and reporting requirements and penalty defaults would apply.

A key benefit of these voluntary arrangements over the medium term would be to develop good practices suited to a wide range of settings that can eventually guide application to a larger set of firms. Put differently, the initial selective projects would serve as a pilot program for the new system. As formal obligations are extended, the arrangements would come to resemble the European regulatory model, with a uniform requirement of participation but responses highly differentiated by locale.

With these practicalities in mind, a scaledup good-jobs industrial policy can be explored. First, the government commits to legislation to address the problem. Next, regulators who have authority for areas directly affecting jobs, like vocational training, introduce incentives to innovate, anticipating the need for additional support services. The requirements could take different forms, including specific employment quantity targets and/or standards.

Where current regulatory authority doesn’t reach, the government would create volunteer public-private programs to advance the frontiers of technology and organization, or provide support services and perhaps subsidies to help firms bridge the gap between their current low-productivity/low-skill positions and participation in the advanced sector. These programs would have to be customized to the needs of particular sectors and locales. Finally, conditional on the success of voluntary arrangements, these practices would gradually be made obligatory for non-participating firms.

Comparison with Current Initiatives

Of the components of a good-jobs program, workforce development has to date achieved the most success. The experience has been nearly the reverse with the job attraction component: local and state politicians typically try to outbid each other to win outside investments in deals that specify all terms of the exchange in advance.

Extension services and cooperative research programs are an intermediate case, with waves of institutional innovation leaving the policy landscape dotted with small organizations that appear to do some good in their domains but lack broader impact. These different outcomes, so far as we can see, reflect the vagaries of policy choices and economic flux. To all appearances, in fact, workforce development should be the most intractable terrain since programs must engage at-risk groups and address the compound problems – financial, educational, familial – that stymie social services. The focus here, therefore, is on workforce development.

Many of the most successful workforce development programs trace back to Project Quest (Quality Employment Through Skills Training), founded in San Antonio in 1992 in response to a wave of plant closings. The displaced workers lacked the skills for new jobs in health care, IT and other sectors, while the service jobs for which they were qualified paid too little to support a middle-class family. Two faith-based organizations, seeing the urgent need to prepare the region’s largely Hispanic population for good jobs, secured municipal funding for Project Quest.

The new project faced a double challenge: on the one hand, it had to identify opportunities in the local labor market, alert the city’s community college system (then inattentive to business needs) to them, and help shape the substance and timing of new courses. On the other, it had to support high-risk learners, most of whom needed to pass remedial courses to qualify for further study, and many of whom carried family and financial burdens on top of anxieties about returning to school.

Project Quest turned to former military leaders with long experience in workforce development. These managers brought with them not habits of military discipline, but rather the culture of continuous improvement that took root in the U.S. military before it became commonplace in the civilian economy. An expression of this culture was the early creation of a dedicated management information system to track the performance of individual students, both to keep their counselors abreast of their progress and to allow continuing review of overall organizational performance.

To be eligible for Quest, students must demonstrate financial need and levels of literacy and numeracy sufficient to complete required remedial courses. Once admitted, students design a bundle of wraparound services and supports, including subsidies for tuition, child care or rent, or services to address problems of transportation, health and domestic violence.

Students meet their counselors individually and in small stable groups in weekly sessions, where they share problems and devise support strategies. The counseling, in combination with information on classroom performance, allows for continuous adjustment of the support bundle.

A recent evaluation of Quest participants nine years after leaving the program found they earn roughly 10 percent more than the control group, and the gap does not diminish over time. Crucially, the difference in earnings is the greatest for participants who were older than the normal school-going age, with children and with additional burdens.

In recent years, as community colleges have been drawn into training partnerships with local firms and the failures of “light touch” interventions to increase completion rates have become more conspicuous, the schools themselves are starting to provide Quest-like services. A leading example is the Accelerated Study in Associate Programs (ASAP) of the City University of New York.

Like Quest, ASAP provides financial support and wraparound services, above all a dedicated adviser for each student. Again, a principal goal is to identify and resolve issues before a student drops out. Nearly 40 percent of the students in the program group graduated by the end of the program, compared with 22 percent for a control group.

Meanwhile, three Ohio community colleges have implemented ASAP, and early assessments show results comparable to those in New York. Growing confidence in the ability to train low-skill workers is encouraging community colleges to partner with large firms like Amazon, where the school offers customized training and student support and the company pays tuition expenses, creates work schedules for participating employees that mesh with school needs and provides a career ladder from unskilled work into management for those who complete the program.

If workforce development initiators are increasingly aware of the need for continuous learning, programs that directly target employment by attracting investment seldom are. The most visible, typically administered by states rather than the federal government, are tax incentives for large investors in return for specific commitments on job creation.

The Foxconn and Amazon deals in Wisconsin and New York, respectively, are recent highprofile examples. The Taiwanese company Foxconn agreed to create 13,000 well-paying jobs in Wisconsin in return for $4.5 billion in government incentives. Amazon promised 25,000 jobs over a decade in return for nearly $3 billion from New York. But both arrangements have blown up amid controversy.

These tax incentives were predicated on markets’ being sufficiently predictable to allow rational employment commitments. If the firm turns out to be unwilling or unable to carry out the terms of the contract – as was the case with Foxconn and Amazon, the former because of unforeseen changes in demand and technology, the latter because of unexpected community resistance – there is little room for revision or renegotiation.

Reason to Hope

We conclude on two positive notes. First, we believe that the prevailing academic pessimism about government- led job creation is misplaced. Much of the research suggests there are few, if any, policies that work consistently to expand good jobs. But that’s to be expected: the dual challenge of dealing with uncertainty and contextualization implies there are no fixed, clear-cut remedies. What is important is to get the governance regime right and allow localities to iterate toward successful practices.

Second, the governance arrangements we have sketched out may also help enlarge the constituency for acting on the problems they address. The same institutions that enable the parties to specify and solve problems under uncertainty also enable them to develop the trust and mutual reliance they need to deepen and broaden their efforts.

We agree that building political will is a threshold condition for a successful industrial policy. But under current conditions – when uncertainty makes the selection of goals necessarily provisional and the revision of ends and means routine – the broad coalition required for an effective good-jobs strategy need not preexist. It is likely to be the result of pursuing the strategy.

A bonus to using this governance approach is its natural affinity with efforts at broad mobilization to address societal problems under uncertainty in other domains. The Green New Deal is the most prominent example.

The GND goes beyond the remedy of carbon pricing to offset the negative externality of climate change to contemplate large-scale investments in green technologies and ambitious programs to foster greater economic opportunities.

Any way it turns – whether confronting environmental problems or creating employment, and most especially doing both together – the GND will wrestle with the application of familiar ideas to the collaborative exploration of the technological frontier that gave rise to the design principles of our good-jobs strategy. It would hardly be a surprise if the GND, combining goals that only recently seemed distinct, reached similar conclusions, creating the core of a broad coalition that responds effectively to the externalities threatening our democracies and our planet.