Credible Reasoning, Contestation, and Dynamic Preference Production, by Anya Bernstein
*This is the second post on a symposium on Jed Stiglitz’s “The Reasoning State.” For other posts in the series, click here.
The Reasoning State presents a broadly interdisciplinary discussion and defense of the administrative state. Jed Stiglitz engages with history, political science, and psychology to argue that credible reasoning about policy matters, incentivized and enforced through external constraints, makes public institutions both more trustworthy and more trusted—and that, in the U.S. federal government, the institution with the most credible reasoning is the administrative state.
Stiglitz’s argument responds to the denigration of administration that pervades our public discourse and court opinions, which paint agencies as unaccountable, even dictatorial. In place of the deep state that so many fear, Stiglitz finds in American administration a crucible of reasoning overseen by many masters who keep it within many bounds. Recognizing that all policy decisions have political valences and involve value judgments, Stiglitz locates administrative legitimacy in the structures that keep agencies deliberating, explaining, and submitting to external review (especially by courts)—rather than in traditional claims that administration is a realm of neutrality and pure scientific expertise.
Stiglitz asks why we have set up our government with agencies, as reviewed by courts, making specific policy decisions, and suggests that this complex, the “reasoning state,” was designed to solve a legislative trust problem. Early republican legislators were big fish in a small and homogeneous pond, known by everyone—everyone who mattered to them, anyway, which in the early republic meant just the discrete and insular minority of the enfranchised. Personal social networks helped voters decide whether a legislator was trustworthy. But then both the electorate and legislative activity expanded and diversified. Modern constituents lack sufficient information to trust legislators’ decisionmaking process; and no external constraints make it reliably trustworthy. Legislators developed the administrative state to lend credibility to their actions: voters can get a fairly reliable sense of a legislator’s overarching, general policy preferences, but don’t have to guess about more particularistic decisions, which have been outsourced to the administrative state.
I am wary of “why” questions about social phenomena. The answers tend to assume that things turned out the way key actors predicted they would, which allows us to impute those effects back onto those actors’ intentions. But actions often have effects that actors neither wanted nor predicted, nor even considered; and there are often more actors in the mix than we recognize. The why format asks about intentions but leaves out a lot of the contingency, unpredictability, small-path dependency, and unintended consequences that pervade human affairs. Still, leaving intentions aside, Stiglitz gives us good reason to think that the administrative state has had the effect or function of producing more credible reasoning about policy matters than one could get elsewhere in our system.
Stiglitz conducts experiments that suggest that requiring rationales reviewed by external parties improves the quality of individuals’ reasoning, which in turn inspires more trust by those affected. Extrapolating, Stiglitz suggests that agency reason-giving subject to judicial review likely improves policy deliberations and yields explanations more acceptable to affected publics. In particular, it makes agencies’ specific policy decisions more credible than Congress’s would be. Unlike agencies, Congress’s reasoning is not constrained, transparent, or subject to review for rationality. Stiglitz concludes that congressional elections give credibility to broad normative goals—which voters can know and evaluate—but not to policy particulars, which are sufficiently opaque and complex that they are best left to more constrained institutions.
These conclusions resonate with arguments Cristina Rodriguez and I make in a forthcoming article drawing on interviews with agency personnel from across the federal government. Consistent with Stiglitz’s claims, our interviewees were highly focused on policy rationales. Across different positions in different kinds of agencies, subjects described their workplace as suffused with ongoing, multilateral reason-giving to multiple audiences, as participants present findings, views, and values to one another in an ongoing process of multi-participant deliberation that ultimately involves career civil servants, political appointees, people in other components and agencies, and the White House bureaucracy, as well as interest groups, general publics, and courts. This deliberative process is, of course, organized in more or less hierarchical ways (depending on the setting). But in participants’ descriptions, reason-giving requirements permeate and cross-cut hierarchy, making simple top-down rule difficult.
Our interviews also support Stiglitz’s suggestion that reviewability pushes people to consider the views of others. Participants described their institutions as engaged in ongoing interaction with external publics and circumstances. They took account of technological innovations, changing practices, and other real-world events that shape the regulated world. And they communicated regularly with the publics they affected, not just through notice-and-comment rulemaking but through focus groups, exchanges, and meetings, as well as through litigation and the assessment of litigation risk that hangs over every important decision. Such practices naturally raise questions about which publics get what access to agency ears. And, as Glen Staszewski and Michael Sant’Ambrogio have written, agencies engage with their publics a lot, but in ways that should be strengthened and made more consistent. Still, our findings bear out Stiglitz’s general point that the structure of administrative law encourages agencies to consider the views of those they affect—and then some.
Stiglitz describes agency process as providing a beneficial alternative to political control by Congress, which lacks constraints on its reasoning. Our research supports this idea—not least because no member of Congress could have the in-the-weeds knowledge of particular policy areas, in all their technical and political intricacies, that the multi-participant workgroups of topic-focused agencies can acquire. Similar reasoning goes for the President. At the same time, we concluded that the variegated mix of different relationships to political bodies that agencies encompass supported the administrative state’s accountability. Our interviewees described political appointees, with their closer relationship to the president, as interdependent with civil servants, who brought longer-term concerns to the process. These groups’ different values and viewpoints—as well as other contrasts such as those between agency lawyers and subject matter experts—contributed to that full flush of reason-giving that both Stiglitz and we find so valuable in the administrative state.
This interacting diversity of viewpoints also underlies the two ways I would expand The Reasoning State. Stiglitz makes a good case that credible reasoning lends democratic legitimacy to the administrative state. But I think we should demand more. As Glen Staszewski and I argue in a forthcoming piece, it is not just reasoning but contestation among different values and viewpoints that enacts the ideals of republican democracy we should strive for. This agonistic republicanism insists that government take multiple viewpoints into account without pretending that there can ever be a perfect, or an eternal, way to accommodate them all. Our administrative state is up to that challenge. But authoritarian populism—including a strain we have identified as judicial populism—threatens that strength by insisting on single answers to inherently polyvalent questions. It is thus crucial to ensure that the administrative state not only provides credible reasoning, but also continues to host the ongoing contestation among diverse views and interests that is the hallmark of a modern democracy.
Finally, Stiglitz frames his discussion in terms of trust, asking whether the state’s actions are trustworthy and whether the public trusts them. He defines trust fairly minimally, as the justified faith that another will act in our best interest. This definition leaves out the way that interests are themselves dynamically produced in the policymaking process. In our agency interview study, Rodriguez and I found that producing a policy preference was itself a group project. The standard image of a political leader who comes in with a firm policy plan and orders subordinates to implement it was not reflected much in our interviews. Rather, participants—at all levels of the organization—mostly had orientations, interests, or ideas that got worked out in the complex, multilateral, contestatory process of policy production.
Moreover, we may often feel have no best interest with respect to some specific policy decision. Whether the Mine Safety and Health Administration adopts one particular way of measuring airborne coal dust or another is not something a given person might have much stake in, nor even an opinion about. Yet the state is obligated to that person, too, to engage in credible, contestatory reasoning. One of the key things the administrative state does is connect larger values (justice, equity, development) with concrete policies that affect more specific publics and situations to varying degrees. In other words, the process of policymaking itself helps develop our ideas about best interests—a recursive, interactive practice that makes the state particularly important as an arena of contestation.
Anya Bernstein is Professor of Law at the University of Buffalo.