Charles Sabel on Good Governance for Good Jobs

Charles Sabel is the Maurice T. Moore Professor of Law and Social Science at Columbia Law School. His current work includes “the elaboration of experimentalist or incremental solutions to apparently global problems such as trade and climate change; an investigation of the current transformation of US administrative law in the face of uncertainty; and new models of economic development emerging with the spread of advanced techniques of ‘industrial’ production to all sectors of the economy in the context of globalization.

Professor Sabel describes himself as “congenitally optimistic.” He’s well aware of the political division, economic inequality, and environmental degradation we face, but he thinks we’re up to the task.

Improving governance

The political, economic, and environmental challenges of the 21st century may seem intractable. Because of their scale and the fact that the costs are borne collectively, these problems require collective action, but even the best governments don't know all the solutions, and, even when their goals are clear, they can make important mistakes. How then can we expect to solve anything, much less deep-seated, global problems?

Sabel recommends that we not only acknowledge the uncertainty and fallibility in governance, but we lean into it. We design our institutions to be vigilant about disturbances, treating them as signs of possible flaws in the design or operation of an institution, and thus creating openings for change and novelty. He mentioned the Andon cord, which allows a worker who spots a defect to stop the assembly line, as an analogy to this sort of deliberate and controlled organizational vulnerability. This triggers an investigation of the root causes of the breakdown, which can lead to redesign of a part of  the product or a stage of the production process. Since decisions in such institutions are understood as revisable (rather than perfectly tuned in advance), policy done this way can learn from experience and progress in the face of uncertainty.

This requires a lot of work: monitoring, communicating, and updating the process, but it can be done. The Advanced Research Projects Agency-Energy (ARPA-E) provides a great example. This US government agency supports research and technology, which are inherently uncertain fields. To manage this uncertainty, in “every stage … of research—the definition of programs of investigation; selection of a portfolio of projects advancing the program purpose; and supervision of individual projects in the portfolio—ARPA-E treats goals as provisional, or corrigible in the light of experience.” Projects are closely tracked, and directors engage in “collaborative setting and revision of goals.” Consistently failing to meet these goals means a project will be scrutinized, and any necessary changes will be made (or the project will be terminated if this isn’t possible). The system seems to work: “Compared to projects in other branches of the DOE doing either basic or applied research, ARPA-E projects have a higher rate of patenting and the same high rate of publishing… [and] they are more likely than the specialized projects to produce both a publication and patent.”

This approach has been called “experimentalist governance” since the process of trying out new policies and adapting them based on experience somewhat mirrors the scientific process of experimentation. However, the experiment analogy is not perfect. There are no universal laws of governance analogous to universal laws in science, and there are no well defined control and treatment groups. In experimentalism, the actual implementation of the policy is the “experiment.” We learn if a particular policy will work in a particular environment by trying it, and the lessons come from watching and reflecting on what happens. In addition, governments can learn from their citizens through interactions outside of the “experimental” intervention. Unlike subatomic particles or amoebas (or human subjects in most social science experimentation, for that matter), members of society can contribute their knowledge to the policy design process and can comment upon their experience during the policy’s implementation. That said, the analogy to experimentation is useful. It brings to mind the key features Sabel endorses; experimentation is tentative, iterative, and “crucially accepts that you’re likely to be wrong, and therefore that it’s necessary to continue the investigation.” 

Experimentalism differs from the way that economists tend to think about policymaking in a number of important ways:

• The goal: Policy analysis in economics requires a well-defined objective: an outcome that can be measured and optimized, like economic growth or consumer surplus. Objectives in experimentalism are open to uncertainty and change. Broad goals (like fighting climate change or promoting technology) won’t change, but within that there are many short term objectives and metrics which are created, given up, or changed. In some sense, the policy process is as much about finding the right goal as it is about figuring out how to achieve that goal.
• Uncertainty & Learning: In economics, uncertainty is almost always assumed to be quantifiable; people may not know exactly what will happen, but they know what could happen and how likely each outcome is. So, learning is simply updating one’s beliefs about these likelihoods. Experimentalism treats uncertainty as more fundamental (more like Knightian uncertainty). It assumes that challenges and opportunities will arise which no one could anticipate, or someone could anticipate but only in principle — not in a strictly quantifiable way. Thus, learning in experimentalism is broader: not only becoming more accurate, but also understanding the nature of the policy environment, like how one can interact with it, how it could be different, and what features are desirable.
• Information: Economics is built in part on the assumption that the key empirical information about people is their behavior. The things people say are just a particular type of behavior and are generally treated with skepticism as a strategic message from a self-interested individual. By contrast, Sabel says experimentalism permits us to draw from people’s “self-awareness, reflexivity, and deliberation”; we can glean information from what people say about policy — not just as “behavior,” but as insight drawn from experience and thought.

That said, economics and experimentalism are not incompatible. In creating policies, economic insights about incentives are extremely important. And precise measurements of carefully identified causal effects are useful for evaluating and updating policies once they are implemented.

Experimentalism and capture

The close collaboration between the people making the laws and those affected by them may seem like it invites corruption. In many cases, the justification for government action is to correct situations in which private incentives are at odds with public interests. In this line of thinking, private actors given the ability to affect policy will do so in ways that are self-serving (thus, by assumption, bad for the public); they will capture the process. So, one natural response is to separate the rule-makers from those affected by the rules. 

However, in experimentalism, the subjects of laws and regulation are treated as active participants in the policymaking process. Sabel even calls them “co-designers.” But that doesn’t mean he’s ignoring capture; he just doesn’t think that keeping citizens at an arm’s length is the best solution. Policymakers must get information from someone, and “if you create a lot of distance, then you create an opportunity for somebody who can vault that distance to have an enormous influence.” Instead, the experimentalist approach to capture uses a few strategies: (1) building institutions which are actively inclusive so that narrow interests don’t have the only voice, (2) maintaining continuous monitoring of policy and reporting results so that a wide audience can detect breakdowns when they happen, and (3) promoting communication and flexibility so that people can work above board to make rules that benefit everyone rather than in back rooms to benefit themselves at others’ expense. Sabel concedes that this is not a silver bullet, but there are no silver bullets; people are clever enough to manipulate any system to their own advantage. The experimentalist approach aims to make it (1) harder to do, (2) easier to spot when it happens, and (3) less advantageous than cooperation.

Making good jobs

The dearth of good jobs — jobs which “provide a middle-class living standard, a sufficiently high wage, good benefits, reasonable levels of personal autonomy, adequate economic security, and career ladders” — is perhaps the most pressing economic issue facing the United States (and much of the developed world). Sabel says that “if you’re not interested in good jobs, you’re not paying attention to the state of our democracy.”

What makes good jobs so important? The most obvious answer is that having a steady income allows people to buy the things they want and need — the standard notion of economic well being. However there are also knock-on effects which are less obvious but just as important and far-reaching. A lack of good jobs contributes to (1) economic inequality, (2) social problems like family breakdown, crime, and substance abuse, and (3) political problems like divisiveness, support for authoritarianism, and nativist populism. Sabel also emphasized the personal meaningfulness of good work: “being able to make the things you want to make is indispensable to your flourishing.”

With all these benefits, why aren't good jobs more readily available? The trouble is that many of the benefits of good jobs are broadly distributed throughout society; the job creators themselves only get a part of the benefit, so (acting alone) they don’t make as many good jobs as would be socially desirable. 

Thus, the task of making good jobs must be shared by public and private actors. We go back to the differences between economics and experimentalism to show how an experimentalist regime might approach creating good jobs.

• The goal: The typical economic approach would require a specific definition of a good job, or at least a clear metric by which one job definitively could be called better than another. What is a “sufficiently high” wage? What is a “reasonable level” of personal autonomy? And what constitutes personal autonomy on the job? These questions are important to experimentalists, too, but they don’t need to be fully agreed upon before policy can be tried, and the answers can vary in different contexts. For the sake of experimentalist policymaking, it is enough to have the central goal of creating good jobs with provisional ideas of what that entails. 
• Uncertainty & Learning: Creating jobs involves uncertainty about investments, technologies, and institutional arrangements (both government and business). In economic analysis, the menu of choices and the likelihood of their outcomes are assumed to be known in advance. Experimentalists would treat these as at best partially known. Not only will known choices lead to surprising outcomes, but entirely new choices might arise — new investment opportunities, technologies, or arrangements that could only be discovered by trying things out. 
• Information: To inform policy, economists would look at similar changes in the past and estimate their effects on behavior, for example the effect of minimum wages on employment or the effect of education on salaries. Experimentalists would take these things into account, too, but they would acknowledge that there’s still uncertainty; these past estimates come from a particular time and place that do not fully account for conditions here and now. Moreover, they have another important resource: the input of business leaders and employees who would be affected by job creation programs.

Taking these together, an economist’s policy recommendation for creating good jobs would typically be something like a subsidy for employers who create jobs that meet certain thresholds regarding wages and stability or a subsidy for workers to get an education that is likely to lead to a good job. The amount of the subsidy would be chosen to optimize the benefits of getting a good job versus the cost of the subsidy and the likelihood of creating/obtaining a good job in response to the subsidy, where both the costs and benefits are calculated and known (at least on average).

An experimentalist’s approach would be more involved. They might consult with business owners to determine the barriers to creating good jobs: Does the market need standardized wages? Do they need credit to invest in technologies which make their employees more productive? Do prospective employees need better vocational training? Any of these could be the case (and it might vary from place to place), and they each call for different responses. A program might start out on a voluntary basis so that firms with limited capacity could continue to operate without incurring prohibitive penalties. For example, firms might be allowed to opt into a system in which they receive funds for technology on the condition that they create a certain number of jobs. The agreed upon number of jobs might turn out to be much too low or too high, so the government could adjust it. The government could also monitor these firms, so that if the program became mandatory later on, the companies forced to join could learn from the experience of the early adopters. Any of the principles of experimentalism could be applied to creating good jobs.

Believing in better

When asked what sort of stories give him hope for policy to make positive change, Sabel gave the example of the Chicago Public School system (CPS).

In 1989, CPS was in bad shape. It had undergone eight strikes in the previous years, and its finances were in shambles. Fewer than half of the students were graduating, and racial disparities along all dimensions were massive. The Secretary of Education publicly declared it the worst urban school system in the country. 

From parents and policymakers to administrators and business leaders, everyone agreed that things weren’t working. So, Illinois passed a law decentralizing control to local school councils. The councils couldn’t fix everything immediately, of course, but they began with the tentative step of hiring new principals. These principals were innovative people and people of color who knew about the problems at hand and could make important (though not earth-shattering) changes like modifying curricula and encouraging new ways of teaching. Seeing initial success, they were promoted into the district administration. As administrators, they saw that they could support principals’ innovative efforts in instruction by taking over their erstwhile duties in day-to-day school management like budgeting and operations.

Over the next three decades, the changes were quiet, but things steadily got better. Academic standards and graduation rates went up. College-going went up. Racial performance gaps fell. CPS still faces many hurdles, but its improvement has surprised even the most informed people.

This story highlights a few lessons:

• No one knew the complete answer to all their problems at the outset. This exemplifies the deep uncertainty which motivates experimentalism, and the solution exemplifies the principles of experimentalism. The state distributed decision making to local actors. The local actors tried out policies they weren’t sure would work. As Sabel says “they invented their own version of this stuff” as they went along. Then they saw something which worked and seized on it. They learned from their experience and modified their institutions accordingly. 
• Experimentalism can work even in the most inhospitable environments. The collaboration and cooperation required to make experimentalism work might seem impossible in a divisive political age, but they’re not. Sabel described CPS in the 80s as a “big, incredibly broken thing in a roiling sea of difficult politics” but it was still able to improve.
• The new and improved system isn’t more expensive than the old one. There’s a tendency to think that big changes require big expenditure, especially since experimentalism requires so much work. While it's true that the transition required time and effort (and money), the new system isn’t more costly than the old. The content of teaching has changed, but the teachers are still working the same amount. Principles are more invested in instruction while administrators are more invested in operations, so the division of work has changed, but the total amount of work has not. In fact, many private companies use management structures which mirror experimentalism. The principles are efficient enough that they can survive (and thrive) under market competition.
• There’s good reason to be optimistic. The process may take a long time, and it may not be perfect, but it’s better than it was, and that’s a realistic goal for any problem.